Sie sind auf Seite 1von 420

CORNELL UNIVERSITY

LATIN AMERICAN STUDIES PROGRAM

ANDEAN PAST
Volume 5 1998

ANDEANPAST Volume 5 1998 Editors: Monica Barnes


Cornell University

and Daniel H. Sandweiss


University of Maine

Guest Cf!-editor, Inca Section:

Brian S. Bauer
University of Illinois

Graphics Editor:

David Fleming New York

Editorial Advisory Board

Richard L. Burger
Yale University

Thomas F. Lynch Brazos County Museum of Natural History

Craig Morris American Museum of NaturalHistory

Copyright c 1998 by the Cornell University Latin American Studies Program ,

The Cornell UniversityLatin American StudiesProgram is the publishinginstitutionfor ANDEAN PAST. Orders should be addressed to: La.tinAmerican Studies Program; 190 Uris Hall, Cornell University, Ithaca, NY 14853; telephone (607) 255-3345, fax (607) 255-8919. Inquiries and manuscriptssubmittedfor future volumesshould be sent to Monica Barnes, 377 Rector Place, II-J, New York, New York 10280; tel. (212) 945-0535; cell phone (917) 992-5880; e-mail 103225.12@compuserve.com or to Daniel H. Sandweiss,Dept. of Anthropology,S. Stevens Hall, University of Maine, Orono, ME 04469-5773; tel. (207) 581-1889; fax (207) 581-1823; e-mail dan_sandweiss@umit.maine.edu.

This publication is partially funded through the U.S. Department of Education Title VI Cornell University/Universityof PittsburghConsortiumon Latin American Studies.

ii

ANDEANPAST Volume 5 1998 TABLE OF CONTENTS Editor's Preface by Daniel H. Sandweiss Reidy Fogel, 1956-1994by Richard L. Burger Daniel Wolfman, 1939-1994by Izumi Shimada
Thematic Section: Inca Studies

v 1 5

The Inca Compound at La Centinela, Chincha by Dwight T. Wallace Reconstmcting the Great' Hall at Inkallacta by VincentR. Lee

9 35

Reconstmcting Andean Shrine Systems: A Test Case from the Xaquixaguana (Anta) Region of Cusco, Pem by Brian S. Bauer and Wilton BarrionuevoOrosco 73 The Temple of Blindness: An Investigation of the Inca Shrine of Ancocagua by Johan Reinhard Ethnogenesis in Huamachuco by John R. Topic Coca Production on the Inca Frontier: the Yungas of Chuquioma by Catherine J. Julien Creating a Ruin in Colonial Cusco: Sacsahuaman and What Was Made of It by Carolyn S. Dean
Miscellanea

89 109 129 161

The Alca Obsidian Source: The Origin of Raw Material for Cusco Type Obsidian Artifacts by Richard L. Burger, Frank Asaro, Paul Trawick, and Fred Stross

185

iii

The Chivay Obsidian Source and the Geological Origin of Titicaca Basin Type Obsidian Artifacts by Richard L. Burger, Frank Asaro, Guido Salas, and Fred Stross 203 The Jampatilla Obsidian Source: Identifying the Geological Source of Pampas Type Obsidian Artifacts from Southern Peru by Richard L. Burger, KatharinaJ. Schreiber, Michael D. Glascock, and Jose Ccencho 225 Unifaces in Early Andean Culture History: The Nanchoc Lithic Tradition of Northern Peru by Jack Rossen 241 Lithic Provenience Analysis and Emerging Material Complexity at Formative Period Chiripa, Bolivia by David L. Browman 301 Textiles from the Lower Osmore Valley, Southern Peru: A Cultural Interpretation by Ran Boytner Corbel Vaulted Sod Structures in the Context of Lake Titicaca Basin Settlement Patterns . by Sergio J. Chave-z Archaeomagnetic Results from Peru: A.D. 700-1500 by Daniel Wolfmanand Richard E. Dodson Addresses of Authors 325

357 409 421

iv

vEditor's Preface
It is a great pleasure to present Andean Past 5, after an interval of four years since Volume 4 (1994) appeared. With 15 articles and two obituaries, AP 5 is our longest volume to date. As in Volumes 3 and 4, we present a special thematic section, this time on Inca studies. This section was inspired by a symposium at the 1993 Society for American Archaeology Annual Meeting in St. Louis, coorganized by Brian Bauer and myself. Brian kindly consented to act as guest co-editor for the thematic section, which comprises seven papers. The miscellanea section includes eight articles on a variety of topics.

Sandweiss: Editor's Preface

First, many papers follow a long-standing and productive avenue of investigation by using architecture to gain insight into Inca technology, history, organization, and even Colonial Period negotiations over ethnic identity. The other focus, also with a distinguished intellectual history in the Andes, involves the use of ethnohistorical information, often in combination with archaeological data, to study the same range of questions.

With sadness, we also include two obituariesof scholarswhose lives and careers ended far too soon. At the time of her death, Heidy Fogel had barely finished her doctoral dissertation under Richard Burger, who has written her obituary. ThoughI met Hei4y on only a few occasions,I was impressedby her seriousnessas we~las by her pleasantpersonality. Daniel Wolfman was only in his mid50s and still had plans to further develop the New World archaeomagneticrecord. Many of the data points used in his Peruvian work came fromprojects run by Izumi Shimada,the author of his obituary. For over a decade, I wouldrun into Dan at leastonce a year, in the Museo Nacional or the Pensi6n See in Lima, or at the Society for American Archaeology AnnualMeetings. I still half expectto see his familiarface and hear his loud call of "jtocayo!" across the room. Thanks to the efforts of Richard E. Dodson and Jeff Cox, we are fortunateto have a well-editedversionof Dan Wolfman'sfinalpaper on the Peruvianarchaeomagnetic curve, which we include as the final article in this volume. This publication is a fitting tribute to Dan, especiallybecause it seems likely that many years will pass before new research supersedes his results reported here. Though the seven papers in the thematic section on Inca Studies cover a wide variety of subjects,two research foci emerge.

In his article on La Centinela,the most importantlate prehispanicsite in the Chincha Valley, Dwight Wallace draws on field data gathered40 years ago when the site was better preserved than today. Wallacefocuseson the Inca sector of La Centinela, though he also makessomeobservationsabout the substantial local structures as well. Careful analysis of the layout of the Inca buildings, especially concerning access and sight lines, allows Wallace to suggest the purposes of state for which the structures were designed. Additional insights about the Inca occupation of Chincha come from observationssuch as the wall segmentwith adobes cut and fit like Inca stonework and hidden behind mud plaster. Wallace draws inferences from this construction technique,where the extra work provided no functional benefit and plaster eliminates visual symbolismas a motivation. Vincent Lee's careful analysis of the Great Hall at Inka1lacta,in Bolivia, offers insight into Inca construction techniquesand motivations. The Great Hall is well enough preserved for Lee to make a reasoned reconstructionof the original structure. The building was one of the largest in the Inca empire, and Lee calculates that it could easily have held assembliesof over 1000people. In his search for the Inca shrine at Ancocagua,Johan Reinhardintegrateshistorical, geographic, and archaeologicalinformation. Early documentsindicate that the site was one of the major temples in the Inca empire and that it had been an important

ANDEANPAST 5 (1998)
.

-vi data on coca in the Andes. The appended primary documentsare publishedhere for the first time. By carefully contextualizingher data, Julien is able to extract information about Inca economicorganizationas it relates to coca processing in the region. In a series of three articles, Richard Burger and his colleaguesreport on the loca': tion of three importantobsidiansourcesin the southernhighlandsof Peru. Material from all three had been identifiedfrom archaeological collections when Burger and his Berkeley collaboratorsbegan this line of research over 20 years ago, but only now can the sourcesbe confidently located in space. Alca is the source for obsidian formerly called the "Cuzco Type"; Chivay is the source of the "Titicaca Basin Type"; and Jampatilla is the source of the "PampasType". In additionto providing the technicalinformationnecessary for other researchers to match obsidian geochemistry to the identified sources, each article also includes a' discussion of the archaeologicaldistributionand possible significance of obsidian from each source. Due to the paucity of analyzed samples from earlier sites, most relevant information refers to the MiddleHorizonor later. However, the recent discovery of Alca source obsidianin deposits dating between about 11,000 and 10,000 BP in a Peruvian coastal site points to the potential for a very long record of use of Alca, Chivay, and Jampatillaobsidian. In his studyof lithicprovenienceat the Upper Formative Period site of Chiripa, on the shores of Lake Titicaca in Bolivia, David Browman also traces the movementof materials across the Andeanlandscape. Browman finds that the sources of minerals and stones for some light-weightobjects may have been as much as 500 km from Chiripa. Heavy building stones, sometimesweighing several tons, were apparently brought from quarries up to 80 km distant. As Jack Rossenpoints out in his article on the Nanchoc Lithic Tradition (NLT), Middle Preceramic sites from northwestern

sacred place before the Inca conquered it. Reinhard makesa compellingcasethat Ancocagua is the archaeological site now known as Mana Fortaleza, located about 140 km southeast of Cusco and recently mapped and excavated by Peruvian archaeologists. Carolyn Dean's study of the postConquest fate of Sacsahuaman, the great Inca fortress and ceremonial center above Cusco, highlights the symbolic importance of Inca architecture in the process of negotiating' identity in Colonial Cusco. She notes that Colonial Spanish, Inca, and non-Inca Andeans all used Sacsahuaman as a symbolic cornerstone in their reconstruction of history and the construction of their place in Colonial society. Especially important in this respect was the defeat there of Manco Inca in 1536. The question of ethnic identity is also central. to John Topic's analysis of late prehispanic Huamachuco in the northern highlands of Peru. Following critena established by Mana Rostworowski for recognizing Andean ethnic groups, Topic employs archaeological data to trace probable group boundaries and interaction in and around the Huamachuco area for pre-In~ times. He uses the results of this study, combined with ethnohistoric data, to show how the Inca manipulated, or even created, ethnic identities in the region.

By continuing to seek and map out the physical and toponymic manifestations of Inca shrine systems in the south-central Andes, Brian Bauer and Wilton Barrionuevo provide data crucial to evaluating ideas about this important aspect of Inca social, political, and religious organization. The Anta shrine system reported here is less well documented than the Cusco system that Bauer has also studied. However, it does provide a useful comparison while suggesting that the Cusco example is not unique. Catherine Julien's detailed analysis of early Colonial coca production in the yungas of Chuquioma, in Bolivia, is a welcome addition to the growing corpus of ethnohistoric

vii-

Sandweiss: Editor's Preface

last volume, all of them positive. It is a pleasure to welcome our new Graphics Editor, David Fleming. Though he formally joined us late in the process of preparing AP 5, his contributionsare already evident in this volume. Congratulationsare also due to Editorial AdvisoryBoard member Craig Morris, who, amongother honors, has recentlybeen elected to the National Academyof Sciencesand the AmericanAcademyof Arts and Sciences. On behalf of the permanent editorial staff, I would like to thank guest co-editor Brian Bauer for his importantrole in acquiringand Ran Boytner's analysis of textiles in . editing the papers in the thematic section on the lower Osmore drainage of southern Peru Inca Studies. I also thank Doris Kurella for is concerned with the archaeologicalrecogni- German proofreading. tion and differentiationof ethnic groups. In In closing, I would also like to recogthis valley, just inland from the port of Ilo, two archaeologicalcultures(Ilo-Tumilaca/Ca- nize, with gratitude, the tremendous efforts buza and Chiribaya)co-existedfor about 300 made by our reviewers, authors, Editorial years early in the Late Intermediate Period. Advisory Board, and above all my fellow Boytner's examinationof textiles assigned to editor Monica Barnes. Together with the the two cultures suggestsa commonhighland continued support of the Cornell University origin for both, but also supportsthe infeJ:ence Latin American Studies Program, it is the made from other data sources that the two hard work and collaborativespirit of all these were distinct groups while co-residentin the people that make Andean Past possible. lower Osmore valley. Daniel H. Sandweiss In a fine example of Andean ethno- Universityof Maine archaeology,SergioChavezprovidesa wealth 23 October 1998 of informationon the constructiontechnology and function of corbel-vaultedstructures and on modem settlementpatterns in the Peruvian altiplanonear Lake Titicaca. He identifiesthe environmental, material, technological, and social correlates of these building types. Chavezthen appliesthe results of his studyto the historic and late prehistoric record of the region, with special attention to the significance of dispersed v. nucleated settlements. Although sod structures (with one possible exception) have not been identified archaeologicallyin the altiplano,Chavezbelievesthat they are very likely quite ancient. Based on his ethnoarchaeologicalresearch, he provides several indirect archaeological criteria that may signal the use of such buildings. As expected for an active institution, Andean Past has seen some changessince our

Peru north as far as Panama are often dominated by unifaciallyworked stone tool assemblages. Thanks to Rossen's detailed analysis of the Nanchoc materials, recovered from sites on the forested western slopes of the Andes in the Upper Zafia Valley of northern Peru, the NLT is one of the best-studiedof theseunifacialassemblages. Amonghis many results, Rossen suggests that unifacial lithic assemblagesindicate more diverse and plantorientedeconomieslinkedto greater sedentism and an onset of intensification.

HEIDY FOGEL, 1956-1994

Richard L. Burger
Yale University

tained while a graduate student barely seemed' to slow her down. Heidy had an intense passion for life, which she expressed through her art, dancing, knitting, and unique style of dress. I always thought of Heidy as a force of riature--unpredictable and unstoppable. Heidy was born in Boston, but spent most of her childhood and teenage years in suburban Newton, Massachusetts. Although she rarely spoke of it, her parents were Holocaust survivors, her mother having been interned in Auschwitz and her father having spent the war in a series of Nazi death camps. Throughout her life, she remained unusually close to her parents at).d two brothers, and continued to maintain tight friendships with her high school classmates. Heidy had natural artistic talent and she honed her skills as a scientific illustrator at the Rhode Island School of Design, from which she received a B.F.A. in 1978. For much of her adult life she thought of herself as an artist, rather than a scholar, and her drafting skills served her well in her own research, as well as being an important source of support during her graduate studies. Shortly after graduating from RISD, Heidy began to pursue her long-standing interest in archaeology by working as an archaeological illustrator and exhibits assistant at the Grosvenor Museum in Chester, England and then as a crew chief for an archaeological survey in Israel, sponsored by the Rockefeller and Israel Museums. While supporting herself as a technical artist in Boston, Heidy continued her involvement in archaeology, drawing upon the resources of a neighboring institution of higher learning, Harvard University. In the summer of 1979 she joined Harvard's Mount Jasper Project and helped supervise the survey and mapping of a quarry site in northern New Hampshire; the following year she went to work as an archaeological illustrator at the Harvard Peabody Museum's Institute for Conservation Archaeology.

When Heidy Fogel died on November 11, 1994 at the age of only 38, Andean archaeology lost its leading authority on the Gallinazo culture. I first met Heidy in 1983 when she came to New Haven to inquire about Yale's M.A. Program in Archaeological Studies. Fresh from her work as staff artist and crew member of the Proyecto Chimu Sur in Casma, she was full of excitement and enthusiasm about Peruvian archaeology both as a quest for scientific knowledge and as a social field teeming with wonderfully eccentric personalities. During her short interview, she filled my office with energy and anecdotes, and her departure left a strange calm, like the one that descends over a devastated landscape following a cyclone. Heidy never lost her astonishing vitality, her love of the field, or her engagement with the people who constitute it. Even the two head-on autoniobile collisions that she sus-

ANDEAN PAST 5 (1998):1-3.

ANDEAN PAST 5 (1998) As a Harvard employee, she began taking classes in 1981 as a Special Graduate Student, including a course on Peruvian archaeology that had a great impact on her. Taught by Gordon Willey, this course caught Heidy's imagination. Professor Willey remained a source of inspiration and counsel for Heidy throughout her life, and his friendship served to link her directly to the Vim Valley Project. Through Heidy's Harvard connection and her contact with Harvard faculty such as Geoffrey Conrad and Garth Bawden, she became aware of the intense excitement being generated by the large projects on Peru's north coast that involved students and faculty from Harvard, as well as many other universities. This led naturally to her departure early in 1982 to spend four months in Casma working on the investigations of Carol Mackey and Ulana Klymyshn at Chimu administrative centers in the Casma Valley. In 1983 Heidy began the M.A. program in archaeology at Yale; although I served as her primary advisor, she also worked closely with Mike Coe and Frank Hole. From her arrival at Yale until her departUre in 1988, she was closely associated with the Division of Anthropology at the Yale Peabody Museum of Natural History. She took on myriad tasks there with her characteristic energy and enthusiasm. She helped catalogue collections, supervised undergraduate assistants and volunteers, and prepared text for and assisted in the preparation of exhibits. In one unforgettable affair, she was suddenly whisked off to Pittsburgh by the FBI to identify property that had been stolen from the Museum. Heidy helped the FBI to crack the thiers artifact coding system and to discover that he had robbed five other prominent museums in addition to ours. The focus of Heidy's own original research gravitated to one of the Yale Peabody's most important Andean archaeological collections-the pottery recovered by Wendell Bennett during his excavations in the Vim Valley. Although Bennett had published some of these materials in his usual prompt and efficient manner, he had only scratched the surface of the collection's potential and Heidy decided to

-2
make the substantial sample of Gallinazo ceramics the subject of her master's thesis. She utilized the stratigraphic information recorded by Bennett as the framework for reanalyzing this pottery, concentrating on forms and their association with decorative patterns rather than using the type-variety method employed by the Vim Valley Project. Her master's' thesis entitled "The Gallinazo Occupation of the Vim Valley, Peru" provided a lucid and well-illustrated three-phase ceramic chronology for the Gallinazo culture that was anchored in well-defined stratigraphic relationships. Unlike Bennett's ceramic chronology, Heidy's sequence could be used to date the full range of elaborate and utilitarian vessels, even those found individually in graves or in small samples on site sUrfaces. This chronological control made it possible for her to trace the changing pattern of settlement in Vim during Gallinazo times and this in turn led to numerous questions about Gallinazo socioeconomic and 'political structure beyond the Vim Valley. On the basis of her success in the archaeology program, she was admitted to the doctoral program in Anthropology at Yale, and she completed her coursework and comprehensive exams in 1987. In 1985 and 1987, she returned with support from the Williams Fund, Sigma Xi, the Hazard Fund, and several Joseph Albers' Traveling Fellowships to Peru's north coast to visit the sites described by Bennett, Willey, and the other members of the Vim Valley Project. Finally, in 1990 she was awarded a Fulbright-Hays Fellowship to spend the year in the Trujillo area re-surveying the Gallinazo sites in Vim and comparing her observations from this valley with sites and materials from neighboring valleys on the north coast. The sequence she had developed allowed her to date the rise of the Gallinazo culture in Vim and the subsequent expansion of Gallinazo traits into the adjacent drainages. Heidy was convinced by her research that she was observing the rise of Peru's first multivalley state and that the Gallinazo Site constituted the large urban capital of this polity. She was also able to use her 1987 restudy of the Vim Valley Project Gallinazo gravelots in the collection of the American

3Museum of Natural History to confirm independently her sequence and to add a new dimension to her reconstruction of Gallinazo socioeconomic organization. In her doctoral dissertation Settlements. in Time: A Study of Social and Political Development During the Gallinazo Occupation of the North Coast of Peru, presented in 1993, Heidy convincingly argues that it was the Gallinazo rather than the Moche culture that pioneered large-scale statecraft on the north coast, and that it was only with the collapse of the Gallinazo state centered in Vim that the focus of power . shifted to the Moche Valley at the beginning of Moche III. Support for her argument came from the redating of Gallinazo sites in Vim, Moche and Santa and the study of changing Gallinazo settlement patterns in each of these . valleys. Glimpses of Heidy's insights were offered at numerous conferences and her ideas helped to spark the renewed interest in the Galliriazo culture that has been evident since the mid1980s. Unfortunately, .she did not publish her findings and her master's and doctoral theses remain unknown to most scholars. She did prepare sections on the Gallinazo Culture (vol. 3, p. 9) and the Gallinazo Group Site (vol. 3, pp. 9-10) for the Encyclopedia of Latin American History and Culture (Charles Scribner and Sons, 1996) and was in the process of finishing an article on the chronology and political organization of the Gallinazo Culture for Latin American Antiquity at the time of her illness. After completing her dissertation, Heidy moved to Washington to be with her husband, Hal VanGieson, an economist she met at Yale while both were graduate students. She took a job as a Senior Archaeologist at Engineering Science, a private firm involved in culture resource management in the U.S. and Latin America. While there, Heidy became engaged in project management, data analysis, and the computer mapping of both prehistoric and historic sites. She wanted to fmd a position at a museum or in academia, but her sudden illness cut short these hopes. In February of 1994, it was discovered that Heidy had an advanced form of cancer. At the time, she was

Burger: Heidy Fogel four months pregnant. Without regard to her own safety, she delayed chemotherapy until after the baby was born. Despite the increasing severity of her illness, Heidy remained optimistic to the end that one day she would be able to return to the work and people she, loved so much. On May 26, 1994 she delivered a healthy baby girl, Julia Rose Van Gieson. In the months following the birth of her daughter, Heidy's health deteriorated rapidly. She died in the same dramatic and powerful way that she lived. Those of us who had the opportunity and privilege of knowing Heidy will never forget the experience. It remains my hope that her lasting contribution to Andean archaeology will be assured by the future publication of her dissertation, so that the enormous efforts she made during her short career can serve as a sound basis for the continued growth of our understanding of the Gallinazo culture and society. Acknowledgments In preparing the above obituary, I have drawn upon materials provided to me by Heidi's husband, Hal VanGieson, and her friend, Sue Benaron.

DANIEL WOLFMAN, 1939-1994 Izumi Shimada Southern Illinois University

On November 25, 1994, Daniel Wolfman died unexpectedly of heart failure in Santa Fe, New Mexico at the age of 55. Dan was at the forefront of archaeomagnetic research in the U.S. and for the past two decades was the principal "mover and shaker" in establishing master curves for Mesoamerica and Peru. ANDEAN PAST 5 (1998):5-8.

His name and bushy, mottled beard may stir up a certain image, but we knew Dan as a congenial, tireless worker in the field, always eager to collect new archaeomagnetic samples. He devoted much of his professional career to establishing master archaeomagnetic curves for various regions in the New World and a

ANDEAN PAST 5 (1998) self-supporting, state-of-the-art laboratory of his own at the Museum of New Mexico, Office of Archaeological Studies in Santa Fe (1988-1994). His methodology and results have been described in a series of important articles (Wolfman n.d., 1984, 1990a; 1990b; Wolfman and Dodson 1998). He loved Latin America, its archaeology, people, and food, and he projected a return trip to Peru in the near future. Dan was born in New York City on February 8, 1939, and received a B.A. with Distinction in Mathematics from the University of Rochester in 1959. His undergraduate education in science would provide a strong basis for his research in archaeomagnetism and archaeometry in general. His interest in archaeology was triggered in 195960 while at the University of Chicago pursuing mathematics as a Woodrow Wilson Fellow. In 1960, with the support of an NSF Summer Training Fellowship, Dan attended field school in Oaxaca directed by John Paddock of Mexico City College. He gained additional archaeological field experience as an assistant in the Wetherill Mesa Project at Mesa Verde and in the Navajo Dam Project, both in 1961. D~ began his graduate training at the University of Colorado in Feburary, 1962, and received his M.A. in anthropology in 1963. He was employed as the Field Director of the Picuris Archaeological Project (1962-1966) in New Mexico, sponsored by the Fort Burgwin Research Centers in Taos, and taught as an assistant professor of anthropology at the University of the Americas in Mexico from 1966 to 1968. He returned to the U.S. and in 1969 worked as an archaeologist for the Nevada Archaeological Survey. It was during 1968-1973, when he worked under Robert L. DuBois as a Research Associate at the University of Oklahoma Archaeomagnetism Laboratory, that Dan decided to dedicate himself to archaeomagnetic dating and the attendant slow process of establishing master sequences in various regions of the New World. It was an opportune moment for Dan because the technique of archaeomagnetic dating was

-6
increasingly recognized in archaeology as an important complement to the rapidly burgeoning radiocarbon dating technique. Recent improvements in computers made processing and publication of data quite manageable. DuBois and Dan collaborated in presenting and publishing a series of joint

papers on archaeomagneticdating of Latin '


American materials. In 1973, Dan presented his doctoral thesis entitled A Re-Evaluation of Mesoamerican Chronology: A.D. 1-1200. This study effectively demonstrated the value of archaeomagnetic dating. Dan was firmly convinced of the value of archaeomagnetic dating as a reliable and reasonably inexpensive technique for both relative and absolute dating. This technique requires independent calibration, commonly with .radiocarbon dating, for the establishment of a regional master curve. However, once the curve is constructed, it can serve to evaluate the relative merits of conflicting radiocarbon dates. That its samples derive from baked features that are highly durable and closely linked to specific human activities increases its value to the field of archaeology. At the same time, recognizing that his primary research goal of establishing archaeomagnetic curves in Latin America could not be accomplished without the awareness and collaboration of archaeological colleagues, Dan quizzed us tirelessly at professional meetings to keep abreast of current and future field work that might provide archaeomagnetic samples. With many colleagues single-mindedly focused on collection of radiocarbon samples and diagnostic ceramics, Dan worked hard to make them aware of archaeomagnetic samples and dating. In Peru, for example, he eagerly gave talks to interested archaeologists illuminating the virtues of archaeomagnetic dating, and trained several American and Peruvian students (e.g., Alvaro Higueras and Glenn Russell) to collect samples. In 1986, he received a travel grant to Peru from the United States Information Agency, Academic Specialist Program to conduct seminars on archaeomagnetic dating techniques and train local archaeologists for sample collection. This effort was preceded

7-

Shimada: Wolfinan both in and out of field. From the 1980s to the time of his death, he served as a member of the Executive Committee of the Society for Archaeological Sciences, Advisory Committee of the Southern Methodist University Radiocarbon Laboratory (then directed by Herbert Haas), and the National Science Foundation Advisory Council for Archaeometric Technology. His untimely' death was a great loss not only to the nascent field of archaeomagnetic dating in Latin America, but also to broader archaeometry and archaeology. Dan's proudest achievement and long-time dream came true in 1993 when the Archaeomagnetic Dating Laboratory under his direction opened its doors at the Museum of New Mexico in Santa Fe. Being able to live in scenic Santa Fe was the icing on the cake. For .many years he struggled on a shoe-string budget to keep his archaeomagnetic research going while teaching at Arkansas Technical University (Assistant Professor of Anthropology, 1973-80; Associate Professor, 198085; Professor, 1985-1988) and at the University of Arkansas (Assistant Professor of Anthropology, 1973-79; Associate Professor, 1979-85; Professor, 1985-1988). At the time of his death, he and his assistant, J. Royce Cox, were busy processing samples to refine and extend the archaeomagnetic curves for Peru, Mesoamerica, and the U$. Southwest. Clearly, Dan left his long-lasting imprint on American archaeology in his sustained commitment to improve dating techniques, particularly archaeomagentic dating, and better integrate archaeometry in our research in and out of field (see also the obituaries by Schaafsma and Schaafsma [1996], with a complete Wolfman bibliography, and by Sternberg [1996]). The Archaeomagnetic Dating Laboratory in Santa Fe is his most noteworthy legacy and Dan's long-term effort to establish the master archaeomagnetic curve for Peru will be carried on by J. R. Cox. It is an effort that both Andeans and Andeanists should continue to support.

and followed by similar trips to Guatemala and Mexico in 1985 and 1987, respectively. Because of its numerous sites with wellpreserved adobe, quincha (wattle and daub) and other clay-bearing constructions and floor features, and long occupation spans, Dan recognized the great archaeomagnetic potential of coastal Peru. Most important in the effort to establish the master archaeomagnetic curve for Peru was Dan's collaboration with R. E. Dodson, then geochronologist at the Rock Magnetism Laboratory, University of California at Santa Barbara. Supported by a grant from the National Science Foundation in 1982-83, Wolfman and Dodson collected samples from archaeological sites on the coast and in the highlands. During this time, 134 samples were collected at 37 sites, with 22 sites on the North Coast contributing 102 samples. This work provided the basis for an archaeomagnetic curve spanning ca. A.D. 6501500 presented in their important reference work (Wolfman and Dodson 1998). Members of the Sican Archaeological Project have fond memories of Dan extolling the virtues of archaeomagnetic datinR while working against the clock to collect 24 samples at the site of Huaca del Pueblo Batao Grande. The site has a "textbook stratigraphy" (ca. A.D. 500 to the present) "with at least 42 occupational floors, many associated with well-preserved hearths, arsenical bronze smelters, and other bumt clay features suited for archaeomagnetic dating. During their various stays with us in Batan Grande, Dan and Rich surprised us with welcome gifts of beef tenderloin brought in from the nearby city of Chiclayo. Their congeniality and comments reflecting their broad scientific and archaeometric experience enriched our otherwise routine field work. Dan returned to Batan Grande in 1989 to collect additional samples from superimposed ceramic kilns in Poma Canal that would push back the archaeomagnetic curve to ca. 1000 B.C. (Wolfman et al. n.d.). Though Dan was best known to us as an archaeomagnetic specialist, his interests and expertise covered many dating techniques. He was an energetic promoter of the integration of archaeometry in archaeological research,

ANDEAN PAST 5 (1998)


References Cited
Schaafsma, Polly and Curtis Schaafsma 1996 Daniel Wolfman 1939-1994. American Antiquity 61(2):291-194. Sternberg, Rob 1996 Daniel Wolfman: 1939-1994. SAS Bulletin 19(3/4). Wolfman, Daniel n.d. A Re-Evaluation of Mesoamerican Chronology: A.D. 1-1200. Unpublished Ph.D. thesis. Department of Anthropology, University of Colorado. 1973. 1984 Geomagnetic Dating Methods in Archaeology. In Advances in Archaeological Method and Theory, Vol. 7, edited by M. B. Schiffer, pp. 363458. New York: Academic Press. 1990a Mesoamerican Chronology and Archaeomagnetic Dating, A.D. 1-1200. In Archaeomagnetic Dating, edited by 1. L. Eighmy and R. S. Sternberg, pp. 237-260. Tucson: University of Arizona Press. . 1990b Retrospect and Prospect. In Archaeomagnetic Dating, edited by 1. L. Eighmy and R. S. Sternberg, pp. 313-364. Tucson: University of Arizona Press. . Wolfman, Daniel and Richard E. Dodson 1996 Archaeomagentic Results trom Peru: A.D. 700-1500. Andean Past. 5: ?-? Wolfman, Daniel, Ricliard. E. Dodson, and J. Royce Cox n.d. Refinement and Extension of the Archaeomagnetic Curve for Peru. Manuscript in preparation. .

-8

THE INCA COMPOUND AT LA CENTINELA, CHINCHA Dwight T. Wallace SUNY-Albany Introduction La Centinela is a large site at the northwest edge of the floor of the Chincha Valley, Peru, very near the ocean and the bluffs that separate the fully utilized flood plain from the higher pampa (Figure 1). The Chincha Valley floor is a triangular, delta-like flood plain, roughly 25 km on a side, of very rich and easily irrigated farmland. The western and southern edges of the area on the bluffs did have additional human settlement in pre-historic times, and this area has a very extensive and dense population today. La Ceptinela dates from the Late Intermediate Period, A.D. 12501470, through the Inca-dominated Late Horizon, from A.D. 1470 until the Spanish Conquest in A.D. 1532. . Working at L~ Centinela in 1900, Max Uhle reported the many building units that formed the extensive ruins (see Uhle 1924: plates 2-5). The site (Figure 2) is dominated by a pyramid at least 18 m high set on a platform approximately 12 m high that forms a terrace at the foot of the steep southern side of the pyramid. The Inca later constructed administrative buildings in front of the southern terrace (Figures 2 and 3) and also across the terrace itself, many using the distinctive Imperial Cusco architectural style; From this focal location a set of five straight roads radiated out across the valley floor, the two outer roads having alignments very close to true east and south (Figure 1). One diagonal road crosses the entire valley floor and continues straight across the pampa to the Pisco Valley, where it meets a road up-valley to the highlands and another that continues south to the Ica Valley (Wallace 1991; map adapted by Hyslop 1984: figure 21.4). These physical features indicate the prehistoric socio-cultural importance of the site, an importance which is further reflected by the fact that it was the largest ceremonial structure
ANDEAN PAST 5 (1998):9-33.

south or east of the famous central coast temple of Pachacamac during the late prehispanic periods. The obviously symbolic layout of the road system and its focus on the main pyramid fit the known function of La Centinela as the pilgrimage center for a famous oracle, serving all areas -- coast and highlands -- to the south and east. It was considered a "daughter" of the even more famous Pachacamac oracular center (Menzel and Rowe 1966; Patterson 1985). The Chincha oracle was established before the arrival of the Inca and continued in use through the period of Inca control. There is also documentary evidence of specialized Chincha merchants maintaining a very farflung trade network (Rostworowski 1970, 1977; Morris 1988; Wallace 1978a, 1978b), giving an economic facet to La Centinela's pre-Inca function as the center of the powerful Chincha socio-political unit. This economic role would also explain why the Inca developed La Centinela as their major political administrative center for the south central coast (but cf Sandweiss 1992:10). Inca use of rectangular adobes identifies their architecture in Chincha, because only tapia, or puddled adobe, was employed before. The main Inca construction area on the southern platform or terrace is shown on the plan in Figure 3. The set of buildings constructed by the Inca during the period they controlled Chincha includes a large rectangular ground level plaza in front of this southern platform, centered on the pyramid (facing the lower right comer of Figure 2). The full layout (see Santillana 1984: figure 1) also included a small set of compounds built directly on the east side of the main platform, just to the right of the plan, but at a somewhat lower platform level. In front of this eastern group is a construction wing that matches the one shown on the lower left of the present plan (Figure 3, see also Figure 2, lower center). There are also additional buildings

ANDEAN PAST 5 (1998)

-10 The back or western end of the compound is taken up mainly by an apparently open area that will be referred to as a veranda (Figure 4: Area 10), from which there is a magnificent view of the beach and ocean (see also the reconstruction drawing in Figure 8). The veranda has a sub-floor water cistern and drain,

around the ground level plaza, a few of which show at the lower right in Figure 3. Rectangular adobes that are visible near the top of the main pyramid, on top of puddled adobe walls, show that some modifications or repairs to the earlier construction were also made during the Inca occupation. The result was probably a new southern front to the main pyramid itself, the focal point of the entire complex. As could be expected for Incadirected construction, the result was a very orderly, well balanced, and visually impressive religious and administrative center, although it should be noted that neither high pyramidal building bases nor pyramids were common Inca architectural features. Most of the observations here concern the compound with the central set of raised rooms (Figure 3: Compound C; Figure 4) that is at the southwest corner of the large southern terrace, already noted as part pf the Inca construction on the terrace. Uhle considered this "palace" a focal structure, and Santillana (1984:19) gives a .cogent argument for the relative importance of this area and its associated courtyard (Figure 3: Courtyard B) and entrance plaza (A). To preview the construction, a set of rooins (Figure 4: Area 6) raised to second story height by a solid basal terrace is a dominant feature of the compound, a view of which can be seen in the photo taken in 1900 by Uhle (1924: plate 4), included here as Figure 5. (Uhle's photo can be compared with my own 1957 photo in Figure 6, showing that hardly anything had been moved in 57 years!) In Figure 7, the reconstruction based on these photos, the tall-niche gallery wall (Figure 4: Area 3) at the east side of the raised room block can be clearly seen. One niche is actually a door that opens into some small rooms (Figure 4: Area 4) against the base of the raised room block. Various double jamb doors and a rare geometric painted mural in one room emphasize the importance of this central warren of raised rooms. An eastern, or frpnt door into the upper room block has access from the niche gallery by a roundabout way up a stairway and along a ledge.

and there are small rooms on both the north '


and south ends of the veranda; the niches and double jamb entries mark these rooms as more than minor storage areas, and one small southern room (Figure 4: Area 9) has steps down into a tank with a drain, undoubtedly to be used for bathing. Uhle (1924:78) traced both drains to outlets outside the compound. There is also a back or western entrance (Figure 4: Area 7) to the raised room block off the veranda, reached by a stairway starting at the entrance to the southern rooms (Figure 4: Area 8). The corridor entrance to the veranda (Figure 4: Area 2) and the two relatively large northern rooms or enclosures (Figure 4: Areas 1, 12) complete the spages in the main compound. Apparently all interior and exterior wall surfaces were painted white over a fine clay plaster surface. Inca and Pre-Inca Wall Art At least two decorated walls are known at the site, one associated ,with the pyramid looming high above the terrace with its Inca construction, the other in the main Inca compound just discussed. Their contrasting decorative techniques and building associations emphasize the contrast between the local preInca source of the Centinela complex and the distinct Inca-style construction and art in the extensively remodeled sector where the Inca established their administrative center. At some unknown date between Uhle's visit in 1900 and the 1957 survey, a small room on the west side and near the top of the main pyramid had been cleared, revealing a 3dimensional frieze (Figure 9) on the inside of the outer walL The frieze is pIano-relief, that is, the flat face has been cut to a fixed depth (about 5 cm), giving only two surfaces: the foreground design and the excised background. One section retained a height of nearly 2 m, the original full height judging

11from the nature of the design. The erosion of the wall into very large blocks makes it clear that it was tapia-constructed, as are all Late Intermediate Period buildings in the Chincha Valley. Tapia consists of adobe (prepared mud) that was either poured directly into forms, much like modem concrete, or was mixed and packed in place between the forms, like modem tamped earth. Each pouring cre~ ated a fairly large rectangular block. The process was then repeated on top of and/or at the side of the previous blocks until the desired height and width were reached. The process left joints that tend to erode so that the size of each block is quite evident. From appearances, the adobe forming the design is firmly attached to the wall, so that it might have been created during the molding of the wall itself. However, it seems more likely that the design was cut into the still plastic surface, was hand modeled by adding clay to the damp surface, or was produced by applying pieces or sections that had been formed in molds. The elements in the design on the wall"are similar to ones found on local textiles, including the scrolled fret in curvilinear fomi (O'Neale and Kroeber 1930: plates 37, 40) and also in squared form (ibid.: plate 43c), the diamond-shaped ray-like figure with eyes and what looks like a V-shaped mouth (ibid.: plate 43c) and the fret of bird heads of the same date and provenience (ibid.: plates 32, 42a). The general diagonal and diamond-shaped lay01,ltis found on textiles, pottery, and basketry, so that the motifs of the wall frieze are well-established as south' coast designs of the Late Intermediate Period and Late Horizon. Scraps of shaped adobe from what were probably similar pIano-relief friezes were found at two other late sites (PV57-20, PV5797) and an entire frieze has been uncovered more recently at Litardo Bajo (PV57-80) during the continuing work in Chincha in conjunction with that directed by Craig Morris (1988). Interestingly enough, adobe sculpture of this type and date has not yet been found elsewhere on the south coast (or, to my knowledge, the central coast), but it is well known and abundant on the north coast, especially at the Chimu capital of Chan Chan

Wallace: La Centinela

(Pillsbury 1993). The piano-relief technique, the diagonal layout, and the use of many geometric and simple animal elements can also be found in the north coast decoration. The similarities could be due to convergences in design and technique of rendering,' because many of the styles in the Central Andes of this later period did consist of uncomplicated and/or geometric designs, often incorporating' images of animals or animal heads, especially simple bird heads. Actually, the pIano-relief type of adobe sculpture is somewhat more limited than might be expected, given that adobe was used in coastal architecture for several millennia. In any case, the recent ethnohistorical evidence for direct trade ties be-' tween Chincha and the north coast (Rostworowski 1970, 1977; Morris 1988; Pillsbury 1996; Sandweiss 1992; Wallace 1978a, 1978b, 1991) suggests the possibility of historical connections between the decorative art of the two areas; an important subject for future research.

In contrast to the local style of pIano-relief adobe friezes and their Chincha style designs, both unknown in strictly Inca style structures, there is a painted wall in at least one room of the raised room block ("murals" in Figures 3, 4); it is attributable to the Inca, both in context and design style. The design (Figure 10; see also Bonavia 1985: figures 113, 114) is strictly an angular geometric one that fits the largely geometric decoration found on Incastyle textiles and ceramics. Inca mural painting is virtually unknown due to the poor preservation of most wall surfaces in Inca buildings, especially in the highlands. Therefore, the unique Centinela example takes on added importance. Painted murals have a long preInca history on the coast. Best known are elaborate examples associated with the north coast Moche culture of the Early Intermediate Period (see Bonavia 1985). A feature of both murals that is important in a closer consideration of their function is that both occur in locations not readily accessible from any truly public exterior space. Rather, both are in places seemingly designed to provide privacy from general viewing. The rooms or spaces in which they occur are them-

ANDEAN PAST 5 (1998)

-12 the reconstruction drawing in Figure 8 show that a "correction" was made. along the line following the east side of the veranda (Figure 4: Area 10), including the entrance from the central passageway (Area 2). Walls were made with a wedge-shaped plan, that is, thicker at one end than the other. As a result, . the only trapezoidal interior spaces are the short entry passage (Area 11) to the small rooms at the northwest corner of the building, and the small entry space (Area 8) that leads to the rooms at the southern end of the veranda. Nearly all other affected rooms and spaces are parallelograms in plan, not rectangles (nor trapezoids), so that opposite walls are parallel. The only possible exception occurs on the veranda, where the upper back wall of the raised rooms follows the northsouth and east-west grid. However, if the an-

selves limited in size, more appropriate for only a few viewers at a time. Both can therefore be considered as elite art. These areas are not directly associated with any archaeological features of more specific function other than a general religious and administrative nature, and neither the design content nor imagery of either mural suggests any symbolism beyond that of a decorative nature. On the other hand, Inca geometric motifs, on textiles in particular, are known to have had associations with specific elite Inca kin units and, in some cases at least for the Inca emperor, specific rituals. As murals, such art was apparently not very common, so its mere occurrence could have provided an aura of elitism and authority. Hiding Architectural Disorder? During the 1957 survey of' Chincha, I made a. plan of the Inca construction on the platform on the south side of the pyramid proper (Figure 3; also adapted by Hyslop 1984: figure 7.10). To start the plan, both wall lengths and c<?mpassreadings for three sides of the compound were taken at the SE and NE corners, but not for the western side, where the 12 m drop off the back terrace did not encourage such measurements. These figures gave three sides of the expected rectangle, but the back wall was not at right angles to the sides. What seemed like botched measurements turned out to be quite accurate on reexamination, the back wall having been aligned with the pre-existing outer terrace wall. The east-west alignment of nearly all the Inca reconstruction, including the south face of the older pyramid itself, is only 2 degrees off a true cardinal alignment, an obviously intentional, symbolic alignment that the Inca are known to have used for important structures (Hyslop 1984:313, figure 31.4: 1990:235). So there is little question why most of the Inca walls were not aligned with the earlier terrace wall: it simply did not fit their strongly preferred orientations and could not have been changed without the major task of rebuilding the high terrace edge. Obviously, something had to give in making the ideally oriented plan fit with the existing outer wall. The plan in Figure 4 and

gled . alignment of the retaining wall for the


back stairs and walkway to the raised rooms were of full height, forming a curtain wall closing off the entire stair and walkway entry to the rear door (see the dotted lines in Figure 8), then all the opposing veranda sides would have been parallel. If this wall had even been high enough to form only a balustrade, as is very likely, then the same effect would have been at least partially given. Hyslop (1990:5-7, 21-15, 234-243) brings together considerable information on Inca architectural forms. Truly rectangular building and compounds (kanchas) are nearly universal. Rounded walls are used on rare occasions, for unquestionably intentional visual contrast, with requisite symbolic connotations. Plazas are often rectangular, but occasionally trapeziform ones (no sides parallel) are probably alignments with astronomical symbolism. True trapezoids are normally limited to door, window, and niche shapes. Hyslop notes that rectangular form is found in Inca orthogonal street and compound wall layouts, but partial deviation or even lack of any apparent order can occur when there are physical hindrances. A fascinating clay architectural model of a trapezoidal structure along with the topography of its site is reported by Sandweiss (1995:151 and figure 129) inside a building of probable Incaic date

13at Tucwne on the north coast. Circular alignments occur at a few notable sites, based on the radial segmentation which follows a conceptual map of kinship organization and associated ritual symbolism. In short, rectilinearity can be shown to be the underlying principle in Inca architectural planning, but intentional deviations for both esthetic and symbolic purposes can be identified, with no apparent resistance to deviations when environmental obstacles make them unavoidable. In short, impose a proper order whenever possible, but give some leeway for choosing the logic of the pattern when practical problems or physical limitations occur. In the Centinela case, the most obvious solution would have been to put an ideally aligned and rectangular compound abutting the misaligned terrace edge, leaving a triangular dead space between the back wall and terrace edge. Or, less ideal but most spatially economical, would have been to follow ideal alignment for all walls ex~ept the western outer one; the angles of the back comers would not be visually obvious from any outside angle. However, the unequal comer angles and non-parallel walls of differing lengths of the veranda would have been noted by anyone with any spatial sensitivity, an architect or master mason in particular. A more complicated "corrective" plan having been chosen, we are left with the inference that the adjustments were consciously made to minimize the perceived irregularities of trapezoidal room shapes in favor of a parallelogram, with facing walls of equal length and with the deviation from the ideal right angles distributed equally between four somewhat off-right angle corners, which adds up to a practical and somewhat more subtle masking of the deviations from the ideal. The clear Inca preference for trapezoidal doors and niches is an interesting contrast with the equally preferred rectilinearity. There is, in fact, evidence for the ultimate contrastive comparison in Inca world view, involving visual and conceptual perceptions of ideal form, as suggested by the order so basic to Inca worldly affairs and the unpredictability (a type of disorder) with which they seem to have

Wallace: La Centinela

viewed nature and its deities. This view was apparently symbolized physically by their highly ordered plans reflected in both sociocultural organization and material culture vs. the many irregularly shaped, natural boulder and large rock outcrops that were often the foci of worship. The intiwatana at Machu Picchu is a particularly interesting case in that it consists of a small natural rock outcrop that' was entirely altered from its natural state, but reshaped into a completely abstract form with many planes, a case of manufactured disorder. As remote as all this may seem from the simple question of why a veranda and some rooms were parallelogram-shaped rather than trapezoidal, the discussion does support the claim that the architect of the central Inca building at La Centinela was not only motivated by a desire to show that he could produce a geometrically neater structure, but that this '''neatness'' would be intended to enhance his product as a fitting center of power for the administration of the Inca state. I urge that similar analyses of other Incaic sites be done to test the interpretations made here. Architectural Access as Power Symbolism Visually dramatizing important buildings or spaces by manipulation of architectural elements is a very widespread practice. Monwnental features such as steeples or elevated bases are familiar examples, although monumentality includes symbolic impact, not merely physical prominence. Power, in one or more forms, is being signified in all cases. Control of access to some physical or abstract seat of power is a common feature, although two distinct factors are in operation: the value of encouraging public participation vs. that of restricting interactions in order to increase organizational efficiency or to heighten perceptions of authority and power. An admire but don't touch or enter approach, or one where the public is barred only from a final area or a central, ultimate visual symbol actually fits both objectives. Domestic privacy still remains a common goal in architectural layouts, but some difference in architectural treatment, such as a simple, remote entry and/or features of the interior rooms, ought to distin-

ANDEAN PAST 5 (1998)

-14
with high values signifying restricted access. Hammond found high indices for areas of connected rooms and patios that lacked monumental construction and had indications of domestic activities. In contrast, temple mounds and their adjoining plazas were highly accessible (at least up to their plaza bases). It appears that the architecturally least accessible places housed only domestic activities, providing an interesting contrast with the Chimu case. At the main compound at La Centinela, there is an inescapable question concerning the strikingly limited access to the raised set of rooms by either the eastern "front" or the western "back doors". To trace the paths (see Figures 3, 5, 6, 7), first is entry up the ramp and through the outer gate into the plaza (Figure 3: A) and then into the inner courtyard (Figure 4: B), both points of restriction and potential control. The single entry to the building compound is via a short stairway and simple doorway (Figure 4: Area 1), which curiously lacks the double Jamb that marks Inca state architecture. The first room or enclosure has many small niches in the rear wall, giving a connotation of importance, whether the niches were for practical use or decoration; no other features suggesting a particular function were visible in 1957. From there, the single, nearby exit leads into the central corridor (Figure 4: Area 2), at which point there are two ways to proceed. To reach the eastern front of the building, a left turn into the corridor (Figure 4: Area 2) leads to a door into the gallery of tall niches (Area 3), which was open to a view from the Courtyard (B) over a low wall. The entry through the fourth "niche" leads to a dead end in a small set of rooms (Area 4), but the path can continue from the gallery with a right turn at its south end (Area 5), a climb up a stairway, a walk along the eastern face of the raised room block on a ledge, then up a stairway and through the simple door into the upper room area. As a front pathway to the raised rooms, this indirect course combines a simple, visually shielded stairway with a public passage along the gallery and across the upper wall face, with another simple stairway and entry that could have also been visible from the courtyard.

guish domestic privacy from restricted entry for areas of symbolic importance. Of course, in a monumental context such privacy has its own symbolism. The very large elite Chimu compounds at Chan Chan have been analyzed in terms of accessibility (e.g., Moseley 1978). The compounds commonly have single entrances, obviously for control of access. Beyond an initial courtyard, there is increasing restriction of entry pathways when proceeding through a warren of structures and spaces. Public contact, probably tribute collection and/or adjudication, is inferred to have been carried out in specially designed enclosures in areas nearest the entrance, while the furthest and most inaccessible areas had elite living quarters, a monumental tomb-to-be, and private religious structures. The Chimu compounds seem to combine a feature of limited accessibility to the king's monumental tomb and its immense pre-burial structures as well as post-burial collection of grave goods, with limited access to provide privacy for domestic activities. Jerry Moore (1992) has applied a line-of-sight analysis to these Chimu compounds, giving a quantitative measure that he inferred as indicating the degree of control over accessibility, thus augmenting the interpretations of the compounds' function. Monumentality can also be seen as a factor at Chan Chan. The mere size and complexity of the compounds are monumental in scale, but in a way apparently more integrated with features of kinship organization, political power, control of the flow of economic goods, and a role in religious ritual that had developed as facets of the total power held by the ruling elite. Such a situation is comparable to the level of complexity reached by Inca imperial organization, rather distinct from the earlier periods of massively monumental religious structures. Hammond's (1972) study of a Classic Maya site illustrates a different use of limited access from that found for a Chimu compound. He calculated an index of accessibility by counting the rooms or spaces it was necessary to pass through,to arrive at various points,

15-

Wallace: La Centinela

This notably round-about path was not dictated by the layout; it obviously could not have been meant to serve for general public access to the raised room block and would have been a rather curious path even if meant to serve only a smaller elite. The one other pathway that can be taken after entering the corridor (Figure 4: Area 2) is to proceed down the full length toward the back of the compound. This course would pass the door to the large room or enclosed patio on the center north side of the compound (Area 12). This room is fairly accessible from the initial entry and has a double jamb entry and small wall niches, connoting some special function; whatever that function, it did not call for frequent and efficient interaction with those in the raised room block. Finally, there is the end doorway in the corridor, leading to the back or western area (Area 10), an apparent veranda open to the west over a low bal-

the very least, with a balustrade, the space in front of the door could have served as a raised dais, allowing an official to appear, speak to a moderate-sized audience, and then retire into the upper rooms. But a good alternative is that access out of the room block was intended primarily to provide a direct and private access into the set of southern comer rooms via an entry area (Figure 4: Area 9); with the veranda being reached only after a full V-turn and re-emergence from the foyer. The veranda itself could have served for daily domestic or restful activities by the upper room occupants as well as other more social or administrative functions for moderate size groups. Concerning domestic activities, I should note that south of the main compound are two smaller compounds at a slightly lower level and. separated from the main compound by a narrow alley (see Figure 3). Access between the compounds is minimal. Families of the administrators, whether lnca or local, could be expected to be of fairly large size and likely to include more than the nuclear core. Vhle assumed that the southern compounds were the main living quarters, and he had good reason to think that the size of the main compound was not large enough to serve as both fullliving quarters of an entire basic kin group and also a main administrative center. However, there is the possibility that the higher Inca administrators did not have their families in residence. In contrast, if one or more of the larger rooms or patios on the north side (Area 12) served for larger meetings or feasts, leaving the veranda area a private one for domestic activities, a regular domestic occupation by a few high officials seems possible, especially if activities such as food preparation were carried out elsewhere. The raised room block itself has a greater number and smaller size of interior spaces than most other Inca buildings of my knowledge. Such spaces might have had ritual or administrative functions, especially given the high rank of the Inca governor. These include storage of ritual clothing and other accouterments, qhipu records, the more valuable and less bulky tribute, such as mullu (pieces of

ustrade (Figure 8). The access is reasonably


direct and visually unimpeded, and the space is large enough for a moderate size group of people. Beyond that, there is no other architectural feature of 1he space itself that would indicate anything more than some semi-formal and semi-private activity such as meetings, feasts, or entertaiIiment of visiting dignitaries or local10rds, or just a private open area for those utilizing the compound rooms. The one exception is the sub-floor cistern, which in itself does not immediately suggest a specific function for the area. It is clear that the western door to the upper story is not designed for a simple, straightforward access from the corridor entrance to the veranda area. Instead, it is necessary to pass along the eastern veranda wall, enter and make two short 90 turns in an enclosed foyerlike space, ascend the stairway, walk along a short passage, then turn in through the door. As noted elsewhere, it is probable that the door and stairway were fully screened by a wall, or at least a solid balustrade. A simple stairway directly to the upper door would have been perfectly possible if simple access to the door had been the goal, so an intentional degree of separation between the veranda and the entry to the upper room block is indicated. At

ANDEAN PAST 5 (1998)

-16 and/or ritual activities, in either case of a more private nature, are suggested. (3) The veranda can be seen as a transitional link between the front courtyard-niche gallery unit and the rear (and probably main) entrance to the raised room block via the straight central hallway, the hallway giving the most direct, formal line of' access in the entire compound. This is a more direct or potentially formal access than the somewhat shorter but indirect one via the gallery. The veranda (and possibly the small rooms at its north side) might be seen as a large foyer-like room or larger meeting/entertainment area for visiting dignitaries, with a dramatic view as an added diversion. Alternatively (or alternately), the veranda could also have served as a multi-purpose area for more relaxed or semi-domestic activities by those stationed in the compound. In any case, it can be viewed as part of a major path of ingress that would put thos~ entering at a point near the main entrance to the southwest comer rooms and in front of what is undoubtedly the main door of the upper room block (the east door lacking the double jambs). That, however, is about as far as the layout draws those entering the veranda area, the southeast exit not giving any visual clue to where it led. As noted earlier, the raised area in front of the upper entrance, if fronted only by a balustrade, could act as a podium, although separating the speakers from their audience. (4) Finally, the two fairly large rooms or enclosures off the main corridor and along the north compound wall form a loose grouping that is distinct from the others. They have easy and direct access from the front entrance and can be expected to have served some direct functions, such as storage-accounting areas for tribute or meeting/feast rooms, a more public and official function suggested by the more formal double-jamb door and niche features.

sacred Spondylus shell) or gold, silver, and copper objects; or the privacy needed for small, high level planning sessions as well as sleeping areas for top officials. Multiple functions are certainly possible. Four areas may have been functional subareas of the Inca compound: (1) The tall-niche gallery may have been a functional adjunct of the eastern courtyard (B), given the connection implied by the open sight line between them over the intentionally short height of the separating wall. The higher floor level of the gallery emphasizes the mainly visual nature of access between these spaces, except for the possible passing of material, such as tribute, over the balustrade. The higher level of the gallery floor would add to the aura of higher authority given by the symbolism of the tall niches and raised rooms. However, the view of the blank outer wall of the upper room, with ~he modest stairway and small door over at one side, seems a curiously bland capping of this scene. (2) The most inaccessible unit is the raised block of rooms. The two southern rooms on the veranda level can be added as an adjunct sub-unit, given what I see as a deliberately arranged privacy and ease of access between them, but only at the expense of what could have easily been direct access to the veranda. The greatest degree of restriction for outside access, the tall niches and double jamb doors in both upper and lower areas, the unique mural in an upper room, and the walls of small niches, the south-facing "window", and the unique bathing tank room of the lower rooms all give the strongest connotations of elite function. The upper warren of small rooms is obviously not meant for interpersonal activities by more than a few individuals, so use by a few high ranking officials and/or underlings carrying out official work is indicated. The southwest comer rooms (Figure 4: Areas 8, 9, and comer room) are particularly interesting, given the bathing tank; either hygienic

17-

Wallace: La Centinela

A Stage Production with Power Niches One further question continues the discussion of the features already mentioned. The starting point is at the tall-niche wall at the eastern side of the main compound. Tall niches have been briefly introduced as having connotations of importance based on their associations with Inca buildings of high statelevel functions. Their implied importance is also strengthened by the fact that the row of four niches was meant to be viewed publicly from the entry courtyard. The wall has also been noted as part of a curiously complicated access that involves following a circuitous route past it to the eastern entrance of the raised room block. The access from the entry courtyard to the niche gallery alone suggests that it was not meant to be available to or used by any public group of even moderate size. What, then, was the purpose of this particular layout? Before following through on that question of access, however, there are some features of the adobe block construction of the wall itself that are pertinent. The outlines of some adobes in the back of one niche are shown in Figure 11, visible because the original adobe plaster has fallen off in places. The rectilinear shape is like all Inca adobes from either the highlands or the coast, and the proportions also fall into a recognizable range, with lengths about 3 to 4 times the heights. However, there is a notable anomaly visible here. Two adobes have been cut into in order to correct for some unequal heights in the coursing. This situation may seem quite minor, even familiar from viewing typical Inca stone masonry. However, it is technologically pointless when used with adobe bricks, a very unInca type of behavior. All dressed stone needs shaping, all the more so for the perfectly fitting blocks of the best Inca masonry; but producing blocks to fit an irregular space is not .automatically more time consuming than producing a simple brick-like one, because there is no short-cut to dressing stone that has no natural cleavage planes. However, producing blocks of identical height, in order to maintain evenly hori-

zontal courses so that joints can be overlapped when laying the next course, would be very inefficient in the time consumed for exact shaping, and also inefficient in use of raw material, because only properly sized blocks or blanks ready for final trimming could be used. For this reason, perfect coursing was restricted mainly to the construction of the' most important buildings (e.g., Hyslop 1990: figure 1.8). Even then block length commonly varied widely, a good indication that attempts to have all masonry block dimensions equal would have been too costly even for a state mason. In any case, careful choice of blocks with different widths could easily avoid any potential alignment of vertical joints. When laying adobe bricks, filling any potential void with mud mortar is much easier than either cutting a dried adobe or preshaping one to fit an irregular space. It is even more pointless when the finished surface is to be plastered over. PrQducing bricks of reasonably equal height is not at all difficult, and laying them using a simple mud mortar makes slight differences in brick height easy to adjust. In addition to the pointless cutting of the adobe blocks, the lack, or very minimal use of a mortar, as in the present case, is not yet noted among the many other cases of the use of adobe bricks at La Centinela. However, the other examples differ signflcantly: they are not part of buildings in the Imperial Inca style. In any case, the result of these irregular adobe shapes gives a visual impression close to that of dry-laid Inca stone masonry. These two cases of cut adobes may be unique, but even so the irregular shaping was obviously the result of a conscious act by whoever had to plan a correction for the uneven coursing, and it was done in a technologically inefficient way. I suggest that it was the reaction of someone accustomed to stonework of the quality used in Inca state architecture and familiar with its attendant problems, and not by anyone trained locally in Chincha, where stone was not easily available and dressed stone construction was unknown. It seems reasonable that this architect and/or artisan may have reacted in this way because he was working with rectangular blocks associated

ANDEAN PAST 5 (1998)

-18 the transaction? If they were to pass into the compound itself, even if no further than the first space, in order to transact some business, what was the purpose of the large viewing space? If everyone visiting the place was meant to be impressed by the view, why was it possible to enter the courtyard and go directly into the building proper without even getting a

with the top quality Inca stonework used in a building destined for important state-related activities. In short, the reaction was as if stone were being worked rather than adobe, despite the common Inca use of rectangular adobes set in thick mortar, a technologically more reasonable approach that also made it easy to cover up irregularities in brick size. It is even likely that this mason was abrading the dried adobe blocks in order to attain their very close fit, just as he would have worked stone. In any case, the situation suggests one more reason to consider the compound as an important place for conducting matters of state. The full size double jamb niches (Figure 4: Area C) are one of the main reasons why Compound C is considered the main center of Inca authority in Chincha. The only other cases of this feature anywhere at the site, to my knowledge (as of 1957), are the one niche in a back room and one in a raised block room, both in this Compound. In Inca state architecture, this. niche type, especially in a long series, is usually found where its appearance emphasizes the importance and monumentality of a structure or area, or where it visually enhances walls lining an interior open space or, especially for retaining walls, a large open courtyard or plaza. The gallery itself is too narrow for such a view from within, establishing the earlier interpretation that the wall was meant to be viewed by an audience in the courtyard, better fitting the Inca pattern. From a strictly architectural perspective, the association of a series of tall niches with walls functioning to retain fill might have given the impression that it was a common extension of the solid platform under the raised rooms, emphasizing the feeling of monumental solidity. Having set up the supposed intentional manipulation of the layout and the structures so that an audience in the courtyard would be faced with a barrier wall, with a governmentstyle tall-niche wall behind, but a blank wall and simple door without the double jamb seal of government approval above (see Figure 7), there is definitely something amiss. What was worth waiting for and looking at? If people were there to pay tribute, was passing it over the wall really a sensible way of handling

view of the niches? The only certaintyis that '


this setup was meant as a carefully managed attempt to impress the viewer with the visual symbol of the focus of great power that resided in the building and in its occupants. There is one possibility that would make sense both of the symbolism and the architectural form of this layout as just discussed. The possible answer comes from one obvious step: putting a roof over the small row of rooms behind the niche wall, using the top of the walls and the ledge across the front side of the room block for support of roof beams or the floor boards themselves. The roof surface would then serve as a platform or stage, as shown in the drawing in Figure f2a. The stage could have various sorts of screens, space dividers, or walls, and could well have been roofed, as shown in Figure 12b. There is definitely a precedent for such a function: the Inca ushnu (Zuidema 1989), a platform found in almost all state-constructed Inca centers, placed in some central location in the main plaza and functioning as a symbol of Inca state power (Hyslop 1990:70). Some ushnus, such as the one in the principal plaza at Cusco, were much more than daises or stages and involved the complicated symbolism discussed by Zuidema. Nevertheless, the simpler raised platform type is widely found at Inca centers established throughout the empire and had a minimal function at least of serving officials for carrying out various formal reviews or rituals, as detailed extensively by Hyslop (ibid. :66-101). The point here is that staged public rituals were a common part of official Inca activities, not that the stage proposed here is necessarily an ushnu, a term which apparently covered a range of forms. A simple ushnu has been identified in the main plaza at Tambo Colorado, an Inca center in the Pisco Valley just south of Chincha (ibid. :85, figure 3.19), as an example near La Centinela.

19-

Wallace: La Centinela

The idea of a stage also makes sense of the very indirect access to the raised room block. It may have had two main functions: the high official(s) residing in the upper structure could easily exit by the eastern door and descend the few stairs to the platform and use it to address the public or receive requests from the audience, as well as oversee any activities carried out in the courtyard, now open to his elevated view. Further contact with the audience would have been well controlled. The access from the lower niche gallery, via the end passageway (Figure 4: Area 5) and stairs, could have been meant mainly for access to the stage from below, by any attendants of the high officials or any guards stationed in the gallery itself. The rooms directly behind' the tallniche wall (Figure 4: Area 4; Figure 7), entered through the one false niche, could have housed a permanent contingent of guards and/or been used to store any tribute passed over the wall by those in the courtyard. The full access from the niche gallery into the raised rooms could have be~n only a convenience at times, not its main purpose, and therefore not designed as an efficient, direct, commonly used methE)d of entering the raised room block from the lower level or from outside the compound. All this argument can be translated into formal symbolic terms. Both the distance and higher elevation maintained between officials and their subjects would translate into figurative "low" and "high" socio-political distance, while the wall that blocked access to the stage itself would shield officials from any direct contact, emphasizing the socio-political gulf between the audience and the officials, and the inaccessibility of the source of power and impossibility of challenging it. The simple emergence of high officials from their second floor aerie, possibly from behind hangings concealing the actual door and stairs, would augment their impressiveness. In addition to the niche wall adding an aura of authority to anyone using it, the stage could have been roofed (Figure 12b) and some true staging, in the modern sense of adding drama and esthetic impact, could have been employed. This may have included pennants flying from the roof, many other props on the stage level, and walls

and/or mat screens covered with decorative wall hangings adding opulence to the signs and symbols of authority. Lower order officials and guards stationed in the niche corridor would have added to the total effect, while separating them, in part symbolically, from the higher officials literally above them and the audience below, in front of the low wall between them. To summarize, the analysis of access suggests that the entire main compound, minus the courtyard, was not meant for public usage in the usual sense; instead, it was at most open only for visiting high dignitaries or state officials with moderate size retinues. Nevertheless, the fact that officials could easily have a public audience of moderately large size without leaving the compound proper gives an interesting twist to the meaning of public. They simply had to descend to the stage overlooking the courtyard, where they were only visually accessible to their audience, a situation which actually emph~izes the aura of restricted access, leaving an interesting type of both managed public access and privacy. While this situation would fit a desire for domestic privacy, only the bathing tank specifically suggests domesticity, so whether the compound proper was laid out for either administrative or domestic privacy or for some combination of these, is still open. There is one further possibility for the function of the compound and courtyard, one which fits the most salient function of the site at a whole: as the home of a very powerful oracle. If it were assumed that the Inca insinuated themselves fully into the activities of this very important pilgrimage site, they may have seen the value of having the actual location of the oracular pronouncements be framed, literally, by a setting of Inca style architecture, and probably would have had control, or at least veto power, over the content of the oracle's output. The stage setting would be perfect for such pronouncements, assuming they were not of too personal a nature. In addition, drama could be added to the act by delivery by a disembodied voice: there is the space under the stage, which, if the entrance were disguised as just another tall niche so

ANDEAN PAST 5 (1998)

-20
Heyerdahl,Thor, Daniel H. Sandweiss,and Alfredo Narvaez 1995 The Pyramids of Tucume London & New York: Thames & Hudson. Hyslop, John 1984 The Inka Road System. Orlando: Academic Press. 1990 lnka Settlement Planning. Austin: University of Texas Press. Menzel, Dorothy, and John H. Rowe 1966 The Role of Chincha in Late Pre-Spanish Peru. Nawpa Pacha 4:63-73. Moore, Jerry 1992 Pattern and Meaning in Prehistoric Peruvian Architecture: The Architecture of Social Control in . the Chimu State. Latin American Antiquity 3(2):95113. Morris, Craig 1988 Mas alIa de las fronteras de Chincha. In La Frontera del Estado Inka, edited by Tom D. Dillehay and Patricia Netherly, pp. 131-140. Proceedings of the 45th International Congress of Americanists, Bogota, Columbia. 1985. BAR International Series 442. Moseley,Michael E. 197E Central Andean Civilization. In Ancient South Americans, edited by Jesse D. Jennings, pp. 179-238. San Francisco: Freeman. O'Neale, Lila M. and A. L. Kroeber 1930 Textile Periods in Ancient Peru. University of California Publications in American Archaeology and Ethnology 28(2):23-56. Berkeley: Universityof California. Patterson, Thomas C. 1985 An Andean Oracle under Inka Rule. In Recent Studies in Andean Prehistory and Protohistory. edited by D. Peter Kvietok and Daniel H. Sandweiss, pp. 159-76. Ithaca: Cornell University Latin American StudiesProgram. Pillsbury, Joanne 1993 Sculpted Friezes of the Empire of Chimor. Ph.D. dissertation, Columbia University. Ann Arbor, Michigan: UniversityMicrofilmsInternational. 1996 The Thorny Oyster and the Origins of Empire: Implications of Recently Uncovered Spondylus Imagery from Chan Chan, Peru. Latin American Antiquity 7:313-340. Rostworowskide Diez Canseco, Maria 1970 Mercaderes del Valle de Chincha en la epoca prehispanica:un documentoy unos comentarios. Revista Espanola de Antropologia Americana 5:13577. Madrid: Facultad de Geografia e Historia, Universidad Complutense. (Reprinted in Rostworowski . 1977, pp. 97-140.) 1977 Etnia y sociedad: costa peruana prehispimica. Lima: Institutode Estudios Peruanos. Sandweiss,Daniel H. 1992 The Archaeology of Chincha Fishermen: Specialization and Status in Inka Peru. Bulletin of Carnegie Museumof Natural History 29. 1995 Life in Ancient Tucume: Sector V. In Heyerdahl et al. 1995,pp. 142-168. Santillana,Julian Idilio 1984 Un asentamiento Inka-Chincha. Rasgos arquitect6nicos estatales y locales. Arqueologia y So-

that the impression were of a solid terrace, would make quite effective staging. There are undoubtedly further interpretations and speculations that could (and probably will) be made, but I have no more at the present. In closing, I will simply note that, even in considering just this one Incaic compound, and despite treating only certain types of features, the underlying theme concerns the use of architectural form and decoration for purposes of facilitating and augmenting the maintenance of the social and political order. The Inca are prime subjects for this kind of inquiry, because their ethnohistory and archaeology show an inordinate concern with order and control. John Hyslop's work on Inca settlement planning (1990) has culled and organized a mass of data and analyses. Like most good work, it reminds us. how much there still is to be done and should inspire us to try all possible types of analysis, particularly ones that search for repeated patterning among the ever increasing detailed data .on Inca architecture and planning. Acknowledgments The subject matter of this paper was presented at the annual meeting of the Northeastern Conference on Andean Archaeology and Ethnohistory held in Albany, New York, in 1985, although it has been considerably reworked from its original form. The field work was part of the 1957-58 site survey of the Chincha Valley (Wallace 1971) supported by the Fulbright Commission in Lima, under Eduardo Indacochea. I was ably assisted on the survey by Luis Guillermo Lumbreras and Isabel Flores. I am indebted to these individuals, and also to Dorothy Menzel for her comments on an earlier draft.
References Cited
Bonavia, Duccio 1985 Mural Painting in Ancient Peru, translated by Patricia J. Lyon. Bloomington: Indiana University Press. Hammond, Norman 1972 The Planning of a Maya Ceremonial Center. Scientific American 226(5):82-91.

21ciedad 10:13-32 & 7 unpaginated figures. Lima: Museo de Arqueologiay Etnologiade San Marcos. Uhle, Max 1924 Explorationsin Chincha. University of California Publications in American Archaeology and Ethnology 12(2):55-94. Berkeley: University of California. Wallace,DwightT. 1971 Sitios Arqueol6gicasdel Peru: Valles de Chincha y de Pisco. Arqueo16gicas13. Lima: Museo Nacional de Antropologfay Arqueologfa. 1978a Economic Systems in Late Intermediate Period Chincha. Paper presented at the Institute of Andean Studies AnnualMeeting,Berkeley. 1978b Late IntermediatePeriod Chincha: Archaeological and EthnohistoricalProblems. Paper presented in the symposium "Economic Complementarity in the Andes", International Congress of Americanists, Vancouver,B.C. 1991 The Chincha Roads: Economics and Symbolism. In Ancient Road Networks and Settlement Hierarchies in the New World, edited by Charles D. Trombold, pp. 253-263. Cambridge: Cambridge UniversityPress. Zuidema,R. Tom 1989 EI Ushnu. In Reyes y Guerreros: Ensayos de Cultura Andina, by R. Tom Zuidema, pp. 402-454. Lima: FOMCIENCIAS.

Wallace: La Centinela

ANDEAN PAST 5 (1998)

-22

.' .,. .' <' .'


,'"

I " "

,~

-'

.. "

..,.l'

,'

',.

'.

'.
'.

j;.
~

~ 0

....

"'"
~

\,.../.,,.-"""-_
J ~. .

~
"" .

~././.

~./ p;

'. , \.. ".

__ _ '~_~ ,_' -

'-'-'_1''Ch " ;..


'

\.

,.'
..,

,f"

..., ,..,

".

' . ).,J' I
I

'

'.--" .'

'.-.

,.,

",.:it

, ,\ ,'" ~..> i ,'.: ..~'y"_ ". "...c.No.4I~ .W,N..."'..~

.., ,." I

.,'" ,..I...'...

'\,

'-.

.2f ~ ~A.
'.

~ r
~h

ff Z 'iff:;.-#~ ,.-/ dr %
""

.2$

.,'

..."..-.. "
'

'-sAUTRER(A

~ ~ .,~~"~'

" ,

., .
.; " ,,

: .
, I , I

..
I I

CEN11NE SANPE

Late Prehlspanic Snes PrehispSMic Roads Modem Roads Edge Of Cultivatjon

. -'_1_"_"_"_.--.........
111111111111

Sluffs' . Base Of Mountains

III 111111111

Figure 1. The Chincha Valley, with La Centinela as the center of a radiating set of straight, ancient roads. The ritual platfonns at Centinela de San Pedro and Salitreria are on the outer anns. The three modem urban centers are also shown. Lateprehispanic sites existing in 1957 are shown by relative size. Map is based on aerial photos by the Servicio Aerofotogratico Nacional del Peru, Proyecto 6351, 1953. Site survey was funded by the Comision Fulbright in 1957-59; map was adapted from Wallace (1971).

tV W I

~
:::::::

~ '.

Figure 2. Air view from the southwest of the Centinela site in the Chincha Valley. The 12+ m high main terrace runs from the high wall at lower left across in front of the 18+ m high central pyramid. The rpad system focuses on the entry plaza at the south front of the main terrace, lower center right, between two lower terrace arms. The main Inca administrative-living compound is at the southwest corner of the high two-stage terrace wall, at lower center left of the photo. Photo courtesy of the Department of Library Services, AMNH, Negative 334733 (photo by Shippee-Johnson).

t.... ~ Q ~ :;. (\) is"

~
~ ~
Main PJ'u mid

/
~
I
I". I I I

~ ~
> 00
~

PI.AIA

0\

R..,

t ( [
~
..

;
..

<II -\0 \0 00 '-"

.0

I
CO"POUMO t;

?
?

---COURTY ARD a

[:l
?

L r
'_.:

! [11
I

.1
I

,.. I

:
L

1m
?

.A

r
[ [

'?

.--.

~ I
AT L,A.CENTINELA, CHINCHA. o 5 1[" '1 __ f8.!1mrRIII ST_
DII'PIIAIHBS II NX1R 1lJCll1UIOUI1O UI'I1I.< IIOA8 _ DISCEAtIPLN<

y INCA BUIt.DINGS ?

PERU

~.ftII!'o':CBS

ClllIIIToIINIHCI WALLS

WAU.$

I
I

OEIISITY OP 1IA'l'CllRl1I01CA11IS ~

IS HIOIIEII. ARROWS. AISB

r
8

IIAIIK$ SIIOW .ARAS WITH 100

Figure 3. Plan of the Inca construction on the 12 m high terrace in front of the pyramid, as of 1957. The main Inca compound (C) is center left. Hatching roughly indicates differences in level, denser being higher.

~ ~

tV V\ I

COMPOUND

Figure 4. The main Inca compound, enlarged from Figure 3. The raised block of rooms is in darkest outline. The tall~nichegallery (Area 3) can be viewed from Courtyard B over a low wall. The veranda (Area 10) has an ocean view. Rooms in the lower left comer include a sunken bath with drain. .

~ is'' ('" ~ '. t"-t ~ Q ~ .... s. ~ is''

~ ~
> 2:
Iood

-3 !.It -. ..... \0 \0 00 '-"

Figure 5. The Inca compound seen from the pyramid, looking southwest, with the ocean as background. Huaca de Tambo de Mora is seen in the distance, while the area between is probably filled with structures. Photo taken by Uhle in 1900, here courtesy of the Phoebe A. Hearst Museum of Anthropology, University of California-Berkeley.

I tV 0'1

27-

Wallace: La Centinela

Figure 6. The Inca compound seen from the pyramid, looking southwest, with the ocean as background. Only Uhle's cleaning of the niche gallery, seen in Figure 5, differs. Huaca de Tambo de Mora is in the distance, while the area between was probably filled with structures. Photo taken by Wallace in 1957. .

ANDEAN PAST 5 (1998)

-28

Figure 7. The Inca compound reconstructed, viewed as in Figures 5 and 6. The gallery with tall double jamb niches (Figure 4: Area 3) is at the east end of the raised room block. The corridor along the shadowed side of the raised rooms leads to the veranda off to the right.

29

Wallace: La Centinela

Figure 8. The Inca compound's west end reconstructed, with the veranda front and center. The raised room entrance is via the stairway at the right and along the, raised walk. The dotted lines show a probable balustrade or wall, making the access from the door to the lower southwest rooms more private. Gradual widening of this wall would make the west end of the raised block appear parallel to the outer veranda balustrade. A sunken bath with steps and drain is at the right. The niches in this comer room cluster include a tall one, marking an elite function.

ANDEAN PAST 5 (1998)

-30

Figure 9. A late local style pIano-relief adobe frieze in a small room or balcony high on the pyramid. Its location allowed for a restricted, elite viewing. Representing a common technique on the north coast, but not elsewhere on the south coast, it was possibly introduced by traveling Chincha merchants.

31 -

Wallace: La Centinela

Figure 10. Part of a three wall mural in the raised room block. The bottom and left end lines are the lower and end borders of the full panel; the upper border is missing. Room walls were white, the mural black and red (hatched), plus a now-eroded green (cross-hatched) that Uhle (1924) described.

ANDEAN PAST 5 (1998)

-32

------

Figure 11. Outlines of adobes in one tall niche in the eastern gallery. The irregular areas are white adobe plaster. remnants. The cuts in the rectangular adobes and apparent lack of mortar imitate highland Inca stone masonry used for important state buildings. They are foreign to the normal laying of adobe blocks.

33

Wallace: La Centinela

FigUre 12. The tall niche gallery shown with roofed interior spaces to form a stage viewable from the eastern courtyard. Officials could easily (and more effectively) descend from the private upper rooms to carry out ritual/judicial/administrative functions. The left stairway would provide access for aides or guards below. Figure 12b adds a possible roof over the stage.

RECONSTRUCTING THE GREAT HALL AT INKALLACTA

Vincent R. Lee, Architect


Institute of Andean Studies

Introduction Preface This paper is an attempt to reconstruct the Great Hall at Inkallacta, the ruined Inca provincial capital (Lee 1991:7-10) near the valley and village of Pocona in the Department of Cochabamba, Bolivia (Figure 1). Some of the masonry elements of the structure remain in remarkably good condition, while others are in various stages of ruin. Original construction components made of long-decomposed organic materials are absent. altogether and must be inferred nom the masonry or nom what we know of both the mechanics and the. spirit of Inca architecture and construction in general.' In addition to analyzing the physical evidence at the site, therefore,.I have revisited the design problems faced by the original builders in light of what Marcia and Robert Ascher have called Inca "insistence" in their study of the quipu (1981: Chapter 3). Simply stated, insistence is the sum of those traits that repeatedly manifest themselves in the activities and artifacts of an individual or a culture. Properly understood, architecture is a form of language that tells as much about the speaker as it does of what is being spoken. As such, it is an especially rich repository of cultural insistence. Among the Incas, building was an important, perhaps primary, means of cultural expression. Whatever ends Inca builders sought to achieve, the means employed seem generally characterized by an insistent concern with organization, order and stability. In attempting to reconstruct their work, I have approached the task with these concepts, especially, in mind.

Inca Great Halls The Great Halls, or kallankas, of the Incas were among the largest buildings of preColumbian America and certainly enclosed the largest interior spaces. They were important features of most Inca sites of consequence and one or more were often erected facing onto the main plaza. In addition to sheltering large public gatherings in foul weather (Garcilaso 1987 [1609]:320), they may have been used as hostels by travelers and passing military units. The Spaniards called them ga/pones, or dormitories, due to their similarity to the large structures in which slaves and farm workers were then housed in Sp~. According to Garcilaso de la Vega (1987 [1609]:320-321), the largest Inca halls were those of Cusco. He claimed that they were as much as 200 paces long, 50 or 60 wide and that one, the Casana, could accommodate a crowd of 3000 people. Nothing of these halls remains today and many scholars discount Garcilaso's report as exaggerated. The Great Hall at Inkallacta is the largest surviving example and covers an area 78 m long and 26 m wide. It is twice the size of its nearest rival, and the Statue of Liberty laid on its side would fit nicely inside. The three-to-one proportions of the plan at Inkal1acta are also unusual. Most Inca halls were long and narrow, more on the order of six- or seven-to-one. The halls at Espiritu Pampa, Hulinuco Pampa and Rosas Pata are typical examples (Figure 2). We know little about the roof structures which once covered Inca buildings in general or Great Halls in particular. Virtually all observers reported that they were of thick, neatly trimmed thatch supported on nameworks of lashed poles (Markham 1862:193-4; Squier 1973 [1877]:395; Pizarro 1978 [1571]:161-2; Garcilaso 1987

ANDEAN PAST 5 (1998):35-71.

ANDEAN'PAST 5 (1998) [1609]:321). The drawings of Guaman Poma de Ayala (1936 [1615]) show various building types, but structural details are almost entirely lacking. His view of the royal palace of Cuyusmango in Cuzco (Figure 3) suggests only that the roof of that hall was rounded over the ridge, rather than rising to a sharp peak, an important point to which we will return later. Huanuco Pampa Craig Morris and Donald Thompson, in, their excavations at HUBnUCO Pampa, uncovered a significant feature in one of the halls there, which they called kallankas. They found a row of seven stone-lined post holes running down the centerline of the largest and northern-most hall (1985:89). No record exists of the exact location of these holes, but Morris recalls that they did not appear to be related to the spacing of the doorways (personal communication). In a photograph of the same hall by Gasparini and Margolies (1980:204), the holes appear to be roughly equidistant apart. A reconstruction drawing (ibid:202) indicates h9W the authors thought posts set in the holes might have supported the roof. Other photographs of the building (ibid:200-204) also show the hall's partly destroyed gable ends, from the steep angles of which the authors surmised an apex height of about 8 m (~6 ft.) above the eaves, or 12 m (40 ft.) overall. Finally, both gables were pierced by two man-sized openings above eave height (Figure 2, Cross Section XX). Careful study of this building discloses several subtleties not noted by previous researchers that are also present at Inkallacta, and are therefore almost certainly intentional design features with some significant purpose. First, the two eave walls are of different heights, the rear side being nearly a meter higher than the front wall, facing onto the plaza. The building is on level ground and no external factors suggest a reason for the difference. Second, although they are equidistant from the centerline of the building, the two gable openings are set at different elevations vertically. The ones nearest the rear wall are about 30 cm (12 in) higher than

-36 those toward the front. Finally, the distance between the post holes is roughly the same as the height of the posts, the width of the building and the length of each slope of the roof. All measure about 10 m (33 ft). Possibly this is an indication that trees available to frame the roof were limited to approximately this dimension, because there is. no source of large timber in the vicinity of
HUBnUCO Pampa.

Inkallacta

The Site and Previous Studies Inkallacta was built by Topa Inca during his conquest of the region, probably in the 1470s, and underwent' repairs by Huayna Capac about 50 years later (Cobo 1983 [1653]:] 54). It is a sprawling site, arranged above and below a large plaza spreading across a gently sloping bench at the confluence of two rivers flowing in deep quebradas. The perimeter of the bench is protected by a defensive parapet and 50 or more buildings of typical Inca design are scattered both within and beyond the fortified area. Of those structures inside the wall, one, called the Great Hall, stands out (Figure 4; Figure 5, Building 1). It is not mentioned specifically in any of the chronicles of the period. However, its enormous size and position of prominence facing onto the main plaza suggest both that it was an especially important building, designed to impress onlookers, as well as accommodate large crowds and serve its functional purposes, whatever they may have been. Inkallacta was presumably abandoned at the time of the conquest and lay more or less undisturbed until 1913, when it was visited by Erland Nordenskiold, a Swedish anthropologist then exploring the Inca frontier in southeastern Bolivia. His description of the ruins (1924) was the first by a European scientist, and was followed in 1927 by the account of a subsequent visit by Don JesUs Lara, a Bolivian from Cochabamba. An updated version of Lara's report was re-published in 1967 and again in 1988. In it, he reviewed various other assessments of the site published

37in the four decades since his 1927 expedition. Among other things, Lara concluded that the efforts of some of his colleagues over the years to identify Inkallacta with the "lost" fortress of Cuzcotuyo (Sarmiento 1907 [1572]:165), and thereby endow the site with a bit of romantic history, were "in error" (Lara 1988: 61). The probable site ofCuzcotuyo is about 250 Ian southeast of Inkallacta, at a ruined hillfort now called Inkapirca (Lee 1992). A detailed inventory of Inkallacta was done in 1976 by Gonzales and Cravotto in connection with its designation as a World Heritage Site by UNESCO (Gonzales and Cravotto 1977). In 1980, Gasparini and Margolies included a description of the Great Hall and a brief analysis of its roof structure in their compre-hensive study of Inca architecture (1980:207-212). "Aside from Santiago Agurto Calvo's generic speculations on large Inca roofs (1987:236-251), Gasparini and Margolies' was the first and, until ~ow, only such study ever done. David Pereira, Director of the Archaeological Museum in Cochabamba, cleared the entire site, including the interior of the Great Hall, shortly before my own visit in the fall of 1990. It was fortunate timing, because we were able to see critical features in the building's fallen west gable end that had previously been obscured by brush. Layout of the Great Hall Before beginning an analysis of the roof and other missing elements of the Great Hall, a close look at its floor plan (Figure 6) and walls (both standing and fallen) is instructive. For descriptive purposes, the building will be assumed to be oriented with its long axis eastwest and its doorways opening to the south, although they actually face about 20 degrees west of south. The site slopes gently to the south as well, such that the rear, or north, wall retains about 4 m (13 ft) of grade. The floor inside the building slopes a third of a meter toward the doorways, through which one steps down another half meter out onto the plaza (Figure 7).

Lee: lnkallacta The long dimension of the building was limited by the terrain. A narrow passage separates its east end from the parapet overlooking the eastern quebrada and the west end is cut into the base of a steep, rocky hillside. A large drainage gutter behind the north wall collected runoff from that slope of the roof and directed it down around the wes~ end of the building and out onto the plaza, well away from the eastern bluff, which heavy runoff might otherwise have undermined. Twelve doorways with windows midway between them are arranged symmetrically on the south facade. No doorway stands on the centerline of the building, but a small platform outside the door just east of center may have signified the principal entrance, or may simply be related to two large boulders that occur there. Lower Walls Below the gables, the masonry throughout is well fitted and chinked pirca, or fieldstones set in clay mortar. With the possible exception of some roughly-shaped comer quoins and lintels, there is no cut stonework anywhere in the building. Nevertheless, it remains in remarkably good condition. The relative isolation of the site has no doubt saved it from the casual vandalism and pilfering of building materials common in more populated areas. Vestiges of a thick coat of hard, salmon-colored plaster remain in several niches (Figure 8) and inside the northwest comer (Figure 9). The many pebbles inserted into the wider joints to provide purchase for the plaster have incidentally served to retard weathering of the exposed mortar. Except for the west gable end, which has collapsed completely, most structural damage to the lower walls has resulted from fallen lintels. They are missing from a number of the 44 equally spaced niches in the north wall, which is otherwise intact, and from all of the doorways and windows in the free-standing south wall (Figure 10), which has suffered badly as a result (Figure 11). The lack of long stones in the debris around the south doorways or elsewhere in the building suggests that

ANDEAN PAST 5 (1998) stones to cap the niches and windows were scarce and blocks long enough to span the doorways may have been unavailable. Wood poles, since decomposed, were probably used instead. Another factor that may have contributed to the deterioration of the south wall is thrust from the south eave of the roof. Unlike the vertical surfaces elsewhere in the building, the south facade is battered noticeably back as if to resist lateral forces. Whether the battering commonly seen in Inca architecture was intended for this purpose is unknown, however examples exist that strongly suggest it sometimes was (Lee 1988b:10-11). In this case, thrust would have accelerated cracking and disintegration once the thatch began to rot and allow water into the top of the wall. On the north side, similar damage was resisted by the greater strength inherent in. the design. Grade retained behind the wall offset the effects of thrust from the north eave and there were no doors or windows below to weaken the masonry. Except on the south, all the walls retain (or retained, at the fallen West end) some amount of grade. Behind the north wall, it varies from about 4 m (13 ft) at the northeast comer to 3 m (10ft) at the northwest end. Down the outside of the gable ends, these depths decrease to a bit more than 2 m (7 ft) at the southeast comer and zero at the southwest. All these walls would have been strengthened by being thicker at their bases than at their tops, yet none shows evidence of battering. Instead, the heights above floor level of the many interior niches vary approximately in proportion to the depth of retained grade behind the walls. The result is that the northern niches seem inconveniently high -- at or above eye-level -- and those near the south corners are less than half a meter above the floor (Figures 7, 9). Also, the pattern of the stonework sometimes shows a faint line in the joints connecting the base of one niche to that of those adjacent to it (Figure 12). Susan Niles has shown that the Incas sometimes built niched walls that maintain full thickness up to the bases of the openings,

-38
and then are stepped out to the plane of the niche-backs (1987). This allowed the builders to finish the niches at their leisure, meanwhile getting on with the rest of the work. This technique also provided a strong base for the wall from the outset. This was exactly the detail needed for the retaining walls of the Great Hall, and both the variable heights of its niches and the pattern of its stonework suggest. that Niles' technique was used there (Figures 13, 14). In fact, the advantage of this system was especially important at Inkallacta, because the backfilling of earth around the outside of the building could have proceeded sooner in the process and simplified access to the walls by plasterers, roofers and other' workers (Figure 15). A curious detail occasionally seen in well preserved, but smaller Inca buildings elsewhere is also present in the Great Hall. Empty pockets still evident in the stonework indicate that meter-long, 20 cm (8 in) diameter wooden poles once spanped diagonally across each interior comer, about 4 m (13 ft) above the floor on the north (Figure 9) and 3 m (10 ft) in the south. The purpose of these poles has never been proven, but it has been suggested that they strengthened the masonry, provided points of anchorage for the roof or were for hanging interior appointments of some sort. In small buildings, any of these ideas might be feasible. At Inkallacta, however, the size of the walls, roof structure and interior space seems to render the poles insignificant for any of those purposes. Instead, the fact that their height above floor level inside roughly mirrors finished grade outside suggests they had some functioI) related to the construction of the building rather than to the finished product. We know from buildings abandoned during construction elsewhere (there is a clear example among the chullpas at the site of Sillustani on Lake Umayo) that Andean stoneworkers, like masons today, built their corners first and then filled in between. Nowadays, batter boards are first erected outside the building perimeter from which lines are strung to establish perfectly square corners with straight walls between them. The

39walls of the Great Hall are straight, but as with Inca buildings elsewhere, the comers are only roughly square. A layout method that would account for this might also explain the comer poles. If the comers were laid up first and fitted with the poles near their tops, then string-lines connecting the poles would establish perfectly straight interior wall faces between only approximately square comers. Such lines might also have been used to assist in laying out niches and other openings by the use of sliding, quipu-like plumb lines to establish their locations for the finish masons (Figure 14).
Upper Gable Ends

Lee: /nkallacta building, but they are progressively lower towards the front or south side. In the Great Hall, their slope approximately matches that of a line connecting the comer poles discussed earlier, as well as the grade outside the building and the bases of the niches inside (Figure 7). Possibly Gasparini and Margolies assumed the visual "sag" of this wall to the' south was the result of uneven settling beneath the structure's foundations, but the lack of any cracking or other damage to the masonry (Figure 19) suggests rather that the slope of all these features toward the front of the building was intentional. Of the four gable comers, the southeast stands tallest and shows about 3 m (10 ft) of intact wall above eave height. In typical Inca fashion, four stone pegs protrude from the exterior face of the wall about a meter apart and almost a meter below the top (Figure 17). For the first two m (6-112 ft), the top of the wall rises at a uniformly steep angle matching the smaller remnant still visible at the southwest comer (Figure 16), as would be expected. Above 2 m, however, the southeast comer lays back to an angle slightly less steep. The fourth and highest of the stone pegs reflects this subtle change, as may be seen from its shadow in Figure 17. None of the other corners being high' enough to confirm whether the subtle change was an intentional feature or just the result of uneven building, this detail was not previously thought significant. After walking over the west end of the building for several days, we suddenly realized that the west gable end had fallen almost entirely intact (Figures 6, 20). The stones extended the same distance out from the original base of the wall as the stonework did above the base of the standing east end, so it was clear that the west gable had not only collapsed in one piece, it had not been distorted by the fall. Even the four openings may be seen in roughly their correct positions (Figure 6). Beyond the fallen stones, a distinct mound of dirt still shows where the adobe upper part of the wall fell. Unlike the east end, where the adobe has melted into a formless pile, the fallen west end retains an

The cross section of a building's roof, and thus the configuration of its roof structure, are usually reflected in the shape of its gable ends. For this reason, most speculation to date has focused on the gable walls of the Great Hall. Although most of the west end has collapsed, the east end still stands more than 8 m (25 ft) high (Figure 19) and portions remain of all four gable comers (Figures 16, 17). By projecting the steep, finished comers upward to their intersections in space, the original height of the apex would have been about 20 m, or 66 ft (Figure 7). By comparison, the still intact ridge of the slightly larger Temple of Wiraqocha at Raqchi rises to only about 12 m (40 ft). Because it is clear from the standing east gable that the stonework gave way to adobe above the 7.5 m (25 ft) level, researchers have wondered whether such a high, largely adobe wall was structurally feasible. Gasparini and Margolies solved the problem by disregarding the standing comers and projecting instead a low-pitched gable of modest height. In addition, they abstracted the layout of the niches and upper gable openings, projecting a symmetry not found in the building itself (compare Figures 7 and 18). The actual layout of the east gable wall recalls the design of the kallankas at Hmmuco Pampa (Figure 2, Cross Section XX). At Inkallacta, four man-sized openings pierce the stonework just above eave level. As at Huaouco Pampa, they are horizontally symmetrical about the centerline of the

ANDEAN PAST 5 (1998) even, rounded shape out to an apex about 7.5 m (25 ft) beyond the stonework. This suggests that the original gables were about 15 m (50 ft) high and were made of adobe in their upper halves. Viewed together with the subtle reduction in the steepness of the southeast gable comer, the rounded profile of the fallen west gable confirms that both became progressively less steep toward the apex of the roof. For the first time, we have a good idea of the design and height of the Great Hall's original gable ends. Assuming that they reflect the shape of the building's roof structure, we are now better able than ever before to analyze how it was framed. Roof Framing It is theoretically possible that the roof was framed by some sort of trusswork spanning

-40 the centerline, beneath the ridge, and the others at the third points to each side. The aisles between the rows would be about 4.5 m (15 ft) wide, except at the front and back, where the thickness of the eave walls would reduce aisle width by more than half. The gable openings would appear approximately centered in the four middle bays, as would the pairs of niches in the end walls below (Figure' 21). Horizontal poles, or purlins, attached to the tops of each line of posts could have been set to reflect the profile of the gable ends by adjusting the post heights accordingly. As if to confirm this idea, the southern-most row of posts would line up with the slight reduction in the slope of the southeast gable comer noted earlier (Figure 22). If a similarly small reduction occurred in line with the second row of posts, the resulting apex at the centerline would have been about 15 m (50 ft) above the floor (Figure 21), exactly where the fallen west gable suggests that it was. Indeed, using straight framing members like poles, there would be no other non-trussed way to frame a roof of progressively decreasing pitch. Similar to the "Dutch" or gambrel barn roofs common in rural America, the system proposed here not only resembles the profile of the Cuyusmango recorded by Guaman Poma (Figure 3), but would have been a good design on several counts. The steep pitch at the eaves reduces thrust onto the tops of the eave walls and creates headroom for the outermost gable openings, while the lower pitch in the center allows for lower gable walls and eliminates unnecessary interior volume, framing materials and thatch. Even so, 12 and 15 m (40-50 ft) high posts would have been required in the inner rows, although they might have been spliced from shorter members the way tall, wooden ship's masts were a century ago. In any case, lateral bracing would have been needed to prevent failure by buckling, especially at the splices, if any. Again, the masonry suggests how this might have been accomplished. The four upper gable openings have been assumed to be windows by most authorities to

from eave wall to eave wall, without interior


supports, as suggested by Agurto Calvo (1987:241), but there are good reasons to think otherwise. We have seen that less than half the width of the Great Hall required a line of posts at Huanuco Pampa and there is no mention of a knowledge of truss design in any of the chronicles. Finally, a clear-span system is most easily supported by eave walls of equal height and would impart no outward thrust at its bearing points, yet the eave walls at Inkallacta differ nearly two m (6 ft) in height and the south wall is battered inward, seemingly against thrust. Assuming, then, that posts were used, their layout remains open to question until excavations uncover firm evidence. Gasparini and Margolies proposed a system with three rows of columns completely unrelated to even their stylized version of the gable ends (Figure 18). From what we now suspect is the true shape and design of the gable ends, it is possible to make a better informed guess. At Hmmuco Pampa, the posts were placed along the centerline of the,building with aisles about 6 m (20 ft) wide to either side (Gasparini and Margolies 1980:204-205). The openings in the upper gables are roughly centered in these aisles (Figure 2, Cross Section XX). If a similar layout was used at Inkallacta, there would be five rows of posts, with one down

41date (Gasparini and Margolies 1980:208). Like those at Hu!inuco Pampa, however, they were large enough to have been small doors, and if so, the stepping downward of their sills makes sense. In so doing, the sill heights reflect the sloping grade outside the building and are between 4 and 5 m (13-16 ft) above the floor inside, about the right height for the first level of bracing between the posts. To be effective, such bracing would have to steady the posts both laterally (N-S) and longitudinally (E-W). If post-to-post braces running N-S were covered with a continuous layer of smaller poles laid E-W at the level of the door sills (Figure 21), the posts would be supported in both directions and an easily accessible layer of scaffolding would result inside the gable ends, simplifying construction of both the high gable walls and upper level pole and roof work (Figure 15). To be strong enough to act as floor beams without becoming excessively thick and to provide bearing for butt jo~ts in the layer of scaffolding poles above, the N-S braces would likely have been double logs, about 30 cm (12 in) in diameter, with one on either side of the posts (Figure 15). To' avoid excessive length, these beams would have spanned only from one post to the next, a distance requiring 6 m (20 ft) poles. This means that they would have necessarily been offset vertically at least one log thickness at each post. This offset could take several forms, but the proposed scheme is suggested by the fact that each door sill steps down about 30 cm, or one beam thickness, from the door to its north. Scaffolding poles bearing on beams offset as shown would therefore exactly match the descending elevations of the door sills (Figure 21). Assuming such scaffolding was used, there is no way to tell whether it extended the full length of the building or, whatever its extent, whether it was retained in the finished building as a permanent loft or removed once construction was complete. If full length and left in place, it would have increased usable floor space by more than 40%--an attractive feature, one would think. Next is an estimate of the longitudinal (EW) spacing of the posts. Two factors suggest

Lee: Inkallacta the layout proposed here. The entire weight of the roof was almost certainly carried by horizontal purlins spanning longitudinally (EW) between the tops of the posts. As with the logs supporting the loft described above, the diameter and length of these purlins were limited by practical considerations such as availability, weight and manageability. Also, there is a direct relationship between their size, strength, span and load. Anyone is given by fixing the other three. Garcilaso claimed that the best Inca thatchwork was typically heavy, sometimes as much "a fathom" thick and extended "a yard" beyond the exterior walls (1987 [1609]:321). Based upon this and other similar reports in the chronicles, 1 have conservatively assumed about a meter of dry ichu grass over a pole framework, which gives a load of about 50 Ibs/ft2. At Hminuco Pampa (Figure 2), the ridge purlins spanned a bit less than 10m (30 ft) and if similarly loaded and done with an average grade (1200 ,psi) of structural timber, would have been about 35 cm (14 in) in diameter. The same' sized logs used at Inkallacta would have spanned only about 6 m (20 ft) due to heavier effective loading resulting from the lower pitch of the roof. With this in mind, we might expect that the posts in the Great Hall were somewhat closer together than those at Hwinuco Pampa. Another difference between the halls of Hu!inuco Pampa and the Great Hall may have influenced longitudinal spacing as well. At Hu!inuco Pampa the posts bear no relationship to the spacing of the doors, but are nearly 5 m (16 ft) inside the building and well out of the way of people entering and leaving. At Inkallacta, the southern-most row of proposed posts is only 2 m (6-1/2 ft) inside the south wall and the E-W post spacing may therefore have been coordinated with the doorways to avoid interference with traffic. Because the doors are 5.5 m (18 ft) apart and this spacing is close to the 6 m purlin span suggested above, the layout proposed here places the posts 5.5 m apart, midway between the doors (Figure 23). At first glance, this seems to create a forest of posts, but the large scale of the building is deceptive. To get a better feel for the space between the posts, refer instead

ANDEAN PAST 5 (1998)

-42
The only example of Inca thatchwork known to have survived into relatively recent times was the dome-shaped roof of the Suntorhuasi at Azangaro, Peru. During his visit there in 1864, George Squier observed a decorative, woven ceiling mat between the polework and the underside of the thatch (1973 [1877]:394). His photograph and drawings of the exterior show heavy, multi-' layered thatch and neither they nor his sketch of the interior indicate any attachment between the roof and masonry, despite what appear to be two projecting "pegs" high on the inside wall (McElroy 1986:102). The features Squier reported were probably also typical of other important Inca buildings. Assuming this, the mat would have been fastened to the polework at the eaves, as it appears to have been at Azangaro. On buildings with pitched roofs, however, the gables at Puncuyoc suggest that the mat extended flush out onto the tops of the gable walls, secured to the recessed "eye bonders" sometimes found there (Lee 1988a). This implies that the top of the polework was flush with the tops of the gables, as shown here (Figure 22). Contrary to the speculations of Bingham (1979 [1930]:78) and others, the polework was not typically attached to the gable walls (Puncuyoc is an exception, due to its exposure to high winds). At Inkallacta, the upper gables were mostly adobe, a material able to support little more than its own weight in simple compression and ill-suited to attachment of anything applying lateral forces. Significantly, there are no eye bonders atop the gables of the Great Hall. Probably, there was no woven ceiling mat above the polework--an understandable omission, given that the building would have been quite dark inside, a loft might have obscured view of the underside of the roof from most of the ground floor, and a mat would have been nearly as large as a football field. Squier also observed that the thatch of the Suntorhuasi was composed of alternating layers of ichu grass and a coarser, net-like lathwork used to hold it together (1973 [1877]:395). Analysis of Puncuyoc suggests

to the sketches with people in them (Figures 15,21,22).


Before turning to the roof covering, there is another important, if largely conjectural, aspect of the framing to be noted. The upper portions of the three tallest rows of posts would have required bracing above the loft level. The lateral X-bracing suggested here (Figure 21) would have been needed longitudinally as well. It is simple, effective and utilizes relatively lightweight poles of manageable length, but there is no way to know what method the Incas actually used. Roof Covering With the timber frame erected, the next step in construction was the thatchwork. Based upon analysis of the unique and wellpreserved masonry detailing of the Incahuasi at Puncuyoc, in the Cordillera Vilcabamba (Lee 1988a), the thatch was applied over a layer of small, horizontal pQles supported on vertical rafters lashed to the horizontal purlins spanning between the posts (Figures 15, 21, 22,24,25). Being .about 7 m (23 ft) long, the rafters would have been spaced to reduce the span of the polework above and to keep their own size and weight within manageable limits. The only'other clue we have as to their probable layout is the configuration of the south eave wall. It was penetrated by numerous doorways, windows and niches (Figures 10, 11) that reduced its strength and resistance to loading from above, especially directly above the lintels. The safest place for the rafters to bear would have been between the openings, where the wall was strongest (Figure 15, arrows), or about 1.4 m (4-1/2 ft) apart. Loaded to 50 Ibs/ft2,they would have been about 20 cm (8 in) in diameter. Horizontal poles about 5 cm (2 in) thick lashed atop the rafters every 30 cm (12 in) would have been needed to support the thatch. If installed progressively as the supporting structure underneath was erected, the poles would also have turned the entire roof into a giant ladder, facilitating the movement of men and materials up onto its higher reaches (Figure 22).

43that this mass extended out over the tops of the gable walls and down to the stone pegs protruding from the exterior face of the gables, where it was secured (Lee 1988a). The pegs at the Great Hall are almost a meter (3 ft) down from the tops of the gables, probably to insure sufficient anchorage into the adobe, a material inherently less secure than well fitted stonework. Finally, consistent with the observations of various writers (Markham 1862:193-194; Squier 1973 [1877]:395; Pizarro 1978 [1571]:161-2; Garcilaso 1987 [1609]:321) both the weather surface and the eaves of the finished thatch would have been neatly trimmed, with the latter providing a generous, sheltering overhang above the south doorways (Figures 21, 22). Construction Sequencing and Coordination As with any large, complicated construction project, building the Great Hall involved the efforts of numerous trades peripheral to the work already described, but equally important to the finished product. Based upon their prolific output, Inca builders must have been particularly good at organizing these trades arid coordinating their work efficiently. Presumably, work progressed simultaneously on as many fronts as possible, as long as the various trades did not interfere with one another. Activities which took place off-site, such as stone gathering, pole and grass cutting and rope making, needed only to be scheduled for proper support of the work at the building. Other trades, however, such as earth movers, plasterers, painters and general laborers had to be coordinated with the masons, carpenters, lashers, and thatchers working within the structure itself. In the discussion of the lower walls, for example, it was suggested that the retaining walls were built first without niches so that earth movers could backfill outside the building while the masons completed the niches inside. This also provided for earlier and easier access to the upper parts of the work than was othenyise possible. Four crews of masons probably started at the comers and increased to eight or

Lee: Inkallacta more as they worked toward the centers of all four walls. A gap may have been left at the center of the south fayade to facilitate erection of the interior posts. Otherwise, the large timbers required would have necessarily been lifted over the completed south wall or threaded through its narrow doorways. Just as the masons worked inward from the comers, other trades probably did the same. Once walls were up, the carpenters, lashers and thatchers could have started from each corner of the roof and built inwards and upwards toward the center, such that four crews for each trade were always working concurrently. Meanwhile, as masons completed the interior niches, plasterers and painters (if any) could have begun finishing the masonry walls. Finally, one of the most troublesome aspects of construction high above the ground is the time, material and energy consumed in moving men and materials up and down the structure and providing, moving and eventually removing the scaffolds necessary to their work. Great savings are achieved if these operations are simplified, minimized or eliminated. At the Great Hall, the building itself was probably designed at least partly with this in mind. The 10ft,with direct access through the gable doorways is one example.

The loweringof the . upper roof pitch and use


of the polework as a giant ladder are others. None of these features required otherwise unnecessary work during construction or the removal of extraneous work afterwards. Conclusion The interior space of the Great Hall was large by any standard, then or now. The ground floor (Figure 23) contained 1737 m2 (18,574 ft2), about the same as a large supermarket or chain store. To this, the loft might have added an additional 1263 m2 (13,505 ft2) of upper level storage or dormitory space. Probably, large public gatherings were reserved for the ground floor because of the heavy loadings involved and because access to the loft was so limited by comparison. How many people might the ground floor have

ANDEAN PAST 5 (1998) accommodated? A lot, probably, depending on how daunting were the conditions outside and how attractive were the activities inside. Absent safety rules limiting occupancy, people can and do crowd very closely on occasion. To prevent this, most modem (US) Building Codes set a limit of 15 ft2 per person in an assembly hall without fixed seating. By that standard, the Great Hall would have had a capacity of 1238 people, with little space left over for whatever activity they were there to witness or participate in. From this we might assume that crowds numbering in the hundreds were not uncommon. As with much of Inca architecture in general, the aesthetic appeal of the Great Hall as reconstructed here is more akin to that of present-day engineering or industrial building than to architecture in the Western sense. Like the Golden Gate bridge or any good New England barn, the Great Hall was true to its purpose (shelter~d assembly, for whatever reason) and to the materials from which it was made, and its designers had the good sense and restraint to let it go at that. Nothing extraneous appears to have been added for decorative effect. The appeal of the design lay not so much in how it looked, though it was undoubtedly a handsome structure (Figures 26, 27), but in the clear and honest relationship between how it looked and what it was. In these plethoric times of pre-occupation with appearance before substance and fascination with guile and glitter, is it any wonder that the simple integrity of Inca architecture is often dismissed as the work of skilled dullards? Upon whom does that assessment ultimately shed the most revealing light? References Cite~
Agurto Calvo, Santiago 1987 Construcci6n, arquitectura y planimiento incas. Lima: CAPECO. Ascher, Marcia and Robert Ascher 1981 Code of the Quipu. Ann Arbor: University of Michigan Press. Bingham, Hiram 1979 Machu Picchu Citadel of the Incas. New York: Hacker Art Books. Originally published in 1930 for the National Geographic Society by Yale University Press, New Haven, and issued as No.1 of Memoirs of the National Geographic Society.

-44
Cobo, Bernabe 1983 [1653] History of the Inca Empire, translated by Rowland Hamilton. Austin: University of Texas Press. Garcilaso de la Vega, EI Inca 1987 [1609] Royal Commentaries of the Incas and General History of Peru. Austin: University of Texas Press.

Gasparini,Grazianoand LuiseMargolies
1980 Inca Architecture, translated by Patricia J. Lyon. Bloomington: University of Indiana Press. Gonzales, Alberto Rex and Antonio Cravotto 1977 Estudio arqueol6gico y inventario de las ruinas de Inkallajta. Paris: UNESCO. Guaman Poma de Ayala, Felipe 1936 [1615] Nueva cor6nica y buen gobierno (del Peru). Facsimile edition. Paris: Institut d' Etbnologie. Lara, Jesus 1988 Inkallajta/Inkaraqay. La Paz and Cochabamba: Los Amigos del Libro. Lee, Vincent R. 1988a The Lost Half of Inca Architecture. Wilson, Wyoming: privately published. 1988b Function, Form and Method in Inca Architecture. Wilson, Wyoming: privately pub. lished. 1991 Seven "Inca Pucaras" on the Bolivian Frontier. Wilson, Wyoming: the author. 1992 Cuzco-tuyo -- the Search for a Lost Inca Fortress. The Explorers Journal 70(4):118-123. Markham, Sir Clements 1862 Travels in Peru and India. London: J. Murray. McElroy, Keith 1986 Ephraim George Squier: Photography and the Illustration of Peruvian Aptiquities. History of Photography 10(2):99-129. Morris, Craig and Donald Thompson 1985 Huanuco Pampa -- An Inca City and its Hinterland. London and New York: Thames and Hudson. Niles, Susan 1987 Niched Walls in Inca Design. The Journal of the Society of Architectural Historians 46(3):277285. NordenskiBld, Erland 1924 Forschungen Und Abenteur in Sudamerika. Stuttgart: Strecker und Schroder. Pizarro, Pedro 1978 [157I] Relaci6n del descubrimiento y conquista de los reinos del Peru. Lima: Pontificia Universidad Cat6Iica del Peru. Sanniento de Gamboa, Pedro 1907 [1572] History of the Incas. London: Hakluyt Society, Second Series 22. Squier, Ephraim George 1973 [1877] Peru: Incidents of Travel and Exploration in the Land of the Incas. Reprint. New York: AMS Press.
.

45-

Lee: lnkallacta

...
.,"~

SOUTH
o 000

AME.fc/CA
t
1000 mtl~J

Figure 1. Location of Inkallacta.

~ ~ ~
l!!!j
""d

> 00
E:1e/r'ltu. Pampa

I)~~'NWd~~
85M
x

~
I

lr~t.. wall

..... \0 \0 00 '-"

CJt -.

fj;

port ho/~l_'*'

;iF

'It

*"
fiuanuco Pcml2Q

x
73M

r ...J.L.
A::>

o /ndUhncr~
~~~~~~:_-~_~~~~-

IT I

mer.:,-J

Ir/Jk~=;:~

'--56J.1

RoJa~fbfu..J

man
CROff JeCTION x- X

Figure 2. Plans of kallankas from different Inca sites.

I ... 0\

47-

Lee: lnkallacta

'"

. ~ -=: .-~

Figure 3. Royal palace at Cuyusmango in Cusco, after Guaman Poma de Ayala (1936 [1615]: 329 [331]).

8t-

(8661) S .LSVd NVJI<INV

49-

Lee: lnkallacta

N. 1110g.

COCHA8AMf!JA J SOLI 1"14

Figure 5. Plan of Inkallacta, with Great Hall shaded.

~ ~
~
"'d

>00
--7tSIvt
r<!ta/f7/nq

-3 .CA
\0 \0 00

'-'
"\
y

wallf

T
~
\D <\I

( p~
I

'--fcl/'~ ,. "~"':""
~j
~.i!:j~./

flun< #~

fallt!17adobe

fal/~17 Jtomf

I ,

i
I

.11;J}"'~\ ..\ .\
1
mal7 ",~\.:::.,./,;,,,,

(..adobt!

~/:

l
l
r.:, ~:

,<

~ ~
.

,~~ . 1
. \ ~ ~~ .. I "C;' /

\: \

t I~;~;''''''

Jlapt!. appx. .3"1/1


I

h. ,.

.-c::.."-

~ ~Y ~ ::'.....

\ \~':'.

~ \
1

~ -I:: ~
~ ~

I drctnage.../

p/atrorm__
r/. bOIJ/d<!rJ

,p/ozaJ

y o mt:to-J
10

GI<EAT !tA.LL AT INKALLACTA, 8oL1VIA /qQO

\- \ norfh \

Figure 6. Existing floor plan of Great Hall at Inkallacta.


I V'I o

.,
VI

.,
...

. "

., \\"\,

OIlE:AT !fALL AT INKAJ.LACTA /990

/.'//'

..,

CRor. JC:CTlONY-Y o
!J f77d~rJ

".
~ ~

..'..//'-prt::J./t:Cft:d 9Obk-'.'\,\,
-"<-'.."...1';.

h1lIJlde

/ ~yono'

.../ .~.'.
'"

"""

rdtJlmng f./Jall

dl'1:1lt1aqc

.lIPOIc; floor .IlofXJ

OPPJr. .:JM

--l T
I

~ h
~
grade behind wall

I In i~

;. ,'\j

,~
--,5101

t--< (\) (\)

'.

~ ~
Figure 7. Existing section through Great Hall at Inkallacta, looking east.
::::::

~ ~

ANDEAN PAST 5 (1998)

-52

Figure 8. Plastered niche in Great Hall at Inkallacta.

53-

Lee: Inkallacta

Figure 9. Plastered northwest comer in Great Hall at Inkallacta.

~
~
t!I!j

~
'"=

> 00
,-..
"-"
\0 \0 00

~ u.

Figure 10. Free-standing south wall of Great Hall, Inkallacta.

I VI ~

55-

Lee: lnkallacta

ANDEAN PAST 5 (1998)

-56

57-

Lee: lnkallacta

> Z ~ ~

~ ~
~
00
!.II

,,- JC TcJ

-..

,-

mOJOnf\
,.

\0
\0

00 --

'-.J

~
' L- .

.I

-'

. /

tJtnnq
lIne!

~
-'
.~.

.,---

WALL!

;:~
I VI 00

:~'

Figure 14. Suggested wall construction techniques using string lines.

v..
\0 I

ROOF

-~,-

,.,~.

t'-<

Figure 15. Suggested roof and loft framing at east end of the Great Hall, Inkallacta.

.. ~ ~ ~ C') is"

(\) (\)

ANDEAN PAST 5 (1998)

-60

Figure 16. Southwest gable comer, Great Hall, Inkallacta.

61Lee: Inkallacta

Figure 17. Southeast gable comer, Great Hall, Inka11acta.

ANDEAN PAST 5 (1998)

-62

197. Inkall:aqta. The cllcularion of the length uf (he NO slopes of the roof. '98. fnkaUaqta. PuWble sauaural 'YICcmlor swtaining the roof of the k.ILtllk.. Buausc of iQ;width. it mu.1C huc h~d tour naVeL

.0..

~I

0
o om 0
o

omo
211

Figure 18. Reconstructions of gables at Inkallacta (from Gasparini and Margolies 1980:211).

63-

Lee: lnkallacta

~ o
~

l!!j

> 00
~

.\0 \0 00

!J8

'-"

Figure 20. Fallen west gable end of Great Hall at Inkallacta (model).

I 0\ oj:>.

0\ VI I

ORcAT
./ M'"lafh (doubiu
.3 M " pok.rofrerJ AM" pok. ptJrlmJ

HALL AT INI<AI.LACTA

allodd<:.rdu.r,ng conJfructlon)
.... e /.4 !>1 o.c. e 4. 'MOC.

ca. /520
CROll JECT/ON

Z-Z
~

/71dcrJ

..r:"

'~'i

'0'

./trmg

Itn(.

n
0

'0"

".-0
.

U1

I rr [J .'.
"

.)
.." . n t " r]'"
. I. .'. .'. .3M floor
"

.411

'"/0<) polrJ

l)

/
~

~;
doub~

.1 IJ ...

!!'

. ,a....

dnp of I7XJf

t I 1...

JOJJtJ t--o ~ ~

'.

~ ~ :::::
Figure 21. Proposed section through Great Hall, Inkallacta, looking east.
~. (')

0\ ...,J I

hoaR.

AREAJ 1737,Nt' (18,700$) IZ6,M' (13,600"') 3000M'(~,.JOO"')

r44i
;

ground: loft: totol:

Z, I
.."',"rYl.':JT-;';;';.r",-__, .ii~.,:'...:>..t-_"'5"'

'..

1'J:"":~-:-:'71:";-:-:::-._;-.'''', . '-Lt.'L' "'~~:;-::r"0L..'::'7~{~._

--~
i I

- --

--_.._'

If
I

4.~M

'..L___J
Jog poftJ.J

L__ _A

-A

. ---t--~ ~--I

;.

"4101"1

I
I

. . . . . . . . . . . . .,-- . . . .. .. .. - _I I
..

I .1 I II II
I I I' I

,: I

Ii

I
I

..

/cd9c! --.--------.

ordukabovc1'

.----.---.-

---.-

~
man

z L---

GIlEAT ttAU AT INKA/.lACTA IJoLIV/A ca. 1520

/0

-\\

.,
norfh

mf!tcr.f

t--<
(\) (\) '.

Figure 23. Proposed floor plan of the Great Hall, Inkallacta, showing suggested post locations

~ ~ ~ ~ ~

~
~
"'C

~
~ u.

> 00-

......... ..... \0 \0 00 '-"

t.
Figure 24. Proposed roof construction of the Great Hall, Inkallacta (model). ~ 00

0\ \0 I

t--. ~ ~ ".

~ ~
Figure 25. Proposed roof construction of the Great Hall, Inkallacta (model).
~ (")
::::::

i3'

~ ~ ~ ~
~
--.
(.II

~ >00-

\0 00 ~

\0

Figure 26. Reconstructed external appearance of the Great Hall, Inkallacta, from the southeast (model).

I o

:I

I
'-J

Figure 27. Reconstructed external appearance of the east gable end of the Great Hall, Inkallacta (model).

t--t (\) ~ ~ ~ ~ (') ~

RECONSTRUCTING ANDEANSHRINE SYSTEMS: A TEST CASE FROM THE XAQUIXAGUANA(ANTA) REGION OF CUSCO, PERU Brian S. Bauer
University of ll/inois, Chicago

Wilton Barrionuevo Orosco Cusco Introduction


Systems of huacas (shrines), organized along ceques (lines), radiating out from village centers were common features of pre-Hispanic communities in the Andes. One such system was located in Cusco and it has received extensive attention from a number of ethnohisto~ rians and archaeologists. Other. pre-Hispanic shrine systems in the Andean higWands have not, however, been clearly identified. In this report, we analyze the distribution of some 22 shrines in the Province of Anta, Departme\1tof Cusco, to determine whether evidence for a pre-Hispanic shrin~ system can be detected in this region of the Andean highlands. The Cusco ceque system, composed of some 300 to 400 huacas conceptually organized along 42 ceques which radiated out from the center of Cusco, has long held a prominent position in ethnohistorical (Rowe 1985; Sherbondy 1982, 1986, 1987; Zuidema 1964), and more recently archaeological, research on Inca ritual complexes (Niles 1987, Bauer 1992). Cusco was not, however, unique among Andean communities in containing such a shrine system. Cristobal de Albomoz (1984 [ca. 1582]:218) indicated that Andean shrines were frequently organized along lines, and Juan Polo de Ondegardo (l916b [1571]:56-57), investigated more than one hundred systems during his travels across the Andes. Around 1571 the Bishop of Charcas, who doubted that ceque systems were as universal as Polo de Ondegardo claimed, was shown such a system in Pocona (Bolivia), by this same Spaniard. Despite the fact that ceque systems may have been common features of pre-Hispanic communities in the Andes, the documentation of such systems outside the Cusco Valley has so far largely eluded archaeologists and ethnohistorians.1 Ethnographic studies indicate that complex systems of shrines exist in modem communities in Bolivia and Chile (Albo 1972; Barthel 1959; Hadingham 1987; Metraux 1935; Morrison 1978). However, the organization and antiquity of these systems, and their possible relationships with the Cusco system, remain to be documented and analyzed. In addition, while a number of archaeologists have used generalized notions of ceques in their interpretations of intra-site remains at pre-Hispanic occupations (Anders 1986; Hyslop 1985; Morris 1990), additional archaeological and historical information is needed before the locations of possible shrines around these pre-Hispanic occupations are identified and site-specific ceque models can be proposed. The identification of. ancient ceque systems in the Andes is understandably difficult because the Spaniards initiated a series of brutal campaigns against indigenous religions, which were largely focused on the discovery and destruction of huacas. The purpose of this report is to illustrate, however, that there may be more information on Andean ceque systems than previously recognized, and that some of that information is contained within the texts written by the leaders of the extirpation of idolatry movements. More specifically, using historic information recorded by Cristobal de Albomoz (1984 [ca. 1582]) and recently collected archaeological data, we will analyze the distribution of some 22 shrines in the modem Province of Anta, Department of Cusco, to determine whether evidence for a pre-Hispanic ceque system can be detected in this region of the Andean highlands.
I

For an example of a long-distance ceque from Cusco

to the pass of Vii canota see Zuidema (1982).

ANDEAN PAST 5 (1998):73-87.

ANDEAN PAST 5 (1998) Bernabe Cobo and the Cusco Ceque System Most of what is currently known concerning Andean ceque systems comes from the 1653 work of Bernabe Cobo, Historia del Nuevo Mundo [History of the New World]. Four chapters of this chronicle (Cobo 1956:169-186; 1980:14-61; 1990:51-84 [1653:Book 13, Chapters 13-16]) were devoted to describing some 328 huacas that surrounded Cusco, and the ceques along which the shrines were organized. Beside providing information on the name, number, and order of shrines on each of the lines, Cobo' s work preserved data on the physical forms of the shrines (springs, caves, boulders, mountain passes, etc.) and on the objects offered to them (gold, silver, cloth, shells, coca, corn, etc.). Cobo indicated that the focal point of the lines was the Temple of the Sun (Coricancha) in central Cusco and that the lines extended outwards from .this structure into the four quarters, or suyus (regions), of the Cusco Valley. Cobo suggested that three of the suyus, Chinchaysuyu.,Antisuyu, and Collasuyu, contained nine ceques each, while the fourth, Cuntisuyu, contained fourteen or fifteen. He also noted that the ceques in each of the four suyus were enumerated in sets of three and that certain kin groups of Cusco were responsible for making offerings to the huacas on specific ceques (Bauer 1992; Rowe 1985; Zuidema 1964, 1983). While Cobo's description of the Cusco ceque system, is by far the largest work on the system, a second, less well known description was written independently by Albornoz around 1582. Albornoz and Andean Shrine Systems Albomoz's account preserves the only detailed description of the Cusco ceque system outside the 1653 work of Cobo. Albornoz was one of the principal leaders in the Spanish campaign against the autochthonous religions of the Andes in the post-conquest era. From 1568 until his death in the early l600s, Albornoz led a series of expeditions in the highlands to identify and destroy native shrines and to punish those individuals and communities who continued to worship them. One of the

-74 largest of these anti-idolatry campaigns was in the Huamanga area (modern Ayacucho) where, from 1568 through 1571, Albornoz crusaded against the millennarian Taqui Onqoy movement. During this period he was personally responsible for the destruction of thousands of huacas and the persecution of a vast number of individuals. Soon after his' Huamanga campaign, Albornoz (1984 [ca. 1582]) wrote an essay, Instruccion para descubrir todas las guacas del Piru y sus camayos y haziendas [Instructions to discover all the huacas of Peru and their camayos (specialists) and wealth]. 2 In this work, Albornoz first described various types of shrines which he identified during his extirpation campaigns, and then he presented lists of shrines in the area of Chinchaysuyu. In the final section of his Instruccion . . ., Albornoz offered advice on how to discover and how to destroy the remaining huacas of the Andes. In the middle of his Instruccion . . . Albornoz described various sl1rinessituated between Cusco and the northern frontier of the Inca empire, as well as those of the central-south coast. The catalogue began with a list of 37 shrines in the Chinchaysuyu region of Cusco, the majority of which also appeared in the Chinchaysuyu section of Cobo's ceque system description (Rowe 1980). The huacas of other highland areas are then described in relation to their distance from Cusco; the shrines of the Xaquixaguana (now called Anta) and Calca valleys -- both located just north of Cusco -are presented next and are followed by those of the Quichuas, Changas and Aymaraes, Soras, Parinacocha, Angaraes and Chocorbos, Hanan Guancas, Tarmas and Atabillos, Guaylas, Guanuco, Carua Conchuco, Guamachuco and Caxamarca, Caxamalca, Paltas, Tomebamba, Puruay, Chica, Chachapoyas, Quito, Angamarca Luytun Cuchu, and Cayambes. After discussing the huacas of these highland regions north of Cusco, Albor2

HenriqueUrbanoand PierreDuviols(1989) have re-

cently provided a new transcription of Albomoz's Instruccion . .. Unfortunately this edition is missing most of the information on the Cusco huacas recorded by Albomoz.

75noz presented brief discussions of coastal shrines in the Lima, Hacari, Piscoy, Yca, Chincha, Luna Guana, and Ychima areas. The apparent purpose of these shrine lists was to illustrate that a large number, and a great diversity, of huacas were still being worshipped. It must be noted, however, that the form of Albomoz's list of shrines in the Chinchaysuyu sector of Cusco varies distinctly from that included in Cobo's chronicle. Cobo began by stating that the Cusco shrines were organized along ceques that radiated out from the Temple of the Sun and that the huacas along each ceque were the responsibility of different kin groups. He then presented systematic descriptions of some 328 shrines in accordance with their positions along 42 ceques. Albornoz's account is far less complex; he simply recorded a list of 37 shrines in the Chinchaysuyu region of Cusco. Nevertheless, Rowe (1980:76) has clearly illustrated that the order of Albomoz's shrines is consistent with information presented. ~y Cobo for Chinchaysuyu, and that Albomoz's work represents an independent investigation of the ceque system. Albornoz and the Huacas ofXaquixaguana (Anta) Unlike Cobo, who recorded only the huacas of Cusco, Albomoz listed the names of huacas in the imperial city of the Inca as well as those in a number of other locations north of Cusco and along the central coast of Peru. Given the fact that Albomoz understood that Andean shrines were frequently organized along lines, and that his Cusco list records the huacas of Chinchaysuyu along their ceques, it is possible that his other huaca lists preserved evidence of other ceque systems. In 1991, field research was conducted in the Anta region to determine whether or not Albornoz's huaca list for this area is organized around similar principles as his Cusco list. The Anta area was selected for intensive investigation because Albornoz provided the names of 22 Anta shrines; a sum which is second only to those of his Cusco list. The research was conducted under the following test assumptions: if an analysis of Albornoz's

Bauer and Barrionuevo: Shrine Systems in Anta Anta huaca list and modem field work data from the same area suggested that the shrines were organized along lines which radiated out from a central place, then there would be strong circumstantial evidence that a preHispanic ceque system existed in this region; if on the other hand, after an analysis of the research data there appeared to be no clear order in the positions of the shrines across the landscape, then it could be concluded that AIbomoz's information provided no apparent evidence for the existence of a ceque system in Anta. It should be noted that this second assumption (i.e., the null hypothesis) does not negate the possibly that a ceque system existed in the Anta region, it simply suggests that there is no obvious evidence for such a system in Albomoz's data. The Province of Anta and Field Methodology The Province of Anta, located approximately 20 km northwest of the city of Cusco, is composed of eight districts: Pucura, Cachimayo, Huarocondo, Zurite, Limatambo, Mollepata, Chinchaypuquio, and Anta which serves as the provincial capital. The province is the best known for itS enormous pampa, through which the royal Inca road from Abancay to Cusco once crossed. It is also known for the pass of Vilcaconga, located between towns of Limatambo and Zurite, where the Inca attacked invading Spanish forces in 1533 and near where Gonzalo Pizarro was defeated by the loyalist Pedro de Gasca in 1548.3 Although little research has been conducted on the pre-Hispanic social organization of the Province of Anta as a whole, it is known that the area immediately surrounding the community of Anta was traditionally organized into four ayllus (kin groups) called: Anta (also called Coilana), Hequeco, Sanco, and Conchacalla, each of which occupied
The pass of Vilcaconga is also mentionedby Cal-

lapifia et al. ('1974 [1542/1608]:56), Cieza de Leon (1976 [1553, Part 1]:135, 1979 [1554, Part 11]:225, Chapter 95]), Santillan (1950 [1564]), and Segovia (1943 [1553]).

ANDEAN PAST 5 (1998) separate villages (Villanueva Urteaga 1982: 192). The locations of these four villages are important because they are each listed several times in Albornoz's description of the huacas of the Anta region. The areas of Sanco and Conchacalla ayllus are located to the northwest and southwest of the community of Anta, and that of Hequeco ayllu is situated to the northeast. The relative positions of these four ayllus are also reproduced in Anta's plaza where it is said that the Hequeco and Sanco ayllus are traditionally associated with the northeast and west sections of the town, and that the Anta (or CoUana) and Conchacalla ayllus are associated with the southeast and west sections. Field work for this study began with project members interviewing local people in areas of Anta thought on the basis of documentary evidence to contain particular shrines. Local inhabitants were interviewed in Quechua to find shrines that.had retained their original names. A toponym was confirmed when three independent informants provided similar answers, A positive identification of a huaca was made when the description provided by Albornoz matched the physical features of a specific object (e.g., a cave, spring, or outcrop) that had also retained the name of a shrine. Unfortunately, because Albornoz provides so little information on the shrines, only a small number of shrines could be identified with certainty. In many more cases, only likely possibilities are presented. Furthermore, we are unable to provide even likely shrine candidates for several of the shrines described by Albornoz as in the Anta area. Field work was also complicated by the fact that only a poor copy of Albornoz's Instruccion . . ., which contains numerous spelling errors, has been preserved. As noted by Rowe (1980:72), "It is quite possible that the copy we have is only a summary of the original, and that the original listed many more shrines." Furthermore, because a comparison of Cobo's and Albomoz's shrine lists for the Chinchaysuyu region of Cusco (ibid.: 76), indicates that there are deficiencies in both of these Cusco accounts, it seem most probable that there were also numerous hua-

-76
cas in the Anta area that were not recorded by Albornoz. Also, his work is not especially early, written one 9r two generations after the arrival of the Europeans. In other words, there are clear and certain limitations in using Albornoz's lnstruccion . . . as a primary source for researching prehistoric ceque systems: it is hard to read, poorly written, and almost cer-' tainly incomplete. Nevertheless, because various chroniclers, Albornoz among them, suggest that ceque systems were widely present in Andean villages, it is a case study well worth exploring. Research Results The 22 Anta shrines recorded by Albornoz (1984 [ca. 1582]:205-206) are presented below, with translations. We have included short commentaries after each shrine description~ noting if our field research provided any additional information on the huaca. Their locations are shown on :figure 1. 1) Oyfiacaca, guaca de los indios guarocondos; es una pefia al pie de un cerro. (Oyfiacaca, huaca of the Guarocondo Indians; it is a boulder at the foot of a mountain.) While the community. of Huarocondo is located on the northern edge of the Anta area, no additional information is available for this huaca. 2) Rutucayan, guaca de Anta, piedra figura de hombre. (Rutucayan, huaca of Anta, stone figure of a man.) Immediately south of the village of Anta is a hill called Rutucayan. A good candidate for the huaca of Rutucayan is an outcrop high on the slope of this mountain which the inhabitants of Anta believe to be in the form of a man. 3) Anta, piedra pacarisca4 de los indios antas. (Anta, origin stone of the Anta Indians.)
4

Most Andean kin groups traced their lineage back to

mythical ancestors who emerged from the earth at sacred locations called pacarinas (origin places).

77A good candidate for the origin stone of the Anta Ayllu is a cliff with a cave, called Runa Pacarisca (origin place of people), south of Anta (Figure 2). 4) Ayaco, piedra guaca de los indios cercanos. (Ayaco, stone huaca of the neighboring Indians.) There is a large mountain near the community of Ayllu Mayo with numerous outcrops called Ayacco which may represent this shrine. 5) Achapay, guaca de piedra muy labrada. (Achapay, a well-worked stone huaca.) One of the most famous carved stones in the Province of Anta is Quilla Rumi (Moon Stone), an elaborate sculpture in {heshape of a half-moon near the base of a large outcrop. Because the area beside this carved stone is called Acchapay, it is very likely that Quilla Rumi is the well-worked stone shrine mentioned by Albornoz (Figure 3). 6) Timpay, cueba'en.un cerro de los indios de Mayo. (Timpay, cave in a mountain of the Mayo Indians.) A very good possibility for this shrine is a cave called Tocco Ccaca (Stone Opening) in the community of Ayllu Mayo (Figure 4). 7) Panara, guaca de los indios dichos mayos. Piedra encima de un cerro. (Panara, huaca .ofthe said Mayo Indians. Stone on top of a mountain.) Survey work in the Ayllu Mayo area found a hill slope called Pacara which may be related to this shrine. 8) Llimillay, guaca de los indios de Canco, en el dicho valle; eran diferentes piedras. (Llimillay, huaca of the Canco Indians, in the said valley, they were different stones.) There is a rocky hill called Llimillay in the Sanco area which may be this huaca.

Bauer and Barrionuevo: Shrine Systems in Anta 9) Uicacayan, guaca de los indios de Hequeco; es una piedra encorvada. (Uicacayan, huaca of the Hequeco Indians; is a curved stone.) Although the community of Hequeco is situated to the northeast of Anta, no additional information is available for this shrine. 10) Mapiguaca, del mismo pueblo de Hequeco; piedra figura de indio. (Mapiguaca, of the same town of Hequeco; stone figure of an Indian.) The hacienda of Mapi is on the northern edge of the Anta Pampa, in the territory of Hequeco. It seems reasonable to assume that the shrine of Mapi was located somewhere in this general region. 11)Pilco guarda, guaca de los indios de Conchacalla, una piedra puesta en un cerro grande. (Pilco guarda, huaca of the Conchacalla Indians, a stone placed on a large mountain.) It can only be suggested, based on current data, that this shrine was located somewhere near the village of Conchacalla, southwest of Anta. 12) Guanacauri, 13) Anaguarque, e 14) Auiraca, guacas en el dicho valle, tres piedras en un cerro en memoria de las del Cuzco. (Guanacauri, Anaguarque, and Auiraca, huacas of the said valley [of Xaquixaguana], three stones on a hill in memory of those of Cusco.) There is a mountain called Huanacauri to the north of the community of Zurite and there are three stone outcrops near its base, which the inhabitants state were once brothers. It is possible that these three stones represent the huacas of Guanacauri, Anaguarque, and Auiraca. 15) Curicancha, en memoria de la del Cuzco, con estatua de piedra. (Curicancha, in memory of the one in Cusco, with a stone statue.)

ANDEAN PAST 5 (1998) A community called Corichanca, on the eastern edge of the Anta pampa, may have been related to this shrine. 16)Tambocancha, casa que fue de un yngay tenia su figura de oro en la dicha casa; llarnase Tupa Ynga Yupanqui. Tenia muchas haziendas y riquezas esta casa y camayos. (Tambocancha, a house which was of an Inca and he had his gold figure in that house; he w~ named Tupa Inca Yupanqui. He had great riches and many treasures in this house and camayos [specialists]). This shrine was almost certainly located at an archaeological site called Tambocancha which contains the poorly preserved remains of several Inca structures around a plaza (Figure 5). 17)Uilca conga, guaca general de todo el Pin'1 y Ie hazia todo el Pin'1 cacchaui y Ie ofrecian y servian. Es .donde dieron la batalla a los espafioles. (Uilca conga, general huaca of all Peru and all of Peru made it cacchauis and they made offerings to it and served it. It is where they did battle against the Spanish.) Vilcaconga is a well-known area at the western end of the Anta Pampa. 18)Maragoac;i guanacauri, piedra donde hazian muchos sacrificios en reverencia del Guanacauri del Cuzco. (Maragoac;i guanacauri, stone where they made many sacrifices in reverence to the Guanacauri ofCusco.) The area of Marca Huaci, which is known for its large Inca settlement, is southwest of Limatambo. Marca Huaci is also mentioned in Molina's (1989 [ca. 1575]:74-75) description of the Situa festival of Cusco.
Albomoz indicates near the end of his Instruccion . . . that the word cachaui held the same meaning as ceque. This is further clarified by Duviols (1984:222 n. 2), who suggests that cachaui has to do with the expiatory rite of Capacocha and comes nom cacha (messenger).
5

-78

19) Guaypon guanacauri, piedra cerca de una laguna. Aqui se horradavan las orejas los indios Cuzcos. (Guaypon guanacauri, a stone near a lake. Here they pierced the ears of the Cusco Indians.) There is a small, steep mountain called Huanacauri beside Lake Huaypon. The location of the shrine on this mountain may be' marked by an outcrop called Pito Ccacca which was described to us as "enchanted". 20) Chinchero guanacauri es una piedra cerca de la dicha guaca de la laguna Guaypon. Tiene otras muchas guacas por allegados anssi. (Chinchero guanacauri is a stone near that huaca of Lake Huaypon. It has many other huacas near it.) On the other side of Huanacauri mountain, is a second cluster of "enchanted" outcrops, called Condorcaca, which may be related to this shrine.
_

21) Pancha guanacauri, una piedra quesUi en un cerro que se llama Paneha, junto a la laguna de Pongo. (Pancha guanacauri, a stone which is on a mountain called Pancha, beside Lake Pongo.) The community of Pongo Bamba is ated beside Lake Piuray, and there is a ridge above the village called Encanto chantment) which may be related to shrine. situhigh (Enthis

22) Racra guanacauri es una guaca pueSta en otro cerro frontero de la sobredicha. (Racra guanacauri is a huaca placed on another mountain in front of the abovementioned one.) No additional information is available for this huaca. Albornoz also tells us:
Trafan consigo los indios otras muchas bestiduras de guacas de fuera de sus tierras y bestfa[n] piedras con ellas y les hazia[n] muchos sacrificios. Y ansimismo mochan los valles de Yucai, Calca y Lamai a estas dichas guacas deste valle de Saquixaguana, sin las que tienen en sus pueblo[s] [yJ probincias que

79se declaran. (The Indians also brought with them many vestments of huacas from outside their territory, and they would clothe stones with them, and, make many sacrifices to them. And likewise the [people of the] valleys of Yucai, Calca y Lamai worshipped those said huacas from this valley of Saquixaguana, in addition to the huacas mentioned, that are located in their villages and provinces.)

Bauer and Barrionuevo: Shrine Systems in Anta In this work, we have examined a list of shrines provided by Albornoz ( 1984 [ca. 1582]) in his essay Instruccion para descubrir todas las guacas del piru y sus camayos y haziendas to determine whether new information concerning Andean shrine systems can be derived from it. Since Rowe (1980) showed that Albornoz's list of Cusco shrines correlates closely with Cobo's description of the Cusco ceque system, it was proposed that Albornoz's other regional shrine lists may also record the existence of other ceque systems. In other words, it was suggested that the order of shrines in Albornoz's lists may reflect uniquely Andean systems of shrine organization (i.e., shrines organized along lines which radiate out from a central point), as they have been shown to do for Cusco. The Xaquixaguana shrine list was selected as a test case, because apart from that of Cusco, it is the largest list of huacas provided by Albornoz. Through field work in the region, possible areas have been suggested for 18 of the 22 recorded shrines. These shrines seem to circle the plain of Anta. Although the limited number of shrines presented by Albornoz for the Anta region makes the definitive identification of ceques difficult, an argument can be made for the existence of two possible lines. The locations of shrines 5, 6 and 7 as well as those of shrines 16, 17 and 18 extend out of the western end of the Anta Valley along the general course of the Royal Road of Chinchaysuyu (Figure 1). It is possible that the locations of these two sets of three shrines mark the courses of two separate ceques. In short, while there appears to have been a Xaquixaguana shrine system (i.e., a set ofrecognized shrines which surrounded the region), there is only marginal evidence that the shrines of this system were organized along lines, similar to those of the nearby Cusco area. The Xaquixaguana shrines and other ritual systems It is apparent, from the information presented by Albornoz, that there was a small scale system of shrines surrounding the great plain of Xaquixaguana during the post-

Summary and Discussion One of the most difficult aspects of studying Andean ceque systems is that there are few detailed data sources on them. The dearth of information on these systems is surprising because a number of literate Spaniards were aware of the Cusco ceques and of analogous systems in other communities. Jose de Acosta (1954 [1580] Book 5,. Chapters 910]:560-561, 562) and Cristobal de Molina (1989 [ca.1575]:126) as well as Pedro de Cordoba Mexia (1900 [ca. 1572]:396) and Viceroy Toledo (1924 [1572]:394) obliquely mention systems or orderings of Andean shrines, but provide no substantive information on their internal organization. Similarly, Juan Polo de Ondegardo (1916a [1585]:43; 1916b [1571]:55-57) indicates that he wrote an account and made a map of the Cusco ceques and that he investigated a large number of other systems in the Andean highlands; yet his known writings provide few details on the internal arrangement of these systems. The Spanish participants of the antiidolatry movements were specifically instructed to record the names and locations of the shrines which they destroyed so that the huacas could be revisited and inspected for evidence of continued use at "alater date. This suggests that additional information on Andean shrine systems may still await discovery in regional archives. For example, Albornoz (1984 [ca. 1582]:207) indicates that he personally destroyed a large number of huacas in the area of the Changas and Aymaraes -- he claims more than two thousand -- and states that their destruction was recorded in parish records. Other leaders, such as Pablo Joseph de Arriaga (1968 [1621]) and Francisco de Avila, and their followers, almost certainly did the same.

ANDEAN PAST 5 (1998) conquest period. While the worship of these shrines was largely restricted to the inhabitants of the immediate region, this is not to say that the system functioned completely independent of the Cusco system or those of other nearby areas. There is even some evidence indicating that there may have been a series of overlapping shrine systems throughout the Andes. For example, at the end of his description of the Xaquixaguana shrines Albornoz states that people of Yucay, Calca, and Lamay also worshipped the shrines of the Anta area as well as those of their own regions. Another example of overlapping ritual systems can be inferred from the presence of Marca Huaci (#18) on Albomoz's Xaquixaguana list. This large Inca site is mentioned by Molina (1989 [ca. 1575]: 74-75) within his description of the Cusco Situa ritual. During this elaborate celebration, held during the ninth month of the Inca calendar,.the imperial city was ritually cleaned. At its close, four groups of one hundred warriors carrying ashes ran out from the central plaza of Cusco along the four royal roads of the empire. When the. runners reached the edge of the Cusco Valley, the ashes were passed to representatives of other ethnic groups who threw them into the major rivers of the region. The runners of Chirichaysuyu went through the Anta area on their way to Tilca, which is above Marca Huaci, and then deposited the ashes in the Apurimac River. There is also evidence to suggest that at least one of the Cusco ceques crossed into the Anta area and thus overlapped with its smaller, apparently less complex system. While the exact course of the ninth ceque of Chinchaysuyu is poorly understood, the eighth shrine of this line stood near the western end of the Anta plain. This shrine, Queachili (Chinchaysuyu Ceque 9, huaca 8), is registered by Cobo (1980 [1653:Book 13,Chapter 13]:29) as a flat place "which is between two hills like a gateway; in it the said victory [over the Chancas] was completed. .." Albornoz (1984 [ca.1582]:204) also includes this shrine within his Cusco list writing, "Oma chilligues, a plain where the Incas had a battle with the Chanca and they defeated them; and the Chanca fled, and they say that they turned into condors and escaped. And thus most Chanca

-80
ayllus are called condor guachos." While most of the early chroniclers of Peru also mention the Chanca war, Cabello de Balboa (1951 [1586: Part 3, Chapter 14]:299) provides especially important information, noting that the land of Quiachilli, where the last battle took place, was "behind" a place. called Ayavira.6 The small village of Ayaviri lies just west of the area ofVilcaconga mentioned by Albornoz as a shrine of Xaquixaguana. This pass area,' which marks a watershed between the Apurimac and Urubamba Rivers, was the site of a series of battles in the Conquest and Early Colonial Periods. Although the name Quiachilli is no long used by the inhabitants of Anta, it is mentioned in local land documents dating from 1566 to 1809 (Archivo General de la Nacion [AGN] Real Hacienda, Tribunal de Cuentas, Composicion de tierras de indigenas Leg. 5, 1643-1717; Archivo Historico Departamental, Cusco [AHD] Cajas de Censos: Leg. 4, 1687-1697; AHD Cajas de Censos: Leg. 20, 1802-1809). Other associations. between the Anta shrines and those of the Cusco area are reflected in the names of some huacas. For example, Albornoz specifically states that three of the shrines in the Anta region (Guanacauri [12], Anaguarque [13], and Auiraca [14]) were named after shrines in Cusco. Furthermore, one of the shrines (15) was called Curicancha (sic Coricancha), a name derived from the famous Sun temple in Cusco which was the focal point of the Cusco ceque system. Although the results of this test case are ambiguous, they are, nevertheless, encouraging. The spatial distribution of the shrines around the Anta region suggests that a system of shrines once existed there, although it may not have been organized in ceques. This system, and other regional shrine systems, funtioned independently of the Cusco ceque system and were maintained by the local groups that venerated them. However, they were also related to more general ritual systems, through the incorporation of major local shrines into
Rowe (1980:9) errs in relating the name of Ayavira with that of Apuyauira (Chinchaysuyu Ceque 9, Huaca 6).
6

81larger systems and through the duplication of names. Our findings indicate that is possible to combine archival and archaeological information to identify Andean shrine systems outside of the Cusco Valley, even when such systems are not explicitly mentioned in the text. Once additional systems have been identified with certainty, they will provide a comparative data base to study the Cusco system and to address broader questions concerning basic organizational principles of pre-Hispanic Andean communities. Acknowledgments This work has profited by critical readings provided by Martina Munsters, Jean-Jacques Decoster, and the anonymous reviewers of Andean Past. Field work was conducted under the auspices of the Cusco Ceque System Research Project. Major funding for the project has been provided by the National Endowment for the Humanities, The Nati9nal Science Foundation, The Skaggs Foundation, The Guttman Foundation, The Institute for New World Archaeology, and the University of Chicago Housing System. References Cited
Acosta, Jose de J954 [J580] De procuranda indoTUmsalute 0 predicacion del evangelio en las Indias. In Obras del P. Jose de Acosta de la Compania de Jesus, edited by P. Francisco Mateos. Biblioteca de Autores Espanoles (continuacion) 73, pp. 390-608. Madrid: Ediciones Atlas. . AIM, Javier 1972 Dinamica en la estructura inter-communitaria de Jesus de Machaca. America Indigena 32(2):773-816. Albornoz, Cristobal de 1984 [ca. 1582] Instruccion para descubrir todas las guacas del Pir(J y sus camayos y haziendas. In "Albornoz y el espacio ritual andino prehispanico", edited by Pierre Duvio]s. Revista Andina 2( 1):169222. Anders, Martha B. 1986 Dual Organization and Calendars Inferred from the Planned Site of Azangaro. Ph.D. dissertation, Cornell University. Ann Arbor, Michigan: University Microfilms. Archivo General de la Nacion, Lima (AGN) Rea] Hacienda, Tribunal de Cuentas, Composicion de tierras de indigenas: Leg. 5, 1643-17]7. "Visita y composici6n de tierras en Zurite, Abancay."

Bauer and Barrionuevo: Shrine Systems in Anta


Archivo Historico Departmental (AHD). Formally Archivo Historico del Cuzco (AHC) Cajas de Censos: Leg. 4, 1687-1697. "Sobre]as haciendas nombradas Quiachille y Ychubamba que posee e] maestro de campo Don Felipe Gutierrez de Toledo." Cajas de Censos: Leg. 20, 1802-1809. "Sobre las haciendas tierras, casas y alfalfar que tengo y poseo que estan en los altos de la fortaleza de esta ciudad nombradas LlaulIipata, Mascabamba, y Sacsahuaman y la Piedra Cansada y otros nombres y dos pares de tenerias que estan en la dicha Fortaleza y asi mismo ympongo sobre las haciendas de Queachili que estan en el valle de Ychubamba desta jurisdiccion." Arriaga, Pablo Joseph de 1968 [162J] Extirpaci6n de la idolatria del PirU [1621]. In Cronicas peruanas de interes indfgena, edited by Francisco Esteve Barba. Biblioteca de Autores Espanoles (continuacion) 209, pp. 191277. Madrid: Ediciones Atlas. Barthel, Thomas S. 1959 Ein FrUhlingsfest der Atacamenos. Zeitschrift jUr Ethnologie 84:25-45. Bauer, Brian S. 1992 Ritual Pathways of the Inca: An Analysis of the Collasuyu Ceques.in Cuzco. Latin American Antiquity 3(3): 183-205. Cabello de Balboa, Miguel 195J [J586] Miscelania Austral, parte 3: Historia del Peru. Lima: Instituto de Etnologfa. Callapina, Supno, y otros Quipucamayos 1974 [1542/1608] Relacion de la descendencia, gobierno y conquista de los Incas, edited by Juan Jose Vega. Lima: Ediciones de la Biblioteca Universitaria. Cieza de Leon, Pedro de 1976 [Part I, 1553 and Part II, 1554] The Incas of Pedro Cieza de Leon, translated by Harriet de Onis and edited by Victor W. von Hagen. Norman, Oklahoma: University of Oklahoma Press. Cobo, Bernabe 1956 [1653] Historia de] Nuevo Mundo. In Obras del P. Bernabe Cobo de la Compania de JesUs, edited by P. Francisco Mateos. Biblioteca de Autores Espanoles (continuacion) 91 and 92. Madrid: Ediciones Atlas. 1980 [1653] Relacion de las guacas del Cuzco. In "An account of the shrines of ancient Cuzco", translated and edited by John H. Rowe. Nawpa Pacha 17: 2-80. 1990 [1653] Inca Religion and Customs, translated and edited by Roland Hamilton. Austin, Texas: University of Texas Press. Cordoba Mexfa, Pedro de 1900 [ca. 1572] Instruction de 10 que ha de haeer el Licenciado Pedro Mexfa, Clerigo presbftero de la Compaflia de Jesus en la visita general que el muy Excmo. Senor Don Francisco de Toledo, Visorrey de estos reynos . .. Revista de Archivos y Bibliotecas Nacionales 3(4): 387-404.

ANDEAN PAST 5 (1998)


Duviols, Pierre (editor) 1984 Albornoz y el espacio ritual andino prehispanico. Revista Andina 2( 1):169-222. Hadingham, Evan 1987 Lines to the Mountain Gods: Nazca and the Mysteries of Peru. Norman, Oklahoma: University of Oklahoma Press. Hyslop, John 1985 Inkawasi: The New Cusco, Canete, Lunahuana, Peru. BAR International Series 234. Oxford: British Archaeological Reports. Metraux, Alfred 1935 Les Indiens Uro-Cipaya de Carangas. Journal de la Societe des Americanistes 27:325-415. Molina (el Cusquefio), Crist6bal de 1989 [ca 1575] Relaci6n de las fabulas i ritos de los Ingas . ., In Fabulas y mUos de los incas, edited by Henrique Urbano and Pierre Duviols, pp. 47134. Cr6nicas de America series. Madrid: Historia 16. Molina (el Almagrista), Crist6bal de See Segovia, Bartolome de Morris, Craig 1990 Arquitectura y estructura del espacio en Huanuco Pampa. Cuadernos 12:27-45, Instituto Nacional de Antropologfa, Buenos Aires. Morrison, Tony 1978 Pathways of the Gods: The Mystery of the Andes Lines. NewY ork: Harper'and Row. Niles, Susan A. 1987 Callachaca: Style and Status in an Inca Community. Iowa City; University ofIowa Press.
.

-82
rectly attributed to Crist6bal de Molina (el Almagrista)]. Sherbondy, Jeanette 1982 The Canal Systems of Hanan Cuzco. Ph.D. dissertation, University of Illinois, UrbanaChampaign. Ann Arbor, Michigan: University Microfilms. 1986 Los ceques: C6digo de canales en el Cusco incaico. Allpanchis Phuturinqa 27:39-73. 1987 Organizaci6n hidraulica y poder en el Cuzco. de los incas. Revista Espanola de Antropologfa Americana 17:117-153. Toledo, Francisco de 1924 [1572] Carta del Virrey D. Francisco de Toledo a S. M. sobre materias de gobierno, hacienda, guerra y eclesiasticos, en respuesta a cartas del Rey del afio anterior. In Gobernantes del Peru: Cartas y papeles, siglo XVI, Volume 4, pp. 380-403. Documentos del Archivo de Indias. Publicaci6n dirigida por D. Roberto Levillier. Madrid: Colecci6n de Publicaciones Hist6ricas de la Biblioteca del Congreso Argentino, Imprenta de Juan Pueyo. Urbano, Henrique and Pierre Duviols (editors) 1989 Relacion de las fabulas y ritos de los Incas. Cr6nicas de America series 48. Madrid: Historia 16. Villanueva Urteaga, H. 1982 Cuzco 1689: economia y sociedad en el sur andino. Cusco: Centro de Estudios Rurales Andinos, Bartolome de Las Casas. Zuidema, R. Tom 1964 The Ceque System of Cuzco: The Social Organization of the Capital of the Inca, translated by Eva M. Hooykaas. International Archives of Ethnography, supplement to Volume 50. Leiden: E. J. Brill. 1982 Bureaucracy and Systematic Knowledge in Andean Civilization. In The Inca and Aztec States, 1400-1800: Anthropology and History, edited by G. A. Collier, R. I. Rosaldo, and J. D. Wirth, pp. 419-458. New York: Academic Press. 1983 Hierarchy and Space in Incaic Social Organization. Ethnohistory 30(2):49-75.
_

Polo de Ondegardo, Juan

.-

1916a [1585] De los errores y supersticiones de los indios, sacados del tratado y averiguaci6n que hizo el Licenciado Polo, edited by Horacio H. Urteaga and Carlos Romero. Coleccion de Libros y Documentos Referentes a la Historia del Peru 3:3-43. Lima: Sanmartf. 1916b [1571] Relaci6n de los fundamentos acerca del notable daiio que resulta de no guardar a los Indios sus fueros, edited by Horacio H. Urteaga and Carlos Romero, Coleccion de Libros y Documentos Referentes a la Historia del Peru 3:45-189. Lima: Sanmartf. Rowe, John H. 1980 The Shrines of Ancient Cuzco. Nawpa Pacha 17:2-80. 1985 La constituci6n inca del CUzco. Historica 9(1):35-73. Santillan, Hernando de 1950 [1564] Relaci6n del origen, descendencia polftica y gobierno de los Incas . . ., edited by M. Jimenez de la Espada. In Tres relaciones de antigUedades peruanas. Asunci6n del Paraguay: Editora Guaranfa. Segovia, Bartolome de 1943 [1553] Relaci6n de las muchas cosas acaecidas en el Peru. .. In Las cronicas de los Mo/inas. Los Pequefios Grandes Libros de Historia Americana, Series I, Volume 4, first document, pp. 1-88. Lima: Librerfa e Imprenta D. Miranda. [Incor-

00 W I

48

Lake Pluray

(0 0)
Cortchanca

~ ;: ~ ~ ~ =i -. ~ ::s ~ ~ ..

o
II Cusco Ceque System Shrine
Figure 1. Possible distribution of ancient shrines in the Anta region.

Anta Shrine

~kml~

~ "'t ~. ~ ~ ~ a ~. ~ ::s ~

ANDEAN PAST 5 (1998)

-84

Figure 2. Runa Pacarisca, the origin place of Anta Ayllu.

85-

Bauer and Barrionuevo: Shrine Systems in Anta

Figure 3. The carved stone of Quilla Rumi.

ANDEAN PAST 5 (1998)

-86

Figure 4. The cave of Tocco Ccacca in the community of Ayllu Mayo.

87-

Bauer and Barrionuevo: Shrine Systems in Anta

Figure 5. The archaeological site of Tambocancha.

THE TEMPLE OF BLINDNESS: AN INVESTIGATION OF THE INCA SHRINE OF ANCOCAGUA Johan Reinhard The Mountain Institute, Franklin, West Virginia and The Field Museum o/Natural.History, Chicago
"In the district of Canas there was a temple which they called Ancocagua: there they made sacrifices according to their blindness." Pedro Cieza de Le6n 1

Ancocagua must be one of the most enigmatic Inca sites mentioned in early colonial documents. The renowned Spanish chronicler, Cieza de Leon (1977 [1554]:107), listed it as the fourth most important temple in the Inca empire. Yet, there is no description of the site, nor of its exact location, and this naturally gives 'rise to some b~ic questions. Where was it? Why was it so important? Given its significance, why did so few of the Spanish writers refer to it? The only way one could hope to answer these questions w~s by gathering together the historical references and by investigating !!Ie region in which the site might be found. Beginning with Cieza de Leon's (1977 [1554]:107, 153) account, we know that an ancient ("muy antiguo") oracle was highly venerated at Ancocagua, and that the temple was somewhere in the province of Hatun Canas.2 Aside from the Incas, people from all around came to worship at the temple. Animals and humans were sacrificed and gold was offered to the deity there. Cie:zade Leon (ibid: 107) heard that gold valued at 30,000 pesos was taken from the temple by the Spaniard Diego Rodriguez Elemosin. Even more treasure was found, and there were reports that gold and silver were buried by the Incas in places still undiscovered.

When I first began searching for clues to the location of Ancocagua in the historical accounts, I did not find its name directly associated with a temple. However, it does appear in historical records from the province of Canas. In a list of communities dating to 1575 one called Ancocaua3 is listed next to Coporaque and Yauri (in Hatun Canas), as it was in lists prepared in 1583, 1599, and 1812 (Glave 1987:64-66) (Table 1; Figure 1).4 Today, the towns called Coporaque and Yauri are 12 Ian apart and in the same wide valley, about 140 Ian. (in a straight line) to the southeast of Cusco in the province of Espinar. Because the communities noted are listed in geographical sequence, the settlemept of Ancocagua must not have been far from them. In addition, by 1581 the people of Ancocagua had been "reduced" (brought together) into a landholding unit that was a part of the community of Coporaque '(Aparicio 1982:96). It consisted of 275 people and 28 tribute-paying Indians, but the Indians would have presumably kept their landholdings outside Coporaque proper. The reason for the "reduction" is said to have been that the Ancocagua Indians poured melted gold down the throat of the Spaniard to whom they owed tribute. Those responsible ran off to other provinces while the remainder
3 "Ancocagua" is also spelled "Ancocaua", "Aconcagua", "Anccocahua", "Anccoccahua", and "Hancocagua" . 4 The orthography of Quechua tenns has varied through the years and depends upon the author. Thus, one can find such spellings as AncocagualAncocaua, MullucaguaIMulluccahua, Suyckutambo/Sucuitambo, etc. I have maintained the use oftenns that appeared to me to be the best lrnown ITomthe literature and have indicated alternative spellings of some words because they appear in various documents and on maps.

I My translation. See Cieza de Le6n (1984 [1553]: 223).

2 Hatun in Quechua means "large" or "principal" (GonzalezHolguin1989[1608]:158).

ANDEAN PAST 5 (1998):89-108.

ANDEAN PAST 5 (1998) settled in Coporaque (Celestino 1982 [1792]: 78). Whether this is fact or legend, in the late 1700s Ancocagua was listed as one of eight groups (ayllus) fonning Coporaque (Hinojosa 1987:232). It was described as having once had its own parish (curato) (i.e., prior to the incident with the Spaniard), but in 1792 it fonned part of Coporaque (Celestino 1982 [1792]:78). It is suggestive that when the lists follow a clear north-to-south progression throughout, as in those of 1575 and 1583 (Glave 1987:64, 66), Ancocagua occurs last. This indicates that Ancocagua originally bordered Coporaque to the south. Support for this hypothesis comes from an examination of the rivers noted in historical sources. In a document of 1586, the Apurimac River is stated to have its origin in a village called Ancocaua, which lay on the road from the city of Arequipa (i.e., to Cusco) (Fomee 1965 [1586]:28). Many tributaries to the south of Coporaque contribute to the origin of the Apurimac River, and the road between Arequipa and Cusco passes through this area (Agurto 1987:42;' Alicia Quirita, personal communication 1994). Thus, it is possible that the place called Ancocaua was the temple of that name.5

-90

We know today that the area south of Coporaque is the source of the Apurimac River. That this was also known by the Spaniards is further demonstrated by a list of rivers presented in 1792 by Celestino (1982:75). According to him, the Apurimac River has its origin in Vilafro Lake, which is 12 kIn west of the town of Cailloma (Caylloma) where the' list begins, and this is, in fact, close to its true origin. Only through the use of modem technology has the source of the Apurimac's furthest tributary been traced to the foot of the mountain Mismi ca. 30 kIn to the south of Cailloma (McIntyre 1991:20).6 The Aconcagua River was first in the list of Celestino, reportedly flowing into the Apurimac after it passed through a steep gorge below Cailloma. Of the rivers that follow, all those that I could

locateon the 1:100,000seriesmaps7 produced


by fern's Instituto Geognifico Nacional8 occur downstream from the AconcagualApurimac confluence (moving Tomsouth to north) in the same order as they do in the list. I could not find a river with the name Aconcagua on any map, but it is probably one of the rivers that originate near the town of Suyckutambo (Sucuitambo) or between it and the pass of Apacheta Rayada some eight km distant (Figure 1). The Safiu River, which follows the Aconcagua River in the list, flows into the Apurimac just below the Inca site of Mauccallacta, approximately 20 kIn south of Coporaque. The next river listed, the Quero, flows in near Coporaque. These rivers both have their origins in the lakes in the mountains above Cailloma near the town of Suyckutambo (4,800 m). This is at a divide where rivers flow in all directions. The majority,
6 The Incas were aware that Mismi Mountain, on which they made important ritual offerings (Ross 1980), straddles the continental divide, and that rivers start flowing to the Pacific Ocean &om its western slopes, and also &om only a few km to the west of Vila&o Lake. 7 These include the Cailloma 31-s, Velille 30-s, Condoroma 31-t, and Yauri 30-t sheets. 8 Formerly the Instituto Geognifico Militar.

5 According to a 1:100,000 map (Peru, Instituto Geografico Militar, Carta Nacional, Cailloma, Hoja 31s), there is a small settlement with the name of Anccocahua seven kIn to the south of the town of Cailloma (Caylloma), at the origin .of the Unculle River (Figure 1). This settlement is not situated on a high point, and, therefore, does not fit the description of the site of Ancocagua. No ruins of significance have been noted in the area (Alicia Quirita, personal communication 1994). In any event, this Anccocahua is not located in the province of Hatun Canas and thus is an unlikely candidate for the temple. However, it is possible that this settlement was named after the temple or the land and people associated with it. There is a mountain called Anccoccahua (see the IGM 1:100,000 map Callalli, Hoja 32-t), but it is 100 kIn to the southeast of Coporaque in the province of Collagua. Thus it is too far away to have been the site of the temple, even if ruins exist on the mountain.

91however, eventually sweep around to merge with the Apurimac. Although we do not know which place the Incas considered to be the exact source of the Apurimac, it seems reasonable to assume they would have perceived it to be in the region south of Coporaque.

Reinhard: Inca Shrine at Ancocagua

fied temple. After some time the rebels ran out of water and were about to surrender. However, it snowed heavily one night, and thus they were able to continue their resistance (ibid.:293).

The Spaniards asked members of the Inca Aside from Inca ruins leading south from nobility who had accompanied them how the Coporaque along the Apurimac River as far as Incas had captured Ancocagua when they batthe archaeological site of Maria Fortaleza tled the people living there some years before. (Quirita and Candia 1994), there is evidence The Spaniards were told that the Incas filled in suggesting an Inca presence higher up the the breach between the two high points using Apurimac Valley. One part of the name Suy- rocks, bushes, and bunches of wild grass. ckutambo, (tambo means "way station" in Employing the same method, the Spaniards Quechua) could indicate an Inca presence at captured the fortified temple (Betanzos 1987 this place, and the name apacheta (Quechua - [1551]:293-294).9 for "cairn") suggests that traditional worship was performed on the pass of Apacheta Rayada. Inca potsherds have also been reported 9 Hemming (1970:185-186, 572 footnote 186) synthefrom the Suyckutambo area (Alicia Quirita, sized material provided by some chroniclers about a battle at a fortified, rocky outcrop called Ancocagua personal communication 1994). More re- located in "bad, humid country". Monica Barnes (persearch needs to be done to establish the Inca sonal communication 1995) kindly provided informapresence in this little-known.area. tion from three sources cited by Hemming: Herrera y One other tributary of the Apurimac, the Callomani (also sp~lled Cayamani) River, also begins near Suyckutambo and meets the Apurimac just above Mauccallacta and immediately before Maria Fortaleza. Because it is the first .significant tributary downstream from the gorge, it might be the one that was called Aconcagua in 1792. When placed beside the information about the community (ayllu) of Ancocagua noted above, the evidence pointed to the temple of Ancocagua as having been located to the south of the town of Coporaque. However, this still left the problem of what the temple looked like, because there are several Inca sites in the area. The discovery of the missing part of the 1551 book by Betanzos (1987 [1551]) has meant we now have important information concerning the physical description of Anco~ cagua. He noted that the temple was on a fortified spur directly opposite a hilltop. Based on events he described that occurred in Cusco, we know that a battle took place at Ancocagua in late 1535. There had been an uprising in the area and local inhabitants had killed a Spaniard. Juan Pizarro led a group of Spaniards and native allies to lay siege to the fortiTordesillas 1615; Pizarro y Orellana 1639; and the "Probanza de servicios . . ." of 1538, the latter published in 1940 in a collection of documents edited by Padre Victor M. Barriga (hereafter referred to as Archivo General de Indias 1940). Pizarro y Orellana relies heavily on Herrera y Tordesillas, which is the most complete account. Only one of the men involved in the battle, Diego de Narvaez, mentioned that the land was humid and bad ("tierra hUmeda y mala") (Archivo General de Indias 1940:49). However, he did not state that it was hot, and it is even unclear if the country described was at Ancocagua or that crossed while he returned to Cusco. This takes on significance when the two other participants, Tomas Vasquez and the priest Pedro de Varco [sic], described suffering from the cold during the campaign (ibid :45-46). The highland region of Hatun Canas was wellknown for its cold climate (Cieza de Le6n 1984 [1553]:224), and we have seen that Betanzos described it snowing at Ancocagua. Given that the battle must have taken place in the South American summer months of late 1535, it makes sense that the wet season would have begun, with snow at higher elevations and rain lower in the valleys. This would make Diego de Narvaez's discussion of humidity understandable. According to the accounts of Herrera y Tordesillas (1615:231-234) and Pizarro y Orellana (1639:199-200), Indians killed Pedro MArtirde Moguer, while Diego de Narvaez (Archivo General de Indias 1940:49) stated that it was Pedro Martin Dominguez. Spaniards under

ANDEAN PAST 5 (1998) Given' the above infonnation, it was now possible not only to narrow down the location of the temple in Hatun Canas, but also the number of sites fitting its description. Unfortunately, the reported presence of Shining Path terrorists in the region meant that I delayed several years in beginning my search for Ancocagua. However, in 1994 I had the good fortune to meet Alicia Quirita, one of two archaeologists who had recently conducted investigations in the area (see Quirita and Canthe leadership of Gonzalo Pizarro attacked the rocky outcrop in 1535. Later Juan Pizarro arrived with reinforcements. They attacked using a siege blanket, but were unsuccessful. Finally a member of the Inca nobility came nom Cusco to help. He had four Spaniards disguise themselves as Indians, shave their beards, and dye. their faces. During the night, he and the four Spaniards (Mancio Sierra, Francisco -de Villafuerte, Pedro de Barco, and luan Flores) were admitted within the first and second entrances of the walls. The five men were able to fight off the Indians long enough for the rest of the Spanish forces to 'enter and capture the fort. Many of the natives leaped to their deaths nom the cliffs. The SpaniaJ'dscaptured 5,000 gold castellanos and used these to help-build the church in Cusco. Although the exact location is not noted in these sources, and they do not mention Ancocagua as having been a temple, these accounts appear to describe the capture of the fortified "temple" of the same name. This is made more likely when we remember that both the battle described by these writers and the one recounted by Betanzos occurred at the time of the uprising of the Collao Indians in late 1535. Both took place on a rocky outcrop with steep cliffs, and both were undertaken due to the Indians having killed an encomendero. Both also note the cold, both involved a Pizarro brother, and both places were named Anco.cagua. Betanzo~ wrote about the battle for Ancocagua a few years after it occurred, and his account does not mention Ancocagua as having been captured by deployment of disguised men. However, this may not have been brought to his attention, and in any event it does not contradict his description of the use of materials to "fill" the space between the hill and outcrop. Indeed, the siege blanket may have been used at this time or it may have been part of a combination of tactics involving the entry through disguise and an assault by means of reaching the upper area of the fort by the method described above.

-92

dia 1994). Lacking the infonnation provided in Betanzos' more c.omplete account, Quirita and Candia (1994:23) hypothesized that the temple of Ancocagua may have been on the summit of the mountain Quimsachata near Coporaque. In August 1994 Alicia Quirita and I were able to examine sites in the region, combining her archaeological survey and my own historical research. . There are numerous ruins in the area. However, few have been described in the literature, and we initially could not find any that fit Betanzos' description of Ancocagua. For example, the important Inca sites of Kanamarca (K'anamarka) and Mauccallacta (Maukallakta) are not on high points (Angles 1988, vol. 2:571-601; Bonnett 1983).10 One of the most significant sites we visited is that of Mullucagua (also spelled Mulloccahua and Muyuqhawa) located on a hilltop at about 4,100 m in elevation and about 12 km to the east of Yauri (Angles 1988, vol. 2:593, 597; Pardo 1948:15.;20). The hilltop is a truncated cone dominating the surrounding area (Pardo 1948: photo 7 between pages 18 and 19). The site is constructed with points of entry protected by bastions, contains funerary towers, and has some fine Inca stonework (Angles 1988, vol. 2:577, 582, 584, 593; Pardo 1948:15-20). It was reportedly one of the fortified places used by the Canas people prior to Inca presence in the region (Celestino 1982:75). However, it did not fit Betanzos' description, nor the location as detennined from the list of communities and rivers described above. Furthennore, the name of Ancocagua was unknown to the local inhabitants we questioned. Ijowever, as soon as we went to the south of Coporaque, we encountered people familiar with the name Ancocagua. It was not used for a particular site, but rather to denote an area, Ancocagua Manturca. This is a plateau located about 10 km south of Coporaque. We were told that it is bounded on the west by the
10 It should be noted that even the bottoms of the valleys in this area are above 3,800 m in elevation.

93Apurimac River and on the east and south by mountains, among which (as labeled on the 1:100,000 .JGM map) were Huicho Pucara, Vila Pucara, Chita, Huayna Condori, Sayuta, and Checorume. Ancocagua Manturca consists of several small settlements in an area of more than fifty square km (see the 1:100,000 map Yauri Hoja 30-t). The altitude of the plateau is around 4,000 m with some houses as high as 4,400 m. Used for grazing animals, this is also the first area (aside from small patches of land along the rivers) where agriculture (primarily of potatoes, but also quinoa) can be carried out to the north of the continental divide and the source of the Apurimac River. It is, therefore, also the first area where a number of permanent settlements can be maintained.
.

Reinhard: Inca Shrine at Ancocagua the outcrop and river bottom are both strewn with boulders and certainly water could be said to trickle amidst the boulders at the river and even on the outcrop when it rains (Figure 2). The physical dimensions of the outcrop are impressive. Quirita and Candia (1994:87) estimate the top area to be about five hectares~ i.e., some 50,000 m2, on which hundreds of ruins exist (Figure 3). According to my altimeter, the summit is at 4,170 m, which is 134 m above the river. Based on Quirita's and Candia's (1994) investigation of the ruins and ceramics, the site became particularly important during the Middle Horizon (ca. A.D. 540 to A.D. 900) when it was occupied by the Huari (Wari)

We could findonly one archaeologicalsite

people.11 It continuedto play a key role in the


region through the Late Intermediate Period (called Collao in this area; ca. A.D. 900 to A.D. 1476) and the Late Horizon (i.e., the Inca Period; ca. A.D. 1476. to A.D. 1534 in this region) (ibid.). There are also ceramics dating from the Early Horizon (ca. 1400 B.C. to 400 B.C.) through the Early Intermediate Period (ca. 400 B.C. to A.D. 540). However, it is not clear just how important a role the site played before the Middle Horizon. Nonetheless, the ceramics indicate that the site may have been occupied for around 3,000 years. One of the best preserved structures is at the western foot of the outcrop about 15 m above the Totorani River (Quirita and Candia 1994:99-102) (Figure 4). It is a large (over 16 m long) rectangular building of Inca origin, as demonstrated by its trapezoidal doors and niches and typical Inca construction tech11 I have only provided a summary of the fmdings made at Maria Fortaleza. Further details on the archaeology of the site can be found in Quirita and Candia (1994) and in a forthcoming publication based on their thesis. They were the first to survey Maria Fortaleza in 1991-92, and they conducted seven I x 1 m excavations: two in Sector A, two in Sector B, two in Sector C, and one in Sector D (Quirita and Candia 1994:104). It must be emphasized, therefore, that archaeological work at the site has just begun.

in this region which fit Betanzos' description, and it. turned out to be the most important: Maria Fortaleza. It would seem to be no coincidence that it is the only major archaeological complex in the region to have a Spanish name. This indicates that the Spaniards may have wanted to suppress the indigenous term for the site. The Spanish toponym is itself suggestive: fortaleza means "fort" and Maria is, of course, the Virgin Mary, implying that the site had a religious element. As noted above, Ancocagua was a fortified temple. According to local tradition, the name dates to the time long ago when a woman yelled out "Jesus, Maria" after being surprised while she was urinating. The Virgin Mary instantly appeared and was converted into a rock, while the urine of the woman became a river. The rock is near the confluence of the Totorani and Apurimac Rivers just below the ruins (Quirita and Candia 1994:44, 88) (Figure 2). This legend could have its roots in an older one, as discussed below. The local name of the site is now T'acrachullo (Tajra Chullo on the 1:100,000 map). According to local inhabitants, this comes from the words t'acra meaning "disarranged rocks" and chul/o meaning "trickling water" (Quirita and Candia 1994:88). The name would seem appropriate for the place, because

ANDEAN PAST 5 (1998) niques (el, Agurto 1987; Gasparini and Margolies 1980) (Figure 5). The majority of the ceramics found in and near the building (located in Sector D as denoted by Quirita and Candia 1994) were of Inca or Collao origin. Four trapezoidal windows are in the front wall and four trapezoidal niches are in the back one. A cleared area (delineated by a retaining wall) is in nont of the rectangular structure, while several circular structures exist behind it. Of interest is that the central doorway is directly in front of a door in the back of the building. The impression is that people may have had to pass through the structure before ascending to the summit of the outcrop. A trail leads from this building to the highest point of the notch between the outcrop and the hill opposite it to the south, and from there on to the summit. At this- point a trail from the Apurimac River converges with it (Quirita and Candia 1994:88). Remains of walls line the trail above tl}.enotch, and one stone gateway still exists about halfway to the summit. The trail reaches an area on the summit with large Inca structures (as in Sector A) and passes near a number of funerary towers (ehullpas) (as in Sector C) (Figure 3). Sector A includes a large enclosed area (eaneha) (ca. 80 x 40 m) with an Inca structure on its western side (Figure 3). According to Alicia Quirita (personal communication 1994), the Inca structure was built on the base of a Huari one, and the ceramics encountered during an excavation in its interior support this conclusion (see Quirita and Candia 1994:109, where more than half the ceramics are Huari). Quirita and Candia (1994:89-90) hypothesized that the eaneha may have been a place where ceremonies were conducted. It is at the center and highest point of the summit area. Huari ceremonial pottery was found in and near it; and the natural bedrock in the cancha contains four circular holes and one rectangular one (ca. 1.5 m in diameter) that could have played a role in rituals. Sector B has structures and ceramics which indicate that it is of Huari origin (albeit with some Inca occupation). Sector C con-

-94
tains several funerary towers (Figure 6) and tombs, mostly pertaining to the Late Intermediate Period (Quirita and Candia 1994:97, 167, 197). Over 200 circular structures exist on the summit, and these also primarily date to the Late Intermediate Period (ibid.:92). Little archaeological work has yet been undertaken at Maria Fortaleza, and Huari ceremonial structures have not yet been identified. However, a religious component at the site during the Middle Horizon is suggested by the presence of fine Huari ceramics (John Rowe, personal communication 1994). Maria Fortaleza was probably much more than a religious center. It is at the farthest limit of Huari expansion to the east (Chavez 1988),I and it likely would have played a role as a control point for trade between the coast and highlands during both the Middle and Late Horizons (Quirita and Candia 1994:196, 201). Taken together, th.e ruins indicate that many people lived at least part of the time on the summit of the outcrop of Maria Fortaleza. The site was occupied from at least the Middle Horizon to Inca times. Religious activi!ies occurred there. What additional evidence exists to support Maria Fortaleza's identification as the Inca temple of Ancocagua? Returning to the information supplied by Betanzos, we need to see if Maria Fortaleza fits his description of the site. The narrowest part of the notch between the outcrop and the hill facing it to the south is only about 14 m wide. The notch naturally begins to wideJ1. as it rises higher. Once level with the summit of the outcrop, it is so wide that it would have been difficult, if not impossible, for the Spaniards to fill it completely, as described by Betanzos. However, Betanzos was not an eyewitness, and if we assume that the Spaniards built up a rampart to reach the less steep sections about 20 m above the notch, then the description makes sense. It is about 73 m from the notch to the highest point of the summit, but the steepest area occurs in the first third of the ascent. It should be added that on all the other sides of the outcrop the drop is vertical for some 60 m or more.

95In any event, Betanzos noted that Ancocagua was built on an outcrop close to a hill equal to it in height, and this fits well with the site of Maria Fortaleza. The outcrop of Ancocagua had no water source on its summit, because Betanzos described that the Spaniards nearly forced the natives to surrender due to the lack of water. This, too, fits the outcrop of Maria Fortaleza. Snow could have easily fallen on the site in the latter part of the year, as it did in mid-September 1991 (Alicia Quirita, personal communication 1994). Maria Fortaleza is in an area called Ancocagua and it is in what was Hatun Canas. In addition to our own unsuccessful attempts to locate another archaeological complex which might be the fortified temple, we could not find anyone, either a local inhabitant or outsider, who has visited the region and who knew of any other important site in the area that fits Betanzos' description of Ancocagua. Thus, taken together, there are several reasons. for identifying Maria Fortaleza as the site of the fortified temple of Ancocagua. It might be of interest to turn now to 'ethnographic and historical information about religious beliefs in the region of Maria Fortaleza. In 1792 it was reported that a mountain above Coporaque was thought to be very powerful and that it contained an enchanted lake (Celestino 1982:77). Much the same belief has been noted in recent times about a magical mountain with a lake on its summit from which treasure is said to have been taken (Hinojosa 1987:230). This mountain is almost certainly the one called Quimsachata (4,759 m), dominating the town of Coporaque. We found that basically the same legends about Quimsachata and its lake exist to the present day. Although no major ruins were seen by a local man who had been to its summit, he did say that people traditionally made offerings there during Carnival (in February) to ask for an increase in livestock. According to Quirita and Candia (1994:23), Quimsachata isconsidered a tutelary deity (apu) of the entire province of Espinar. It stands out prominently to the north when viewed from Maria Fortaleza. Not far from Maria Fortaleza (ca. 30 km to the northwest of Coporaque) is the mountain

Reinhard: Inca Shrine at Ancocagua called Huaylla Apacheta. It is believed today to influence livestock fertility and to have a lake on its western slopes where "many cows graze" in its depths (Roel 1966:28). This is a belief which conforms to one found throughout the Andes, including during the Inca horizon. It is thought that domesticated livestock originated out of lakes and that mountains were their ultimate owners (Arriaga 1968 [1621]:115; Duviols 1974-76:283; Flores 1988:249-250; Gowand Gow 1975:142, 148; Isbell 1978:59). It also accords with beliefs about mountains throughout the highlands. There pastoralism plays an important role, as has been the case in the province of Canas since at least the Inca Period (Cieza de Leon 1984 [1553]:224), and at Coporaque for centuries (see Hinojosa 1987:230 for a report dating to 1689). Ritual battles still take place ca. 40 km to the north of Coporaque. These are believed to augment crop fertility, and offerings, along with any human blood that might be shed in the battles, are thought to go to Pachamama (Mother Earth) and the mountain deities (Barrionuevo 1971:79,82; Gorbak et al. 1962:278, 287, 290). As we have seen, there is also land suitable for the cultivation of potatoes and, to a lesser extent, other native crops such as quinoa, on the plateau of Ancocagua Manturca and along the river valleys. Mountains not far distant, such as Ausangate, continue to be worshiped for agricultural fertility (Gow and Condori 1982:40-43; Reinhard 1991:81). Thus, agricultural and livestock fertility were principal reasons for the establishment of shrines on the summits of mountains throughout much of the Andes (Reinhard 1985). River confluences were often considered sacred by the Incas (Betanzos 1987 [1551]:72; Murua 1946 [1590]:312). They were used as places to wash and be cured of illnesses (ibid. 1946:312) and to perform rituals relating to the dead (Guaman Poma 1980 [1615]:272). In the Cusco region, they are still viewed as being powerful and dangerous places (Allen 1988:205). Maria Fortaleza was built on an outcrop which overlooks the confluence of

ANDEAN PAST 5 (1998) two major rivers at a critical ecological point, and this surely would have added to its sacredness. There is little ethnographic information available on Maria Fortaleza itself, but what does exist is significant. In local tradition, selected elements of the physical terrain are perceived in terms of a body metaphor. Maria Fortaleza is believed to be at the center of the body, i.e., the navel. The head is the Callomani River, the right arm is formed by the Llaska River, and the left by the Totorani River. The right foot is the archaeological site of Inti Pucara and the left foot is composed of the ruins of T'eraqara Pucara, located on the east and west sides of the Apurimac River, respectively (Quirita and Candia 1994:42-43). Thus its head is to the south (the source of the Apurimac River) and lower members are to the north (the direction of the .flow of the Apurimac).

-96 Valley to the south (Valderrama and Escalante 1988:206). Today, Maria Fortaleza is still perceived to be a female, and we were told that offerings are made on its summit in August, and during Carnival, for the fertility of crops and livestock. Offerings are also made to Pachamama (Mother Earth) and to the surrounding moun-. tain deities (e.g., Quimsachata, Choquepirhua, Machula, and Laramini) for the same reason.

Every day, coca leaves are offered to the


mountain deities so that they will give the people strength during the day (Quirita and Candia 1994:46-47). It is common throughout the central Andes to make' such offerings to deities associated with the land (Allen 1988). The meaning of the name Ancocagua may .also provide some clues as to site. The word appears to be of Aymara origin, the language spoken in this region at the time of the Spanish conquest (Bertonio 1984 [1612], vol. I:A2). In Aymara anco (hanco) means "white" and cahua (i.e.: cagua , the "g" and "h" being used interchangeably in the Spanish writings of the time) means "the last or hindermost of a village" (ibid., vol. 2:32, 118). The use of cagua makes sense when one considers the location of Maria Fortaleza. It is indeed at the hindermost part of the area called Ancocagua Manturca.12 According to Girault
12 The chroniclers consistently spelled Ancocagua with a single "c" and not "cc" when referring to the temple fortress. The double "cc" is distinctive in both Quechua and Aymara and indicates forceful pronunciation (el, Betanzos 1987 [1551], vol. 2:32,41; Gonzalez Holguin 1989 [1608]:9-10). Betanzos, in particular, was fluent in Quechua and would presumably have caught the difference in pronunciation. Thus it seems unlikely that "ccagua" (the Aymara word for "tunic,"; Adelson and Tracht 1983:50) was the word intended, despite its apparent appropriateness in the context of Ancocagua, i.e. as being like a "white tunic." (This might, however, explain the spelling of the mountain Anccoccahua discussed in note 5 above.) I initially searched for possible meanings of Ancocagua in Quechua, but could not find spellings and meanings that I felt were reasonable alternatives to the Aymara etymology presented here. There is the possibility that because in Quechua "anca" means "eagle"

The use of a .body metaphoris known for


other regions of the Andes (Bastien 1978; Valderrama and Escalante 1988:206) and is clearly a way of providing meaning to significant geographical and. cultural features found in an area. In this case the focus is on rivers and ruins which are united through the metaphor. Based on the long tradition of the body metaphor in Andean beliefs (Bastien 1978; Salomon and Urioste 1991 [ca. 1600]), Maria Fortaleza's role as a symbolic center could date back at least to the Inca period. Indeed, Cusco itself was perceived by the Incas to be the "navel" of the world (Garcilaso 1966 [1609]:93). This metaphor suggests one role that the deity of Ancocagua may have played. We saw above how the name of Maria Fortaleza is believed to have come from the Virgin Mary, who was converted to stone. The urine of the woman who called out her name was the source of a river, i.e., the Apurimac River after its confluence with the Totorani River at the foot of the outcrop. A symbolic link between urine, water courses, and fertility has been noted in other regions of the Andes, including Chumbivilcas not far to the west (Roel 1966:25-26) and the Colca

97(1988:140), anco also has the meaning of "fertile," and this appears to have been the case during the Inca period (Bertonio 1984 [1612], vol. 2:118). Use of the term anco becomes even more understandable when one sees the white coloring of the rocks which make up the outcrop of Maria Fortaleza. Based on the use of the name Maria and the beliefs noted above, it would seem reasonable to assume that the deity of Ancocagua was perceived to be female and that it was closely associated with the Apurimac river. Maria Fortaleza's strategic situation relative to trade between the higWands and coast may have also carried with it religious connotations. Deities of the hills and mountains were widely invoked for success in trade, both during the Inca period and in modern times (Reinhard 1990:167). In addition, as we have seen, the summit complex of Maria Fortaleza is within view of both a river confluence and sacred mountains, including snow-capped peaks over 100 km away,. which we know were especially sacred to the Incas, and stirely long before them (Reinhard 1990, 1991, 1995, in prep.). Whateve~ other roles it played, Maria Fortaleza was likely used as a place of worship for obtaining the fertility of crops and animals. and success in trade. Its ecological
and cagua could be a variant of "qhaway," "to look," Ancocagua could mean something like "eagle lookout" (Monica Barnes, personal communication 1994; cJ, also Angles 1988, vol. 2:593 for his use of"qhawa" for "cagua"). However, I doubt this was the case, since the "qh" is like the "cc" noted above, and the "a" and "0" were not switched in any of the spellings of the name by the chroniclers. These vowels are distinguishable as phonemes in Quechua (as opposed to "u" and "0" which are allophones of the same phoneme) (Gonzalez Holguin 1989:9-10; Salomon and Urioste 1991:79, note 307). Furthermore, the name, if Quechua, would lack a nominalizer. Many of the names of the huacas of the Cusco ceque system are also best read in Aymara (Beyersdorff in press), supporting my interpretation of Aconcagua as an Aymara toponym. Given the Aymara presence in Hatun Canas and the common usage and spelling of both words forming the name Ancocagua in that language, I believe that it is not necessary to search elsewhere for the etymology of the term.

Reinhard: Inca Shrine at Ancocagua setting in an area suited for pastoralism and at the beginning of the agricultural zone would have reinforced the importance of such worship. These factors all would help explain Maria Fortaleza's regional significance, but it still leaves open the question of why the Incas considered the site to be so important.13 Indeed~ Cieza listed it as the fourth most important temple in the Inca empire. He would appear to have been influenced in his list by someone with a vested interest in Ancocagua, for surely ceremonial centers such as the Island of the Sun and Pachacamac, to name only two, could, with more reason, have been on it. But this does not detract from Ancocagua having been held in very high regard among the Incas. So why was this the case, when any number of other temples in the empire could have been selected? With Ancocagua there is support for a conceptual link between the temple and the origin of the Apurimac River. As we have seen, the source of the Apurimac River is located to the south of Coporaque, near the town of Cailloma (Figure 1). The Apurimac was a river higWy venerated by the Incas (Garcilaso 1966 [1609]:527). A temple was built on its banks near a steep gorge downriver from the town of Curahuasi, just prior to its descent into the tropical lowlands. The main idol was female, and the deity Apurimac spoke through this idol (Cobo 1990 [1653]:108). Indeed, during Manco Capac's rebellion against the Spanish, the deity spoke to the Inca emperor in the presence of a Spaniard, Francisco Martin, who was being held as a. prisoner. The temple's guardian was a woman of the
13 It was reported by Celestino that silver mines were worked in ancient times in one of the two mountains near Suyckutambo, and many other mines were reported in the area (Celestino 1982 [1792]:78). The Incas worshiped mountains for their minerals (Cobo 1990 [1653]:45). However, mines were located in many parts of this region, and it would seem unlikely that this was a major reason for the Incas to have considered the temple of Ancocagua to be of such importance.

ANDEAN PAST 5 (1998) Inca lineage, and it was one of the most famous of the Inca sanctuaries (ibid:l08). The river's name meant "the speaking captain or leader," and it was also called Capac Mayu, a name indicating that it was a major river of the Inca realm (Garcilaso 1966 [1609]:527). The use of the phrase "muy antiguo" by Cieza de Leon to denote the temple of Ancocagua, the statement of Betanzos that the Incas had to defeat local people at Ancocagua, and the archaeological remains, including fine ceremonial pottery dating to the Middle Horizon, all point to Ancocagua as a place of worship for the people of the area prior to the arrival of the Incas. The Incas may have added to its importance by relating it conceptually to the source of the Apurimac River. It is suggestive that the temple of ViIcanota, noted as the third most important temple in the Inca empire by Cieza de Leon (1977 [1554]:106), was located ~t the pass of La Raya, the source of the ViIcanota River and at the divide of two major river systems (Reinhard 1991:38, 1995). Cusco, the center of the Inca empire, lies between the Apurimac and ViIcanota river systems (Figure 1), and we know that the Incas paid considerable attention to hydrology and attributed ritual importance to river sources (Sherbondy 1982).14 Rivers were also fertility sources which partook of the powers and sacred characters of the mountains at their sources (Reinhard 1991:37-38). Maria Fortaleza is situated on an outcrop overlooking the first wide valley which the Apurimac entered that was both of agricultural and pastoral significance. Along with the fact that many minor tributaries contribute to the origin of the Apurimac River, it is possible that the place called Ancocaua, noted in 1586 as being at its source (Fornee 1965 [1586]:28), was the temple of that name. The Incas may
14 We know the significance that dualism had in Inca thought, and the Incas may have also viewed the temple of Ancocagua as fonning a logical counterpart to the

-98
have perceived that it was where the Apurimac River truly began, and thus that it was located at the conceptual, if not the precise, origin point. Important rivers converge at or near Maria Fortaleza, and this would also explain why the name Aconcagua was applied to the first river noted as one of its tributaries in the late 1700s (Celestino 1982 [1792]:75). Inca sites in and near the valley below Maria For-' taleza indicate that the area was of significance to the Incas. Assuming that Ancocagua was already a place of worship for the indigenous peoples, the Incas would have gained prestige and economic/political control by building up the site and making substantial contributions to it. This is a tactic that they used in several other areas (Albornoz 1984 [ca. 1582]; Salomon and Urioste 1991 [ca. 1600]; Reinhard 1985). In summary, I have used the following evidence to identify Maria Fortaleza as Aconcagua: 1. It is located in the province called Hatun Canas at the time of the Spanish conquest, and it is in the precise area which can be deduced from the list of communities and rivers noted in the historical sources. 2. Its physical description matches that provided by the chroniclers in that it is situated on an outcrop with steep cliffs, and opposite a hilltop at nearly the same elevation. It is fortified and virtually inaccessible. 3..It contains some of the most important Inca and pre-Inca archaeological remains in the area, including artifacts suggesting a religious significance for the site. 4. The climate fits the description of the chroniclers, and the site is in an area where brush and branches could be obtained (as described for the siege); such vegetation was not available at higher elevations. 5. The name Ancocagua is used to denote the area in which the site is located. 6. The site is the only one that has a Spanish name, a name which in turn indicates both the site's fortress aspect and suggests a religious role. 7. Local tradition points to the religious significance of the site and its location. 8. Its location, in terms of both sacred and physical geography, provides a reasonable explanation for why the Incas considered it to be of such religious importance. 9. The meaning of the name of Ancocagua is appropriate for the site. 10. No other

temple ofVilcanota.

99known site in the area meets the description of the chroniclers. Given all the evidence, it seems likely that Maria Fortaleza is Ancocagua. This still leaves us, however, with the question of why Ancocagua was noted by few chroniclers, yet was described as being so important by Cieza de Leon.15 Writing in the mid-1500s, both Cieza de Leon and Betanzos were among the earliest chroniclers of Inca religion following the Spanish conquest of 1532. Cieza de Leon (1977 [1554]:107) noted that Ancocagua had been looted only three years after the Spaniards won the battle for Cusco. By the late 1500s, when several accounts of Inca religion were being written, it would have been a distant memory. The site at Maria Fortaleza had been abandoned after the Spanish conquest (no colonial remains were found on the. summit), and the Spanish presence in the area was maintained instead in settlements built in the valley below the site. By all appearances, they simply renamed the site, placed a cross on it (according to local lore), and presumably restricted access to it, because there is little evidence of a later occupation. Currently, only local pastoralists take their sheep to graze on the summit. It was a common practice for the Spaniards to build a chapel or church on an important sacred site, but they did build a church at the site of Mauccallacta not far downriver from Ancocagua, and this would have served the Spanish settlers and native peoples (Angles 1988, vol. 2:598, 601; Bonnett 1983:56). Ancocagua apparently played no enduring role in Inca mythology, because it was not especially
15 The chronicler Vasquez de Espinosa (1948 [1617]:558) stated that the temple of Viracocha at Cacha (Rajchi) was called Ancocagua. His account was written in 1617, much later than those of Cieza de Le6n and Betanzos, and he clearly confused Ancocagua with Rajchi (Cieza de Le6n 1984 [1553]:223). An idol of a deity called Ancocagua (written Hanco Caua in the original text) was confiscated by Catholic priests in the late 1500s far to the west of Coporaque (MilIones 1990:274), but it is unclear if this was meant to represent the deity worshiped at the temple of Ancocagua.

Reinhard: Inca Shrine at Ancocagua noted by the Incas interviewed by the Spaniards after the mid-1500s. Ancocagua was not an exception in this regard, because Vilcanota and Coropunal6 (noted as being third and fifth in Cieza dC? Leon's list of the five most important Inca temples) were also not described in any detail in later chronicles (Reinhard 1995, in prep.). Although Vilcanota was mentioned occasionally, this seems to have been due to its having been situated on a major route used by the Spaniards, and to Inca priests having annually made pilgrimages to it from Cusco (Molina 1959 [1575]:38). Being situated in relatively isolated areas and having their exact locations unknown, both Ancocagua and Coropuna received little attention from historians and archaeologists. Although this article helps to clarify some issues relating to Ancocagua, while admittedly raising others, the temple's meaning has yet to be well established. Nonetheless, this examination of the evidence relating to Ancocagua serves to focus attention on a site of great importance to the Incas, and it helps further our understanding of Inca beliefs in one of the lesser known regions of their empire. 17

16 Near the Mountain Coropuna in southwestern Peru (Reinhard in prep.). 17 The temple of Ancocagua may also contribute to our understanding of an Inca ceremonial site on the mountain of Aconcagua over a thousand miles distant (Schobinger 1991, 1995). At 6,960 m, it is the highest mountain in the Andes and is located in the southern tip of the Inca empire in west-central Argentina. As we saw in note 3 above, Aconcagua was an alternative spelling for Ancocagua in the accounts of both Cieza de Le6n and Betanzos. We know that the Incas sent colonists to populate areas in distant parts of their empire and also that these colonists reestablished their traditional deities in the new areas (Albornoz 1984 [ca. 1582]:198). Thus the possibility exists that colonists from the region of Ancocagua could have been responsible for both the name of the mountain and the ritual sacrifices performed high on its slopes during the Inca period.

ANDEAN PAST 5 (1998) Acknowledgements The writing of this article was made possible thanks to a fellowship (No. ROM22297-91) from the National Endowment for the Humanities. My research on sacred geography in the Cusco region has involved numerous visits undertaken since 1981. The organizations that supplied the principal financing for this research are Rolex Montres, the Organization of American States, the National Geographic Society, and the Social Science Research Council, with additional support provided by Joseph and Sharon Richardson, and Bob and NancyMerritt. Research in the Department of Cusco was greatly facilitated by the National Institute of Culture and its directors Gustavo Manrique, Oscar NUfiez del Prado, Danilo Pallardel, and Femin Dias. For the past few years I have been a Senior Research Fellow of The Mountain Institute, Franklin, West Virginia and a Research Associate of the Field Museum of Natural History, .Chicago. I would like to express my gratitude to all of these people and institutions for their support. References Cited .
Adelson, Laurie and Arthur Tracht 1983 Aymara Weavings. Washington, DC: Smithsonian Institution Press. Agurto, Santiago 1987 Estudios acerca de la construccion, arquitectura, y planeamiento incas. Lima: Camara Peruana de la Construcci6n. Albornoz, Crist6bal de 1984 [ca. 1582) Instrucci6n para descubrir todas las guacas del PirU y sus camayos y haziendas. In "Albornoz y el espacio ritual andino prehispanico", edited by Pierre Duviols. Revista Andina 2(1):169222. Allen, Catherine 1988 The Hold Life Has. Washington, DC: Smithsonian Institution Press. Angles Vargas, Victor 1988 Historia del Cusco Incaico. 3 volumes. Lima: V. Angles. Aparicio, Manuel 1982 Historia de la Provincia de Canchis. In Kanchi: La Provincia de Canchis a traves de su historia, edited by Vicente Guerra, pp. 85-121. Lima: Editora Humboldt. Archivo General de Indias 1940 Probanza de servicios del Presbitero Rodrigo Bravo para presentar a S.M. y al Consejo de Indias, Cusco, 23 de diciembre 1538. Legajo 93, No.8, Rama 4 [old numeration]. In Documentos para la historia de Arequipa, /534-/575: Documentos de

-100
los archivos de Arequipa y Sevilla, edited by Padre Victor M. Barriga, Mercedario. Volume 2, pp. 4253. Arequipa: La Colmena. Arriaga, Pablo Jose de 1968 [1621] The Extirpation of Idolatry in Peru, translated by L. Clark Keating. Lexington: University of Kentucky Press. Barrionuevo, Alfonsina 1971 Chiaraqe. Allpanchis 3:79-84. Cusco. Bastien, Joseph 1978 Mountain of the Condor: Metaphor and Ritual in an Andean Ayllu. New York: West Publishing. Bertonio, Ludovico 1984 [1612] Vocabulario de la lengua Aymara. . Cochabamba: CERES. Betanzos, Juan de 1987 [1551) Suma y narracion de los Incas, edited by Maria del Carmen Martin Rubio. Madrid: Ediciones Atlas. Beyersdorff, Margot in press Suggested Glosses of Huaca Names. Appendix 3 in The Sacred Landscape of the Incas, by Brian S. Bauer. Austin: University of Texas Press. Bonnett, Percy 1983 Sitios arqueol6gicos de Espinar: Restos de Mauka-LIaqta. Arqueologia Andina 1:56-61. Celestino, Pedro 1982 [1792] Descripci6n orogrMica de la Provincia de Tinta. In Kanchi: La Provincia de Canchis a traves de su historia, edited by Vicente Guerra, pp. 74-78. Lima: Editora Humboldt. Chavez, Sergio 1988 Archaeological Reconnaissance in the Province of Chumbivilcas, South Highland Peru. Expedition 30(3):27-38. Cieza de Le6n, Pedro 1977 [1554] EI Seiiorio de los Incas. Lima: Editorial Universo. 1984 [1553] La cronica del Peru. Lima: Editorial Inca. Cobo, Bernabe 1990 [1653] Inca Religion and Customs, translated and edited by Roland Hamilton. Austin: University of Texas Press. Duviols, Pierre 1974-1976 Une petite chronique retrouvee: Errores, ritos, supersticiones y ceremonias de los yndios de la prouincia de Chinchaycocha y otras del Piru. Journal de la Societe des Americanistes 63:275-97. Flores, Jorge 1988 Mitos y canciones ceremoniales en comunidades de Puna. In Llamichos y Paqocheros: Pastores de llamas y alpacas, edited by Jorge Flores, pp. 237-251. Cusco: Centro de Estudios Andinos. Fornee, Niculoso de 1965 [1586] Descripci6n de la tierra del corregimiento de Abancay. In Relaciones Geograficas de Indias Peru, edited by Marcos Jimenez de la Espada. Volume 2, pp. 16-29. Madrid: Ediciones Atlas. Garcilaso de la Vega, el Inca 1966 [1609] The Royal Commentaries of the Inca and General History of Peru. Part One, translated

101by H.V. Livermore. Austin: University of Texas Press. Gasparini, Graziano and Luise Margolies 1980 Inca Architecture, translated by Patricia J. Lyon. Bloomington: Indiana University Press. Girault, Louis 1988 Rituales en las regiones andinas. La Paz: CERES. Glave, Luis 1987 Sociedad, poder y organizacion andinas: el sur peruano hacia el siglo XVII. In Comunidades campesinas: cambios y permanencias, edited by Heraclio Bonilla, pp. 61-94. Lima: Ediciones Exodo. Gonzalez Holguin, Diego 1989 [1608] Vocabulario de la lengua general de todo el Peru /lamada lengua Qquichua 0 del Inca. Lima: Universidad Nacional Mayor de San Marcos. Gorbak, Celina, Mirtha Lischetti, and Carmen Paula Muiioz 1962 Batallas rituales del Chiaraje y del Tocto de la Provincia de Kanas (Cuzco-Peru). Revista del Museo Nacional31 :245-304. Lima. Gow, Rosalind and Bernabe Condori 1982 Kay Pacha. Cusco: Centro de' Estudios Rurales Andinas. Gow, David and Rosalind Gow 1975 La alpaca en el mito y el ritual. Allpanchis 8:141-174. . Guaman Poma de Ayala, Felipe 1980 [1615] El primer nueva cor6nica y buen gobierno. 3 volumes, edited by John V. Murra and Rolena Adorno, with Quechua translated by Jorge Urioste. Mexico: Siglo Veintiuno Editores. Hemming, John 1970 The Conquest of the Incas. New York: Harcourt Brace Jovanovich. Herrera y Tordesillas, Antonio de 1615 Historia general de los hechos de los castellanos en las islas i tierra firme del Mar Oceano. Madrid: Juan de la Cuesta. Hinojosa, Ivan 1987 Poblacion y conflictos campesinos en Coporaque (Espinar): 1770-1784. In Comunidades campesinas: cambios y permanencias, edited by Heraclio Bonilla, pp. 229-256. Lima: Ediciones Exodo. Isbell, Billie Jean 1978 To Defend Ourselves: Ecology and Ritual in an Andean Village. Austin: University of Texas Press. McIntyre, Loren 1991 Amazonia. San Francisco: Sierra Club Books. Millones, Luis (editor) 1990 El Retorno de las Huacas. Lima: Instituto de Estudios Peruanos. Molina, Cristobal de 1959 [1575] Ritos y fabulas de los Incas. Buenos Aires: Editorial Futuro. Murua, Martin de 1946 [1590] Historia del origeny genealogia real de los Reyes Incas del Peru. Biblioteca Missionalia

Reinhard: Inca Shrine at Ancocagua


Hispanica. Volume 2. Madrid: Consejo Superior de Investigaciones Cientfficas, Instituto Santo Toribio de Mogrovejo. Pardo, Luis 1948 Dos fortalezas antiguas poco conocidas, Huaccrapucara. y Molloccahua. Revista del Instituto y Museo Arqueol6gico 12:3-23. Cusco. Pizarro y Orellana, Fernando 1639 Varones i/vstres del Nvevo Mundo. Madrid: Diego Diaz de la Carrera. Quirita, Rosa Alicia and Maritza Rosa Candia 1994 Arqueologia de Maria Fortaleza (Tacrachullo), Cuenca del Apurimac-Espinar. Unpublished thesis. Cusco: Universidad Nacional San Antonio Abad del Cusco. Reinhard, Johan
.

1985 Sacred Mountains:An Ethno-Archaeological

Study of High Andean Ruins. Mountain Research and Development 5(4):299-317. Boulder, Colorado. 1990 Tiahuanaco, Sacred Center of the Andes. In The Cultural Guide of Bolivia, edited by Peter McFarren, pp. 151-181. La Paz: Fundacion Quipus. 1991 Machu Picchu: The Sacred Center. Lima: Nuevas Imagenes. 1995 House of the Sun: The Inca Temple of Vilcanota. Latin American Antiquity 6:340-349. in prep. Coropuna: Mountain Temple of the Incas. In Architecture and Ritudl Space as Sacred Landscape, edited by D. Gundrum, M. Aviles, and R. Connolly. Roel, Josafath 1966 Creencias y practicas religiosas en la Provincia de Chumbivilcas. Historia y Cultura 1(5):25-32. Lima. Ross, Peter 1980 El Cerro Mismi, fuente ultima del Rio Amazonas. Revista del CIADAM (Centro de Investigaciones Arqueologicas de Alta Montana, Argentina) 4:27-29. Salomon, Frank and George L. Urioste 1991 [ca.1600] The Huarochiri Manuscript. Austin: University of Texas Press. Schobinger, Juan 1991 Sacrifices of the High Andes. Natural History . 100(4):62-69. 1995 Aconcagua un enterratorio incaico a 5.300 metros de a/tura. Mendoza: privately published. Sherbondy, Jeanette 1982 El regadio, los lagos y los mitos de origen. A/lpanchis 20:3-32. Valderrama, Ricardo and Carmen Escalante 1988 Del Tata Mallku a la Mamapacha. Lima: Centro de Estudios y Promocion del Desarrollo. Vasquez de Espinoza, Antonio 1948 [1617] Compendio y descripci6n de las lndias occidentales. Smithsonian Miscellaneous Collections. Volume 108. Washington, DC: Smithsonian Institution.

ANDEAN PAST 5 (1998)

-102

Table 1. Tribute paying population according to landholding units (repartimientos) in the provinces of Canas and Canchis in 1575 and 1812. From Glave 1987.
Repartimientos Checacupe - Hilave Cangalla Combapata Combapata Chiara Tinta Cacha San Pablo (Charrachapi) Sicuana Lurucache Maranganf Yanaoca Languisupa Layosupa Checasupa Pichigua Yaure Coporaque Ancocaua 1575 1812 211 368 70 171 441 445 129 1,162 179 200 350 157 239 515 510 651 393 172 6,363

488 115 161 118 671 323 59 400 322 120 679 256 227 322 922 660 239 29 6,111

TOTAL

103-

Reinhard: Inca Shrine at Aconcagua

Location of site of Marla Fortaleza in the Upper Apurrmac Valley Southern Peru
. ToWns and Inca sites .. Mountains ~ Lake {-.., Snowfields

-"

'

...'

.,

,'-

,'"p372m , ',', ,"~',:" " '.1"" , ,I ,,

\ , Ausangate

,-.. ,'1 I
..

N
~.

.J , , , (... (

,,

, '\ \" )

,~ *',,

t ..., l. Chimbol/a 5472m

ViIs'ro Lako

~
"

. Ca10m

Mismi~ ~
.5596m

/ J)
.f ~ -g

~
0 KM 50

Figure 1. A map of the region in which the temple of Ancocagua was located, including sites, towns, and geographical features noted in the text.

ANDEAN PAST 5 (1998)

-104

Figure 2. The rocky outcrop of Maria Fortaleza rises above the confluence of the Apurimac and Totorani Rivers. The main archaeological complex is found on the summit of the outcrop in the upper center of the photograph. .

105-

Reinhard: Inca Shrine at Aconcagua

MARIA FORTALEZA Approx. 7440'$ 7200'W Co;)


Circular structures

~ Structures

..

Funerary towers and tombs

.-:: .::::.Rock.

e.. . ". Walls ,,-" Trail

Contour Interval 20 meter$

Figure 3. Plan of the ruins of Maria Fortaleza (after Quirita and Candia 1994).

ANDEAN PAST 5 (1998)

-106

Figure 4. The notch between the outcrop of Maria Fortaleza (at the upper left) and the hill to the right is clearly visible. A large Inca structure can be seen in the lower left-center of the photograph.

107-

Reinhard: Inca Shrine at Aconcagua

Figure 5. This Inca structure stands in the route of access to the swnmit ruins of Maria Fortaleza.

ANDEAN PAST 5 (1998)

-108

Figure 6. One of the many chullpas (funerary towers) found on the summit of the outcrop of Maria Fortaleza. Person in photo is Alicia Quirita.

ETHNOGENESIS IN HUAMACHUCO

John R. Topic
Trent University

Introduction At the time of the Spanish conquest, the Inca Empire incorporated numerous ethnic groups speaking a variety of languages and characterized by distinctive dress. Many commentators have noted that, while the Inca encouraged the use of Quechua as a common language and spread the cult of the Sun as a common religion, they allowed ethnic groups to continue to worship local deities, speak the local languages, and maintain many traditional aspects of land tenure and political organization. Indeed, Spanish chroniclers suggest that the Inca went beyond passive acceptance of diversity to require those performing labor service for the state or those moved pelJ11anently to distant settlements by the state to maintain the traditional dress of their ethnic group. In this paper, I will examine the development of ethnic identity in a single case, that of Huamachuco in the north highlands of Peru. I particularly wish to trace the changes in ethnic consciousness brought about by the Inca administration. Before the Incas, we have only archaeological information upon which to base our interpretations. Archaeological data can document ethnic divisions where they are sharply drawn, for example between the Recuay and Moche cultures in the Nepefia Valley during the Early Intermediate Period (Proulx 1983). With the Inca, however, we can begin to apply mythical and political information derived from historic sources. In the case of Huamachuco, the archaeological distributions of cultural attributes associated with ethnic identity in the Andes suggest interactions with Cajamarca to the north and Conchucos to the south that resulted in a continuum of cultural variability. Archaeological, linguistic, and mythical evidence does not suggest sharp ethnic boundaries. There is, ANDEAN PAST 5 (1998):109-127.

however, more evidence of sharing between the Conchucos and Huamachuco areas than between the Cajamarca and Huamachuco areas. The Inca administration imposed on this dynamic situation a much more structured territorial, social, and political identity. Huamachuco is located in the northern sierra of Peru, at the extreme southern end of the Condebamba Basin (Figures 1 and 2). The Condebamba River flows nortll to join with the Cajamarca River. The conjoined rivers form the Crisnejas River, which flows eastward to the Marafion River. . Huamachuco is separated from Conchucos and the headwaters of several rivers that flow down to the Pacific (the Chicama, Moche, 'Vim, Chao, and Tablachaca Rivers) by high altitude grasslands and passes more than 4000 m in elevation. Ethnicity in the Andes There are three prominent perspectives on ethnicity in the Andean lit~rature: that, having developed in the remote past, ethnic groups are essentially autochthonous; that ethnicity is causally related to ecology; and that ethnic identity can only be understood in the context of state formation and collapse. These are not inherently contradictory positions and it is quite possible to accept all three premises simultaneously. The view that ethnic groups are autochthonous fits well with European conceptions of folk cultures and anthropological constructs like tribes, linguistic families, and archaeological traditions. Most scholars working in the Andes have, at least at times, implicitly assumed the long-term existence of ethnic groups in the Andes. Spanish sources like Polo (1940 [1561]:131), Sarmiento (1907 [1572]), Garcilaso (1966 [1609]), and Guaman Poma (1980 [1615]) describe Inca expansion as a process of conquering pre-existing ethnic

ANDEAN PAST 5 (1998) groups, and ethnohistorians (Murra 1982:238; .Pease 1982; Rowe 1946; 1982; Rostworowski 1990) have tended to accept the general veracity of those sources. John Rowe (1946:185, 256; 1982:110) points out that the Inca both combined small groups to form administrative units and broke up large states like Chimor, but, in general, they respected existing ethnic divisions. Maria Rostworowski (1990) thinks that the ethnic group constituted the highest level of integration during the Late Intermediate Period, but that this unit varied greatly in size. Franklin Pease (1982) cautions us not to consider ethnic groups as static entities and focuses our attention on the system of relationships that operated at different levels of integration; his analysis points out the vagaries of ethnic boundaries. John Murra (1982:238), on the other hand, does see the ethnic group as a perduring social construct that became the basic building block of the Inca administration. He (1980:85-86; 1975: chapters 1 and 3) goes somewhat further to imply that much of Inca organization was ethnic organization writ large (state mitmaq policy being modelled on ethnic vertical economies and state redistribution on ethnic patterns'of reciprocity). While the ethnohistorians often identify rather small sefiorfos as ethnic groups, archaeologists, myself included, have sometimes focused on larger units and treated Rowe's (1946) list of Inca provinces somewhat uncritically as a list of ethnic groups present during the Late Intermediate Period. Archaeologists have considered the general continuities within regions over large time periods as evidence for long term cultural stability and limited population movement; "archaeological traditions" (Willey 1945; Willey and Phillips 1958: 34-39), identified by the persistence of particular material attributes, are good examples (stirrup spouts on the north coast, double spout and bridge vessels on the south coast, keros in the south highlands, continuities between Moche and Chimu or Nasca and lea, etc.). When we see the great art styles of the Early Intermediate Period (Moche, Cajamarca, Recuay, and Nasca), together with Tiahua-

-110 naco, then we may talk of political unity but we alsc;> think of cultural unity. In his classic study, Fredrik Barth (1969: 19-20) pointed out that ethnic groups sharing the same general territory will often occupy different ecological niches. The ecologically determined nature of ethnicity is seen in the Andean area in terms such as yungas, chaupiyungas, and quechua, which apply to both' people and to ecological zones. While these terms do not correspond to named ethnic groups, they were used in a general way to distinguish broad ethnic differences (Rostworowski 1990:13). Moreover, there are general cultural differences between at least the yungas and quechuas that can be defined on the basis of both historic and archaeological information (Murra 1980; 1975; Rostworowski 1975; 1989b; cf Rowe 1948). In a somewhat more specific way, archaeologists and ethnographers recognize a relationship between geographical regions (the altiplano, for example, (Bolton 1979 and a degree of shared culture. In their own "histories", highland peoples sometimes described themselves as taking over the quechua ecological zone from previous inhabitants, who were killed or chased down into the lower-elevation chaupiyungas (J. Topic 1992; Taylor 1987). Alternatively, quechua dwellers were occasionally considered the original inhabitants (huari) who were intruded upon by high altitude herders (llacuaz) (Duviols 1973). It is important to note that these mythological histories emphasize descent from a founding ancestor as the legitimation of a people's right to control an ecological zone. In the emic perspective, then, ayllu affiliation and ecological location jointly determine a significant degree of ethnic identity. A rather different view is taken by some theorists who see ethnogenesis (i.e., the creation of an ethnic cultural identity) as a historical process related to the formation of states and colonial empires (Vail 1989; Gailey and Patterson 1987). Gailey (1987) notes ways in which ethnic identity can be forged out of resistance to state formation and state sponsored

111ideologies; in this view, ethnic identity represents "authentic" culture while state 'propaganda engenders "spurious" culture. Patterson (1987) applies this model to the Inca case, pointing out that imperial policies resulted, in different instances, in both ethnocide and ethnogeneSlS. Vail (1989), writing about the development of ethnic consciousness in southern Africa in colonial and post-colonial times, has proposed a historical model that relates the development of ethnic ideologies to the needs of a variety of actors; in this model foreign and indigenous intellectuals, colonial administrators, emergent bourgeoisie, and ordinary villagers all found that the creation of ethnic identities served their needs even though the various actors were often in opposing situaparallels in the Inca Empire: the conscious codification of custom, tradition, and language provided ideological support for the curacas, who needed to legitimize their claims to authority over the "ethnic" group; the Inca were then able to use -thelocal elite as an efficient means of indirect control while, at the same time, obfuscating the major structural changes' imposed on local organizations; commoners, especially men who were away from families and fields on service for the state could rely on the local elite, operating on the basis of "custom" and "tradition", to protect their interests at home. The Recognition of Ethnic Identity Barth (1969:13-14) notes that there is no simple correlation between shared cultural attributes and ethnic identity. Ethnic groups use cultural attributes as signals of their identity, but the critical feature of ethnic groups is selfascription by the members of the group and ascription by others. Cultural attributes signal the social boundaries of the ethnic groups and may include dress, language, house type, generallife-style, and basic values of morality and excellence (ibid: 14). The defining characteristics are often beh~vioral or cognitive and definitions are often in a state of flux (e.g., Clark 1994; Hill 1992; Mahmood and Armstrong 1992; Rasmussen 1992). The cross-

Topic: Ethnogenesis in Huamachuco cultural variability in attributes that might signal ethnic affiliation means that it is impossible to specify universal archaeological correlmesofethnicidentity. Referring specifically to groups in the Andean area, Rostworowski (1990:16-21) sug':' gests that the following cultural attributes were considered key identifiers of ethnic groups; these attributes, of documented significance in early historic times, would also have been important in late prehistoric times.
a. Unity a/Origin and Beliefs. Such unity is expressed by a people's shared belief that their ancestors all emerged from the same origin place (pacarina). Common origin may also be related to worship of particular divinities (huacas) which pertain to the whole ethnic group. The mummies (ma//quis) of the ancestors are a further symbol of unity. Unity of Language or Dialect. There were numerous languages in the Andean area before the Inca conquest. Cultural and ethnic identity was related to language and, in fact, the indigenous names for these languages, like runa simi, often mean the "speech of people". Common dress. Pedro Pizarro (1978 (I57]]: chapter ]6) st1Jtesthat the various ethnic groups were known by their dress. Acosta (1940 [1590]:302 [Book 6, Chapter 16]) notes that it was illegal to change or modify the ethnic costume. Molina ("El Cusquei'io" 1943 [ca. 1575]:9) specifies that the ethnic costume was similar to the dress worn by the principal huaca of the group. Cieza (1984 [1553]) and Guaman Poma (]980 [1615]) provide the best details on ethnic costume. Socia-political Unity. Political units were of varying sizes but Rostworowski interprets the largest political groupings of the Late Intermediate Period to encompass no more than a single ethnic group.

tions. . Many of these African situationshave


.

b.

c.

d.

These attributes will be used to try to trace the development of ethnic identity in the Huamachuco area through time. It is, of course, impossible to consistently identify all these attributes at all times. Moreover, the attributes that can be identified may lead to conflicting definitions of ethnic boundaries, and the criteria for the definitions are often in flux.

113with the Santa River.3 The shrine and oracle of Catequil was located at San Jose Porcon, near the center of the province. A river named after Catequil's mother is about halfway between the oracle and the pacarina. There are several hills and quebradas along the north and northeastern frontiers of the province that still bear the toponym "guachemin"; these probably commemorate the places where the guachemines were driven out of the province. Interestingly, the quebradas named "guachemin" all descend abruptly to the hot chaupiyungas zone. Alfredo Torero (1989: 228-29) points out that guachemin may be equivalent to guaxme (fisherman) of Domingo de Santo Tomas (1560:s.v.), and that Guaman Poma calls coastal fishermen uachimis and uachime yunga. Thus, the creation myth also defines the territory of Huamachuco as ecologically sierra and the people -as ethnically serranos and contrasts them to people adapted to life along the seacoast. Four other huacas listed in the account reinforce this ecological and ethnic classification: Nomadoy, 'Pomacama, Vlpillo, and Quimgachugo (Figure 1). Albornoz (1967 [ca.1582]t also mentions the first two and states that they are the principal huacas of Llampa 'and Guacapongo, two of the four indigenous guarangas (Inca administrative groups of about a thousand households) in the province of Huamachuco. These two huacas
3 Albomoz (1967 [ca. 1582)) says that the pacarina of Huamachuco was called Guaracayoc. Guaracayoc (Quechua) may be translated as a person [adept] with a sling and may be a title of Catequil; the Augustinians list Guaracayoc as one of the nine principal huacas of Huamachuco, but do not provide further comment (1. Topic 1992). 4 Albomoz.(1967 [ca. 1582)), Arriaga (1968 [1621)), Betanzos (1987 [1551)), and Calancha (1974-81 [1638)) also provide other infonnation about the huacas of Huamachuco which sometimes complements and sometimes conflicts with the infonnation provided by the Augustinian priests. The Augustinian account, however, is the most complete source and the only one based on extended and early observations in Huamachuco. For further discussions of this infonnation see J. Topic 1992 and 1994 and Topic and Topic 1993.

Topic: Ethnogenesis in Huamachuco seem to have been located at low elevations in deep river valleys just above the chaupiyungas zone. The other two huacas can be identified with the highest peaks within the territories of each of these two guarangas. Together, each pair of huacas define the upper and lower elevational extremes of the territories of their guarangas (Topic 1992:74). The ecological information in the myth is related to the replacement of the original population with a new population. In this ~ase, and in other similar cases (Calancha 1976 [1638]:933-34; Taylor 1987: Chapter 8), the claim to the territory is based on the destruction of the original population and/or the forced removal of the previous inhabitants to a different ecological zone. This type of origin myth draws sharp distinctions between the occupants of adjacent ecological zones, who are depicted as enemies. Now in late prehispanic times, the importance of Catequil as an oracle was acknowledged throughout the Inca empire and this stature causes some confusion amongst the historical sources about the origin and history of the deity. There is also confusion about the events surrounding the destruction of the shrine of Catequil by the Inca. Although Catequil is described by the Augustinians (San Pedro 1992 [1560]:173-74) as ". . . el ydolo mas temydo y honrado q. avia en todo El peru adorado y reveren~iado desde quito hasta El cuzco . . .",5 the creation myth demonstrates a special association with Huamachuco. Sarmiento (1907 [1572]:165-6) refers, in passing, to Catequil as the huaca of Cajamarca and Huamachuco, but later (p. 176), while describing its destruction by Atahualpa, he refers to it as the oracle and huaca of Huamachuco. Albornoz (1967 [ca. 1582]:31) lists Catequil6 under the heading "Provincia de Guamachuco
5 The most feared and honored idol that there was in all Peru, adored and revered from Quito to Cuzco. 6 He writes the name of the oracle as "Apocatiquillay" and places it near Uruchalla. For the probable locations of two different tambos (way stations) called Uruchal, see Topic and Topic 1993, figure 2.1. The two tambos are northwest and southeast of San Jose Porcon, where the Augustinians say the shrine was located.

ANDEAN PAST 5 (1998) y Caxamarca" and affirms that it was one of the most important in the whole empire but says specifically that it is "guaca . . . de los indios guamachucos." Despite Cieza's assertion that Cajamarca and Huamachuco worshipped the same gods, there is no evidence that Catequil was particularly important in Cajamarca before the cult was spread by the Inca. On the other hand, Arriaga (1968 [1621]:203) and Calancha (1976 [1638]: 1062-63) say that Catequil was especially worshipped throughout the province of Conchucos as well as in the province of Huamachuco (Figure 2). They relate the presence of Catequil in Conchucos to the destruction of the shrine at Porcon and this relationship is the only means of dating the presence of Catequil in Conchucos. Interestingly, there are two versions of the story of the destruction of the shrine at Porcon, one based on witnesses from I1uamachuco and the other based on witnesses from Conchucos. In the Huamachuco version, Atahualpa destroys .the shrine just as the Spanish arrive in Peru and make thei:t:way to Cajamarca (San Pedro 1992 [1560]; Betanzos 1987 [1~57]; Sarmiento 1907 [1572]); in this version the pieces of the idol were eventually found by the Augustinians, who ground them up and threw the powder into the river. In the Conchucos version, it is the son of Topa Inca who destroys the shrine at Porcon. Arriaga (1968 [1621]) makes Huascar the son of Topa Inca, while Calancha (1974-82 [1638]) corrects the genealogy so that Huayna Capac is the protagonist. Huayna Capac (the son of Topa Inca) set fire to the shrine in Porcon but the priests were able to rescue the idol and bring it to Cabana and Tauca where they built a new temple for it. There is some question whether this idol was eventually found and destroyed by Fray Francisco Cano or whether it was hidden by the natives. There are interesting contradictions here. On the one hand, both stories agree that Catequil originated in Huamachuco. On the other hand, the Conchucos version has Catequil ar-

-114 riving in Cabana before the shrine in Porcon would have been destroyed, according to the Huamachuco version. The chronology of the Conchucos version is not particularly credible because it was probably Huayna Capac who was responsible for spreading the cult of Catequil to Ecuador.7 Moreover, the Huamachuco version specifies that the idol was. shattered by Atahualpa, while in the Conchucos version the idol is rescued and moved without being fragmented. Finally, it is certain that the Augustinians in Huamachuco saw pieces of an idol that was claimed to be Catequil in the middle of the 16th century; if these were, in fact, fragments of the idol from Porcon, the unbroken idol in Cabana at the beginning of the 17th century must have been a different piece. I suggest the following interpretation of the contradictions. I suspect that Catequil, as well as being an oracle, was a deity related to thunder (San Pedro 1992 [1560] f. 6v8; Silva Santisteban 1986a:23) and that, in this guise, he was worshipped widely throughout Conchucos and Huamachuco. Catequil, as oracle, developed later in the province of Huamachuco from this celestial deity. Here he was also incorporated into the creation myth as a culture hero and had his shrine at Porcon. It is noteworthy that the Conchucos version does not talk of Catequil functioning there as an oracle nor is there the claim that the idol, once moved to Cabana, enjoyed a widespread reputation. The confusion in the historical sources is caused, in part, by Spanish attempts
7 The spread of the cult of Catequil to Ecuador is the subject of ongoing research and is peripheral to this discussion. At the very least, however, there is one account (Sanniento 1907 [1572]:165) which states that Huayna Capac had the huaca "Cataquilla" or "Catequilla" of Huamachuco and Cajamarca with him in Quito. 8 San Pedro (1992 [1560]: f. 6v): . . . q.s grande El CatamLo q. tienen a cataquil y el temor/ porq. dizen q.s el q. haze los rayos y truenos y Relampagos los q.ales/
haze tirando con su honda

. ..

[The respect and the fear

they have for Cataquil is great because they say it is he who makes the lightning bolts; and thunder, and flashes of lightning. These he makes by throwing them with his sling.]

115to reconcile the particular oracle at Huamachuco with a more general celestial deity. As we will see below, the confusion is also compounded by the Inca and colonial administrative grouping of Huamachuco with Cajamarca. This leads Albornoz (1967 [ca. 1582]: 31) to list Catequil, and the other Guamachuco huacas, under the heading "Provincia de Guamachuco y Caxamarca" and then, immediately list the huacas of Cajamarca under a separate heading "Caxamalca." In the same way, Sarmiento (1907 [1572]:165-6 and 176) refers to "cataquilla" as the huaca of Cajamarca and Huamachuco and later, more specifically, as the huaca of Huamachuco. The most likely chronological implications of these dynamic associations of Catequil with different places and groups of people is that Catequil has his greatest antiquity as a widespread sky deity in Conchucos and Huamachuco. He was later, perhaps in the Late Intermediate Period, established as an oracle at San Jose Porcon. During'the Late Horizon, he was firmly associated with Huamachuco, where Inca kings ~ame to consult him and to destroy him. By the time of the destruction of the shrine of Catequil,on the eve of the Spanish conquest, his cult was widespread in the north; and by 1560, if not before, Catequil had become identified as a culture hero relating to the creation myth for people occupying the Incaic and early Colonial province and encomienda of Huamachuco. Unity of Language or Dialect It is well known that the indigenous language of Huamachuco was Culle (Silva Santesteban 1986b; Torero 1989). Although the language was spoken into the early years of this century, there are only a few words recorded (Silva Santisteban 1986b). Again, largely based on Cieza's comment, there has been a tendency to consider Culle also to be the indigenous language of Cajamarca (Silva Santesteban 1986b). In fact, there is little evidence to support that view. The single piece of documentary evidence for Culle being spoken in Cajamarca was

Topic: Ethnogenesis in Huamachuco published by Jorge Zevallos Quinones (1948; Silva Santisteban 1986b; Torero 1989:224). The document is an expediente from 1774 in which Miguel Sanchez del Arroyo, priest of the town of Ichocan and the Condebamba Valley, states that he understands "la culle por curiosidad e industria y por haver administrado los santos sacramentos entre los que la acostumbran ablar . . ."9 Ichocan has always pertained to Cajamarca, but in 1774 (and until 1854) the Condebamba Valley was part of the province of Huamachuco (Espinoza 1971:3031) (Figures 1 and 2). During the same epoch, Martinez Compafi6n (1978-85[1789] Volume 2:iv) specifically refers to Culle as "(la) lengua culli de la Provincia de Guamachuco." On the other hand, Culle was still spoken in Pallasca, located in the old province of Conchucos, in 1915 (Rivet 1949). Linguistically, then, there may be closer connections between Huamachuco and Conchucos (and Huacrachuco). Several years ago, I noticed that the Augustinians (San Pedro 1992 [1560]:205) referred to the earth as "pachamama y chucomama". This led me to question the common derivation of the term chuco from the Quechua for hat or headdress (e.g., Garcilaso 1966 [1609]:476) and consider it instead, along with pus (Martinez Compafi6n 1978-85 [1789]: Volume 2:iv), to be a Culle word meaning earth or place. As a toponym, the word chuco (or variants) commonly occurs throughout the former province of Huamachuco (i.e., the territory around Otuzco, Santiago de Chuco, Huamachuco, and Cajabamba) (Figure 2). It also occurs in Pataz, Pallasca and Corongo but is rare in the Cajamarca area and largely limited to the southern part of the modem Department of Cajamarca, adjacent to Cajabamba. Torero (1989:226) has independently recognized the toponymic significance of chuco but has carried his analysis far beyond my own. From a study of documentary sources and, especially, topographic maps (Carta Nacional) he has reconstructed the distributions
9 "He understands Culle because of his curiosity and industry and for having administered the holy sacraments among those who are accustomed to speak it."

ANDEAN PAST 5 (1998)


.

-116 Huamachuco cultures had very distinct ceramic assemblages, but there were interesting patterns of borrowing and sharing. Huamachuco ceramics were primarily utilitarian. Deco-ration was infTequent and executed quickly and carelessly. Cajamarca ceramics included very high proportions of decorated vessels, highlighted by the painted kaolin wares of the Cajamarca Cursive style. Cajamarca ceramic pieces were valued in Huamachuco and were imported in significant quantities. Potters in Huamachuco also imitated the Cursive style during the end of the Early Intermediate Period and Middle Horizon, using locally available materials to produce mediocre copies of the technically far superior Cajamarca originals. Huamachuco potters never mastered the firing of kaolin clays to attain the degree of hardness reached by Cajamarca potters, and the kaolin pedestal bowls produced in Huamachuco have soft paste and eroded painting. A minor insight into the ease with which Huamachuco ceramicists borrowed stylistic elements is offered by the common occurrence in the EIP and MH of heavy strap handles in brown paste, slipped in red and decorated with impressed circles and incised lines. These "Cajamarca Coarse Red" handles (Terada and Onuki 1982: plate 38a) caught the fancy of Huamachuco potters, who produced them on local clays. Architectural differences seem to include both the types of buildings constructed and the masonry style used. In Huamachuco, the typical domestic structure is a long narrow multiroomed building called a gallery (McCown 1945; Topic 1986). In the gallery, the rooms are arranged in a single file and, usually, the doors of the rooms all open to one side of the building.. The side to which the doors open is often a patio, enclosed by the building which curves around it. A common type of public building is the niched hall; this is an immense roofed volume often measuring 6 m x 40 m or more in plan and with ceiling heights of as much as 9 m. The masonry style is quite distinctive and includes long and shortwork corners and ordered chinking. The architecture of Cajamarca from the late EIP, MH, and Late Intermediate Period (LIP) is not well known or described, but only one site, Coyllbr, seems

of severallanguagesor dialectsin the northern

sierra of Peru. While some, like Den, are based on very little evidence, Culle is represented by a number of different word parts. He arrives at the conclusion that Culle was spoken throughout the province of Huamachuco and into at least part of the province of Conchucos, but that it was not spoken in Cajamarca (Torero 1989:218, 222). He (ibid:244) believes that the language Cieza heard spoken in both Huamachuco and Cajamarca was Quechua IIA (Yungay). A similar, though more geographically limited analysis, suggests the presence of a linguistic boundary between Cajamarca and Huamachuco in the Chicama valley (Krzanowski and Szeminski 1978). Again, as with religious beliefs, there seem to be more similarities in language between Huamachuco and Conchucos than between Huamachuco and Cajamarca. Unity of Dress (Style) In addition to Cieza's statement, Pedro Pizarro (1978 [1571]:73) also states that' the Cajamarca and Huamachuco men had similar headdresses, consisting of long tresses wrapped with wool cords. I am unaware of extant pictures or further descriptions which would a~lowme to enlarge on these two, very brief, statements regarding dress. It is possible, however, to infer a bit more on the use of style as an indicator of ethnic identity. In terms of the archaeology of Cajamarca and Huamachuco, there are clear differences in ceramics, architecture, and, to some extent, burial patterns extending back to the Early Intermediate Period. In this section, I am treating the terms "Cajamarca" and "Huamachuco" as archaeological cultures rather than Incaic provinces. Archaeologists have relied heavily on ceramics to identify groups, boundaries, and interrelationships; the durability of ceramics in the archaeological record and the richness of variability possible in their manufacture and decoration make them a very useful medium to sort out questions of group identity. Through the Early Intermediate Period (EIP) and Middle Horizon (MH), the Cajamarca and

117to be related to the Huamachuco architectural style (Reichlen and Reichlen 1949; Julien 1988). Burial patterns are probably not as distinct as architectural or ceramic styles. Still, one of the typical patterns in Huamachuco -- secondary burial of bones in the walls of niched halls -- is not known from Cajamarca, while ventanillas, or rows of tombs cut into the soft trachite, are known from Cajamarca but not from Huamachuco. Obviously, both patterns may be varients of, or derivations from, burial in caves, and the distribution of ventanillas may be further affected by the availability of suitable trachite outcrops. In contrast, there appear to be closer archaeological ties to Conchucos, south of Huamachuco, but these relationships are somewhat tenuous. There are a few trade pieces and some ceramic influence from the Pashash style of Conchucos in the Huamachuco area during the late Early Intermediate Period, but these are much less prominent than the influence iTom Cajamarca. Stone carvings, on the other hand, indicate more intense interaction between the areas of Huamachuco, Santiago de Chuco, and Cabana (Grieder 1978; Kroeber 1950; McCown 1945; Schaedel 1952). It is intriguing that the masonry style so typical of the Early Intermediate Period and Middle Horizon in Huamachuco occurs in the Preceramic Period at La Galgada (Grieder and Bueno 1985). A similar masonry style is widespread in Ancash during the late Formative (Daggett 1983; Pozorski 1987; Wilson 1988). There are two sites which have buildings similar to those described above for Huamachuco: the Rondan Circular Construction from near La Pampa (Terada 1979) is similar to the domestic galleries and at Yayno near Pomabamba the few photographs and brief description available (Tello 1929:30-36) indicate an important site with both buildings and masonry similar to that of Huamachuco. While the Rondan Circular Construction seems to date from the Late Intermediate Period, the masonry style used at Yayno may indicate an earlier date.

Topic: Ethnogenesis in Huamachuco The buildings at La Galgada, which are related to the Kotosh Religious Tradition (Burger and Burger 1980), are not particularly similar to known buildings from Huamachuco, but still may be related. They are basically square in plan, but with rounded comers an4 niches in the upper part of the walls. McCown (1945: figure 12, e and f) shows a plan of two buildings at Cerro Campana, an Early Horizon site in Huamachuco, that might be similar. .While these buildings have not been excavated, the thick walls suggest that they may once have had niches. If so, these two buildings may form a developmental link between La Galgada and the later Huamachuco niched halls. In this regard, it is interesting to note two further points about La Galgada. First, multiple burials were placed in the niched buildings as they were sequentially abandoned, filled in, and new buildings constructed on top (Grieder and Bueno 1985). This use as burial places is different in detail but possibly related to the incorporation of burials in the walls of niched halls at Huamachuco. Second, La Galgada is located near the foot of Cerro Huacate, the pacarina ofHuamachuco. Beyond La Galgada, 'burial patterns are, again, intriguing but not very useful. There is a mausoleum at Cerro Amaro near Huamachuco (Topic and Topic 1984) that is similar to the mausoleums at Wilka Wain and Honcopampa (Bennett 1944; Isbell 1991). However, EIP and MH burial structures, usually referred to as chullpas but often with multiple chambers like the mausoleums, occur sporadically in the north highlands from Chota (Shady and Rosas 1976) to Huaraz. It should also be noted that burial in walls has been described for Cuelap (Reichlen and Reichlen 1950). Stylistic information suggests, then, that there were no sharp ethnic boundaries between Cajamarca, Huamachuco, and Conchucos during the late EIP and MH. Huamachuco accepted ceramic influence from both Cajamarca and Conchucos while also producing a distinctive style. Moreover, within the area that later became the Incaic province of Hua-

ANDEAN PAST 5 (1998) machuco, there was never a single unified ceramic style. Nevertheless, architecture and stone carvings suggest a stronger stylistic boundary between Cajamarca and Huamachuco than between Huamachuco and Conchucos during the ElP and MH. Throughout the areas under discussion, the LIP ceramics are poorly defined. Ceramic assemblages lack the high proportions of decorated wares or exotic shapes that characterize contemporary groups like the coastal Chimu. Ceramics are utilitarian serving, cooking, and storage vessels with little apparent symbolic or ritual importance. In the immediate Huamachuco area, two distinct styles are dated to the LIP (both initially defined in Krzanowski 1986). The Huamachuco Incised style is characterized by jar forms in orange paste decorated with impressed circles, usually on applique bands. This style has antecedents in the late EIP and MH, during which jars in soft orange and grey pastes had frequent applique decoration; often the jars are face-neck vessels with modelled earspools and facial features, and sometimes hands holding a flute to the lips. This later Huamachuco Incised style lacks the anthropomorphic designs, and is executed on considerably harder pastes. The style overlaps into the upper Chicama and, to a lesser extent, upper Moche valleys. The Huamachuco-onWhite style (see McCown 1945: plate 22 c, d, and e) includes both jars and bowls, executed in a hard grey paste with heavy inclusion of crushed rock. Surfaces are often given a white or cream wash, on which large red and/or black circles, spirals, and meanders are sloppily painted. This ceramic style occurs at some high altitude sites to the west, but is most common in the Huamachuco area. LIP ceramics on the western slopes of the Andes, between Huamachuco and the coast, are most typically characterized by thickwalled jars made of hard brown and red-brown pastes, with large flaring rims; these jars lack any decoration other than a broad red band or red slip on the rim. Chimu sherds are quite common on LIP sites in the west slope area, but absent from the Huamachuco area proper.

-118 Masonry styles are variable, with few sites in the west slope area (e.g., Huasochugo and Cerro Sulcha) and in the Huamachuco area (Cerro Grande) continuing to use the earlier masonry style. The long gallery buildings were no longer being constructed. The archaeological evidence from the LIP, cannot be aligned easily with information from the early Colonial Period about the location of the guarangas or of provincial boundaries. Huamachuco Incised pottery is somewhat correlated with Llampa guaranga and the persistence of the Huamachuco masonry style, with its long and short work comers and ordered chinking, is loosely associated with Guacapongo guaranga. The sharpest division, running roughly along the continental divide, is between the area with Chimu trade sherds and the area lacking Chimu trade sherds, and this' cuts through at least Llampa guaranga and possibly Guacapongo guaranga. Socio-Political Unity .

Several years ago Theresa Topic and I (Topic and Topic 1985) felt that there was strong evidence for the formation of a state during the EIP and MH with its capital at Marcahuamachuco. This evidence was largely architectural in nature, consisting of settlement patterns and masonry style. The area in which the architectural evidence occurred corresponded well with the Incaic boundaries of the province, but included outliers like Coyllor, near Jesus in Cajamarca, and Cungush, near Cabana in Conchucos (Alberto Bueno M., personal communication 1986). Since then, we have conducted much more research in Huamachuco and we were also able to test excavate some of the outlying sites. Today, we still think that there was considerable cultural unity throughout much of this area. However, we are no longer convinced that the region was politically unified. We would now stress confederation instead of consolidation and, indeed, would relate the process more to a developing ethnic consciousness within the province than to conquest. A number of autonomous curacas were no doubt players on the political scene, con-

119stantly negotiating for power on behalf of their communities and kin groups, using alliance, negotiation, and ritual display to attain desired ends. The center of these negotiations and shifting alliances for many centuries was the immediate Huamachuco area. Marcahuamachuco, which covers some 240 hectares, was the largest site in the north sierra. It seems to have functioned, at least in part, as a ceremonial center where the bones of the ancestors, buried in the walls of the niched halls, were worshipped (1. Topic 1986; 1991). Another shrine, Cerro Amaru, was related to water and was the site with the greatest amount of imported pottery, obsidian, and other exotic materials (Topic and Topic 1984; Topic and Topic 1992). Cerro Sazon, which covers at least 20 hectares, may have been the secular center. Additionally, the site of Viracochapampa (32 hectares) was under construction. These developments, however, took piace during the late EIP and the MH. During the LIP, there is a definite shift in economic focus toward the west slope of the Andes. Groups in this area were looking toward the coast: the largest sites in this area, Carpaico, Cuidista, Chamana, Cerro' Sulcha, and Huasochugo, are on roads leading up from the coast and are near the points where those roads first climb out of the chaupiyunga to the zone of rainfall agriculture. In contrast, LIP sites in the Huamachuco area are not located near the major roads. While Marcahuamachuco is still partially occupied, there is no new monumental construction; the three next largest sites fall in the 3.2 to 6.4 hectare range and are smaller than the largest sites in the west slope area. In both the Huamachuco and west slope areas the site size hierarchy suggests political unification only at the level of the curacazgo. During the Late Horizon socio-political units were hierarchically arranged. One level 'of the hierarchy can be documented to' some extent by the encomienda grants made by Francisco Pizarro. The earliest of these grants were made before the Spanish had any detailed knowledge of the country and relied on Inca informants (John Murra, Franklin Pease,

Topic: Ethnogenesis in Huamachuco Maria Rostworowski, and John Rowe personal communications 1991; Porras 1978:394). Huamachuco was given in encomienda as a single unit (Castro 1992:xxx; Loredo 1958: 255-258; Hampe 1980:101); this encomienda is, in fact, the basis for the definition of the Province of Huamachuco that I have been using. Cajamarca was also originally granted hi encomienda as a single unit but was later (by 1548) split into two encomiendas (Lee 1926:18-19; Urteaga 1942:12-13; Loredo 1958:255-258; Hampe 1980:103). The information for Conchucos is not as clear: Francisco Pizarro was the original encomendero but I am not sure whether he held all of Conchucos or only a part; in 1543, after the death of Pizarro, Conchucos was divided between two Spanish encomenderos but part was also claimed by Dona Francisca, Pizarro's daughter (Cook 1978; Rostworowski 1989a:37; Hampe 198b:l00-lOl). In all cases the grants of encomienda were made by assigning native lords and their Indians, not. territories, to the encomenderos, so that encomiendas consisted of groups of people who were recognized as some sort of unit by the Spanish. However, within the larger encomiendas, such as Huamachuco and Cajamarca, there were subdivisions. In northern Peru, these divisions were referred to by the terms used to designate the decimal administrative units, guaranga (1000) and pachaca (100), though the populations of the units in Huamachuco do not seem really to have been very close to the decimal ideal. Huamachuco was divided into four indigenous guarangas and also included a guaranga of mitmaq (people permanently moved by the Inca from one part of the empire to another) of highland origin, a guaranga of mitmaq of coastal origin, and a guaranga of chaupiyungas, who lived in their home territory but were administered from Huamachuco (Espinoza 1974). Each guaranga contained a varying number of pachacas. While the pachacas and guarangas were grouped and ranked to form larger administrative units, there is no social unit that can be clearly identified as an ethnic sefiorio; sharply defined ethnic boundaries can neither be distinguished between individual pachacas nor

ANDEAN PAST 5 (1998) between whole provinces like Conchucos and Huamachuco. In addition to creating a multi-ethnic mosaic by moving mitmaq into Huamachuco and including chaupiyungas in the administrative unit, the Inca may have restructured the indigenous guarangas themselves. Espinoza (1974:34) has argued that one (Lluicho) of the four indigenous guarangas was created by the Incas; on the basis of huaca distributions, I (Topic 1992) have made the argument that, in fact, two (Lluicho and Andamarca) of the four guarangas resulted ITomInca restructuring of the local population. This point is important chronologically: the two guarangas (Llampa and Guacapongo) with huacas defining their upper and lower elevational limits were coalescing in the LIP, as indicated by ceramic and architectural evidence respectively; the Incas

-120

century. On the other hand, the grouping of mitmaq ITomGuambos, Cajamarca, and Huamachuco helped to create a feeling amongst the mitmaq that the larger Inca administrative unit comprehending all that area had some real social significance to them. In both ways, the Inca policy developed an identification of the individual as a member of larger social units, and broke down the parochial tendency of the segmentary lineage. Conclusions Archaeological, linguistic, and ethnohistoric data provide more evidence of a longterm shared tradition between Huamachuco and Conchucos than between Huamachuco and Cajamarca. The shared features, however, were not developed to the point that perduring and politically unified ethnic groups can be identified. Instead, as Pease (1982) suggests, there was a complex system of interrelationships that operated at different times and levels of integration; we can recognize attributes, like language, religion, and style, that might be used to define ethnic groups, but the boundaries between the groups are vague and the degree of cohesion fluctuates.

recognized them, codified their . boundaries,


and created two new guarangas in the east in order to arrive at a quadripartite division. The Inca attached Huamachuco for 'administrative purposes to Cajamarca and Guambos while Cbnchucos was attached to Hwinuco or Huaylas (Cieza 1984 [1553]:226, 234; Pizarro 1978 [1571]:220-21) (Figure 2). On the surface, this division between Huamachuco' and Conchucos would appear to be simply an administrative convenience. One piece of information, though, suggests that it was part of an ongoing strategy of restructuring populations. Mitmaq ITom Huamachuco were moved to Chimbo, in Ecuador, together with mitmaq ITom Guambos and Cajamarca (Miguel de Cantos 1965 [1581]:255).\0

The Inca administration manipulated ethnicity, at least in the Huamachuco area. They split closely related groups, like Huamachuco and Conchucos, into different administrative units. They modified the guaranga groupings within the resulting provinces. Chaupiyungas groups, defined mythically as enemies, were appended to the unit. They then lumped Huamachuco together with less related groups, such as Cajamarca and even Guambos, to This example provides an illustration of form a larger administrative unit and they used two possible ways in which mitmaq policy mitmaq policy to create a "spurious ethnicity" affects ethnic identity. First, the mitmaq ITom. (ef Gailey 1987) to support their administraHuamachuco who, before leaving home, tive restructuring. would have identified more closely with their There are conclusions to be drawn on sevguaranga or pachaca, now identified themselves as "mitimas de guamachuco" and con- erallevels: tinued to do so until the end of the eighteenth 1. Theoretically. In the case of Huamachuco, ethnic identity was imposed from above rather than developing out 10Another indication that the Inca spread the cult of of a process of indigenous resistance to Catequil to Ecuador is the location of a hill in Chimbo the Inca state. If, in fact, at the time of called by the Ecuadorian variant of the name: "Catethe Spanish conquest the people of quilla".

121Cajamarca and Huamachuco spoke the same language (Cieza 1984 [1553]), dressed the same (ibid.; Pizarro 1978 [1571]), and shared Catequil as a principal huaca (Sanniento 1907 [1572]; Albornoz 1967 [ca. 1582]), it would be a demonstration that ethnicity had been extremely manipulated by the state. I suspect, though, that the people of Cajamarca and Huamachuco still preserved more elements of distinctiveness than the chroniclers suggest and that there are other explanations for the similarities observed: that perhaps they were beginning to speak a shared dia.lect of Quechua as Torero (1989) suggests and that the shared aspects of religion and dress were overly simplified glosses on the part of the Spanish. 2. Methodologically. The. information presented here suggests that we cannot . assume that any level of the Inca administrative hierarchy corresponds to an indigenous ethnic group. This means, moreover, that we cannot use lists of Inca provinces or encomienda grants as a shortcut to the reconstruction of Late Intermediate Period polities and social groupings. On the other hand, the example presented here should encourage us to use archaeological and documentary information in a complementary way to study the impact of the Inca administration on indigenous perceptions of self.

Topic: Ethnogenesis in Huamachuco lowing them to make major structural changes. Acknowledgments Archaeological research in Huamachuco was supported by the Social Sciences and Humanities Research Council of Canada and authorized by the Instituto Nacional de Cultura. The research was co-directed by Theresa Lange Topic. This paper has benefited greatly from participation at a Dumbarton Oaks summer seminar that allowed me to read widely and discuss these ideas with the other participants: John V. Murra, Franklin Pease G. Y., Charles Hastings, and Inge Schjellerup; I thank both Dumbarton Oaks and the participants in the seminar. John H. Rowe and Maria Rostworowski have also always been very generous in providing advice and insight on matters of Inca ethnohistory. Dan Julien, Luis Millones, and two anonymous reviewers provided very useful copunents during the review process. At Trent I would like to thank Lisa Rankin and the late Otto Roesch for advice and comments. Theresa Lange Topic has played a critical role in the preparation of this paper.

References Cited
Acosta, Padre Joseph de 1940 [1590] Historia natural y moral de las Indias, edited by Edmundo O'Gorman. Mexico and Buenos Aires: Fondo de Cultura Economica. Albomoz, Cristobal de 1967 [ca. 1582] Un inedit de Cristobal de Albomoz: La instruccion para descubrir todas las guacas del Pin'i.con sus camayos y haziendas, edited by Pierre Duviols. Journal de la Societe des Americanistes 56(1):7-39. Paris. Arriaga, Pablo Joseph de 1968 [1621] La extirpacion de idolatrfa en el Pen'i.. In Cronicas Peruanas de Interes Indigena, edited by Francisco Esteve Barba, pp. 191-277. Biblioteca de Autores Espafioles 209. Madrid: Atlas. . Barth, Fredrik 1969 Introduction. Ethnic Groups and Boundaries: the Social Organization of Culture Difference, edited by Fredrik Barth, pp. 9-38. London: Allen and Unwin.

3. Functionally. Even though imposed and artificial, the Inca promotion of ethnicity served the needs of the indigenous groups as well as the goals of the Inca overlords; this is why Inca policy was successful. The local people gained increased access to different ecological zones, averaging of risk over larger areas and populations, and participation in a wider social universe. Local elites had their status confirmed and authority legitimated. The Inca gained an administrative tool that appeared to honor, indeed enforce, local tradition, while in fact al-

ANDEAN PAST 5 (1998)


Bennett, Wendell C.
.

-122
pp. 35-56. Washington, D.C.: Archaeology Section!American Anthropological Association. Gailey, Christine W. and Thomas C. Patterson 1987 Power Relations and State Formation. In Power Relations and State Formation, edited by Thomas C. Patterson and Christine W. Gailey, pp. 1-26. Washington, D.C.: Archaeology Section! American Anthropological Association. Garcilaso de la Vega, EI Inca 1966 [1609] Royal Commentaries of the Incas" translated by Harold V. Livermore. Austin: University of Texas Press. Grieder, Terence 1978 The Art and Archaeology of Pashash. Austin: University of Texas Press. Grieder, Terence and Alberto Bueno Mendoza 1985 Ceremonial Architecture at La Galgada. In Early Ceremonial Architecture in the Andes, edited by Christopher B. Donnan, pp. 93-109. Washington, D.C.: Dumbarton Oaks. Guaman Poma de Ayala, Felipe 1980 [1615] EI Primer nueva coronica y buen gobierno, edited by John V. Murra and Rolena Adorno, with Jorge L. Urioste. Mexico: Siglo Veintiuno editores. Hampe M., Teodoro 1979 Relacion de los encomenderos y repartimientos del Peru en 1561. Historia y Cultura 12:75117. Lima. Hill, Ann Maxwell 1992 Chinese Funerals and Chinese Ethnicity in Chiang Mai, Thailand. Ethnology 31:315-330. Isbell, William H. 1991 Honcopampa: Monumental Ruins in Peru's North Highlands. Expedition 33(3):27-36. Philadelphia: The University Museum, University of Pennsylvania. Julien, Daniel G. . 1988 Ancient Cuismancu: Settlement and Cultural Dynamics in the Cajamarca Region of the North Highlands, Peru, 200 BC to AD 1532. Unpublished Ph.D. dissertation. Austin: University of Texas. Krzanowski, Andrzej 1986 The Cultural Chronology of Northern Andes of Peru (The Huamachuco-Quiruvilca-Otuzco Region). Acta Archaeologica Carpathica 25:231264. Krzanowski, Andrzej and Jan Szeminski 1978 La toponomia indigena en la cuenca del Rio Chicama (Peru). Estudios Latinoamericanos 4: 1151. Warsaw: Academy of Sciences of Poland, Institute of History. Kroeber, A. L. 1950 A Local Style of Lifelike Sculptured Stone Heads in Ancient Peru. In Beitriige zur Gesellungs- und Volkerwissenschaft: Professor Dr. Richard Thurnwald zu seinem achtzigsten Geburtstag gewidmet, edited by Ilse T1:)nnies, pp. 195-98. Berlin: Verlag Gebr. Mann. Lee, B. T. 1926 Encomenderos y encomiendas. Pueblos de Mambamarca, Pumamarca, Chonda, Ichinga, y

1944 The North Highlandsof Peru. Anthropologi-

cal Papers of the American Museum of Natural History 39(1). New York. Betanzos, Juan de 1987 [1557] Suma y narracion de los Incas, edited by Maria del Carmen Martin Rubio. Madrid: Atlas. Bolton, Ralph 1979 Who are the Qolla? A Note of Clarification. EI Dorado 4(3): 49-58. Greeley, Colorado. Burger, Richard L. and Lucy Salazar Burger 1980 Ritual and Religion at Huaricoto. Archae-

ology 33(6):26-32.

Calancha, Fray Antonio de la 1974-81 [1638] Coronica moralizada del orden de San Augustin en el Peru . . ., edited by Ignacio Prado Pastor. Lima: Imprenta de la Universidad Nacional Mayor de San Marcos. 6 volumes. Cantos, Miguel de 1965 [1581] Relacion para la real audiencia de los repartimientos y numero de indios y encomenderos que hay en el corregimiento de ChinbO. Relaciones Geograflcas de Indias, Peru 2, pp. 248-253, edited by Marcos Jimenez de la Espada. Biblioteca de Autores Espafioles 184. Madrid:.Hijos de M.G. Hernandez. Castro de Trelles, Lucila 1992 Relacion de la religion y ritos del Peru hecha por los padres agustinos. Lima: Pontificia Universidad Catolica del Peru. Cieza de Leon, Pedro de 1984 [1553] Cr6nica,del Peru: Primera parte, edited by Franklin Pease G. y. Lima: Pontificia Universidad Catolica del Peru. Clark, Sheila Seiler 1994 Ethnicity Embodied: Evidence from Tahiti. Ethnology 33:211-228. Cook, Noble David 1976-1977 La visita de los Conchucos por Cristobal Ponce de Leon, 1543. Historia y Cultura 10:2345. Lima: Instituto Nacional de Cultura. Daggett, Richard E. 1983 Megalithic Sites of the Nepefia Valley, Peru. In Investigations of the Andean Past, edited by Daniel H. Sandweiss, pp. 75-97. Ithaca, New York: Cornell University Latin American Studies Program. Duviols, Pierre 1973 Huari y Llacuaz: Agricultores y pastores. Un dualismo prehispanico de oposicion y complementaridad. Revista del Museo Nacional39. Lima. Espinoza Soriano, Waldemar 1971 Geografia historica de Huamachuco. HistOriay Cultura 5:5-96. Lima. 1974 Los Sefiorios etnicos del valle de Condebamba y provincia de Cajabamba. Anales Cientijicos de la Universidad del Centro del Peru 3:11-371. Huancayo. Gailey, Christine W. 1987 Culture Wars: Resistance to State Formation. In: Power Relations and State Formation, edited by Thomas C. Patterson and Christine W. Gailey,
.

123Cuysmango, en la provincia de Caxamarca, encomendados en Melchor Verdugo. 1535. Revista del Archivo NacionaI4(l): 1-21. Lima. Loredo, Rafael 1958 Los Repartos: Bocetos para la nueva historia del Peru. Lima: Libreria y Imprenta D. Miranda. Mahmood, Cynthia K. and Sharon L. Armstrong 1992 Do Ethnic Groups Exist? A Cognitive Perspective on the Concepts of Culture. Ethnology 31:1-14. Martinez Compaii6n, Baltasar Jaime 1978-85 [1789] Trujillo del Peru, Siglo XVIII. Madrid: Ediciones Cultura Hispanica del Centro Iberoamericano de Cooperaci6n. McCown, Theodore 1945 Pre-Incaic Huamachuco: Survey and Exavalions in the Region of Huamachuco and Cajabamba. University of California Publications in American Archaeology and Ethnology 39. Berkeley and Los Angeles: University of California Press Molina, Crist6val, tIelCusquefio" 1943 [ca. 1575] Relaci6n de las fAbulasy ritos de los Incas. In Las Cronicas de Los Mo/inas. Los pequefios grandes libros de historia americana, First Series 4, pp. 1-101, second pagination. Lima: Imprenta D. Miranda. Murra, John V. 1975 Formaciones economicas y politicas del mundo andino. Lima: Instituto de Estudios Peruanos. 1980 The Economic Organization of the Inca State. Research in Economic Anthropology, Supplement 1. Greenwich, Connecticut: JAI Press. 1982 The Mit'a Obligations of Ethnic Groups to the Inca State. In The Inca and Aztec States, 14001800.: Anthropology and History, edited by George A. Collier, Renato I. Rosaldo, and John D. Wirth, pp. 237-262. New York: Academic Press. Patterson, Thomas C. 1987 Tribes, Chiefdoms, and Kingdoms in the Inca . Empire. In Power Relations and State Formation, edited by Thomas C. Patterson and Christine W. Gailey, pp. 117-127. Washington, D.C.: Archaeology Section!American Anthropological Association. Pease G. Y., Franklin 1982 The Formation of Tawantinsuyu: Mechanisms of Colonization and Relationship with Ethnic Groups. In The Inca and Aztec States, 1400-1800: Anthropology and History, edited by George A. Collier, Renato I. Rosaldo, and John D~Wirth, pp. 173-198. New York: Academic Press. Pizarro, Pedro 1978 [1571] Relacion del descubrimiento y conquista de los reinos del Peru, edited by Guillermo Lohmann Villena. Lima: Pontificia Universidad Cat6lica del Peru. Polo de Ondegardo, Juan 1940 [1561] Informe del Licenciado Juan Polo de On.degardo al Licenciado Briviesca de Mufiatones sobre la perpetuidad de las encomiendas en el Peru. Revista Historica 13:125-196. Lima.

Topic: Ethnogenesis in Huamachuco


Porras Barrenechea, Raul 1978 Pizarro. Lima: Editorial Pizarro S. A. Pozorski, Shelia 1987 Theocracy vs. Militarism: the Significance of the Casma Valley in Understanding Early State Formation. In The Origins and Development of the Andean State, edited by Jonathan Haas, Shelia Pozorski, and Thomas Pozorski, pp. 15-30. Cambridge: Cambridge University Press. Proulx, Donald A. 1983 Territoriality in the Early Intermediate Period. Nawpa Pacha 20:83-96. Berkeley: Institute of Andean Studies. Rasmussen, Susan

1992 Disputed Boundaries: Taureg Discourse on


Class and Ethnicity. Ethnology 31:351-365. Reichlen, Henri and Paule Reichlen 1949 Recherches archeologiques dans les Andes de Cajamarca: Premier Rapport de la Mission Ethnologique Franyaise au Perou septentrional. Journal de la Societe des Americanistes 38:137-174. Reichlen, Henri and Paule Reichlen 1950 Recherches archeologiques dans les Andes de Haut Utcubamba: Deuxieme Rapport de la Mission Ethnologique Franyaise au Perou septentrional. Journal de la Societe des Americanistes 39:219246. Rivet, Paul 1949 Les langues de l'ancien diocese de Trujillo. Journal de la Societe des Americanistes 38:1-52. Rostworowski de Diez Canseco, Maria 1975 Pescadores, artesanos, y mercaderes costefios. en el Peru prehispanico. Revista del Museo NacionaI41:311-349 Lima. 1989a Dona Francisca Pizarro: Una ilustre mestiza 1534-1598. Lima: Instituto de Estudios Peruanos. 1989b Introducci6n. In Costa peruana prehispanica, pp. 15-18. Lima: Instituto.de Estudios Peruanos. 1990 Las macroetnias en el ambito andino. AI/panchis 22(35/36):3-28. Cusco: Instituto de Pastoral Andina. Rowe, John H. 1946 Inca Culture at the Time of the Spanish Conquest. In: The Handbook of South American Indians 2, edited by Julian H. Steward, pp. 183-330. Washington, D.C.: Smithsonian Institution, Bureau of American Ethnology, Bulletin 143. 1948 On Basic Highland Culture. In A Reappraisal of Peruvian Archaeology, edited by Wendell C. Bennett, p. 20. Menasha, Wisconsin: The Society for American Archaeology and the Institute of Andean Research. 1982 Inca Policies and Institutions Relating to the Cultural Unification of the Empire. In The Inca and Aztec States, 1400-1800: Anthropology and History, edited by George A. Collier, Renato I. Rosaldo, and John D. Wirth, pp. 93-118. New York: Academic Press. San Pedro, Fray Juan de 1992 [1560] La persecuci6n del demonio: Cronica de los primeros agustinos en el norte del PerU(J560), transcribed by Eric E. Deeds. MAlagaand Mexico:
.

ANDEAN PAST 5 (1998)


Editorial Algazara and Centro Andino y Mesoamericano de Estudios Interdisciplinarios. Santo Tomas, Domingo 1560 Lexicon, 0 vocabulario de la lengua general del Perv, copuesto por el Maestro F Domingo de S. Thomas de la orden de S. Domingo. Valladolid: Francisco Fernandez de Cordoba. Sanniento de Gamboa, Pedro 1907 [1572] The History of the Incas, translated by Sir Clements Markham. Cambridge, England: The Halduyt Society. Schaedel, Richard P. 1952 An Analysis of Central Andean Stone Sculpture. New Haven, Connecticut: Ph.D. dissertation, Yale University. Shady Solis, Ruth and Hennilio Rosas la Noire 1976 Enterramientos en chu//pas de Chota (Cajamarca). Lima: Publicaciones del Museo Nacional de Antropologia y Arqueologia, Serie: Investigaci6n del Campo 1. Silva Santisteban, Fernando 1985 Cajamarca en la historia del Peru. In Historia de Cajamarca: 1, Arqueologia, edited by Fernando Silva Santisteban, Waldemar Espinoza Soriano, and Rogger Ravines. Cajamarca, PI>.9-27: Instituto. Nacional de Cultura, Filial de CajamarcalCorporaci6n de Desarrollo de Cajamarca. 1986a El reino de Cuismanco. In Historia de Cajamarca: 2, Etnohistoria y /ingiiistica, edited by Fernando Silva Santisteban, Waldemar Espinoza Soriano, and Rogger Ravines, pp. 15-33. Cajamarca: Instituto Nacional de Cultura, Filial de CajamarcalCorporaci6n de Desarrollo de Cajamarca. 1986b La lengua culle de Cajamarca. In Historia de Cajamarca: 2, Etnohistoria y lingiiistica, edited by Fern~do Silva Santisteban, Waldemar Espinoza Soriano, and Rogger Ravines, pp. 365-369. Cajamarca: Instituto Nacional de Cultura, Filial de CajamarcalCorporaci6n de Desarrollo de Cajamarca. Taylor, Gerald, editor 1987 Ritos y tradiciones de Huarochiri del siglo XVI/. Lima: Instituto de Estudios Peruanos and Instituto Frances de Estudios Andinos. Tello, Julio C. . 1929 Antiguo Peru: Primera epoca. Lima: Comisi6n Organizadora Del Segundo Congreso Sudamericano de Turismo. Terada, Kazuo . 1979 Excavations at La Pampa in the North Highlands of Peru, 1975: Report No.1 of the Japanese Scientific Expedition to Nue/ear America. Tokyo: University of Tokyo Press. Terada, Kazuo and Yoshio Onuki 1982 Excavations at Huaca/oma in the Cajamarca Valley, Peru, 1979. Tokyo: University of Tokyo Press. Topic, John R. 1986 A Sequence of Monumental Architecture from Huamachuco. In Perspectives on Andean Prehistory and Protohistory, edited by Daniel H. Sandweiss and D. Peter Kvietok, pp. 63-83. Ithaca: Latin American Studies Program, Cornell University.

-124
1991 Huari and Huamachuco. In Huari Administrative Structure: Prehistoric Monumental Architecture and State Government, edited by William H. Isbell and Gordon F. McEwan, pp. 141-164. Washington, D.C.: Dumbarton Oaks. 1992 Las huacas de Huamachuco: precisiones en torno a una imagen indigena de un paisaje andino. In La persecucion del demonio: cronica de los primeros agustinos en el norte del Peru (1560), by Fray Juan de San Pedro, pp. 41-93. M{llaga and' Mexico: Editorial Algazara and Centro Andino y Mesoamericano de Estudios Interdisciplinarios. 1994 El izamiento del gallardete en Huamachuco. In En el nombre del Sefior: shamanes, demonios, y curanderos del norte del PerU, edited by Luis Millones y Moises Lemlij, pp. 102-127. Lima: Biblioteca Peruana de Psicoanalisis. Topic, John R. and Theresa Lange Topic 1985 EI horizonte medio en Huamachuco. Revista del Museo Nacional, Volume 47:13-52. Lima. 1992 The Rise and Decline of Cerro Amaru: An Andean Shrine During the Early Intennediate Period and Middle. Horizon. In Ancient Images, Ancient Thought: The Archaeology of Ideology, edited by A. Sean Goldsmith, sandra Garvie, David Selin, and Jeanette Smith, pp. 167-170. Calgary: The University of Calgary Archaeological Association. 1993 A Summary of the' Inca Occupation of Huamachuco. In Provincial Inca: Archaeological and Ethnohistorical Assessment of the Impact of the Inca State, edited by Michael A. Malpass, pp. 1743. Iowa City: University ofIowa Press. Topic, Theresa L. and John R. Topic 1984 Huamachuco Archaeological Project: Preliminary Report on the Third Season. June-August 1983. Peterborough, Ontario: Trent University Occasional Papers in Anthropology 1. Department of Anthropology, Trent University. Torero, Alfredo 1989 Areas toponfmicas e idiomas en la sierra norte peruana. Un trabajo de recuperaci6n linguistica. Revista Andina 7(1):217-257. Cusco. Urteaga, Horacio H. 1942 Algunas provisiones de Pizarro sobre encomiendas. Provisi6n de Don Francisco Pizarro gobernador del Peru, Afios 1534-1540. Revista del ArchivoNacional 15(1): 129-216. Lima. Urton, Gary 1990 The History of a Myth: Pacariqtambo and the Origin of the Incas. Austin: University of Texas Press. Vail, Leroy 1989 Introduction: Ethnicity in Southern African History. In The Creation of Tribalism in Southern Africa, edited by Leroy Vail, pp. 1-19. London: James Currey, and Berkeley and Los Angeles: University of California Press. Willey, Gordon R. 1945 Horizon Styles and Pottery Traditions in Peruvian Archaeology. American Antiquity 11(2):4956.

125Willey, Gordon R. and Philip Phillips 1958 Methodand Theoryin AmericanArchaeology. Chicago: University of Chicago Press. Wilson, David 1988 Prehistoric Settlement Patterns in the Lower Santa Valley, Peru: A Regional Perspective on the Origins and Development o/Complex North Coast Society. Washington, D.C.: Smithsonian Institution Press. Zevallos Quifiones, Jorge 1948 Primitivas lenguas de la costa. Revista del Museo NacionaI17:114-119. Lima.

Topic: Ethnogenesis in Huamachuco

ANDEAN PAST 5 (1998)

-126

50km I

Figure 1. The province of Huamachuco. The approximate locations of the four indigenous guarangas and the guaranga of Chaupiyungas are shown. Place names associated with the creation myth are indicated. Large letters indicate the locations of some of the principal huacas: A = vlpillo, B = nomadoy, C = pomacama, and D = quimgachugo.

127-

Topic: Ethnogenesis in Huamachuco

Figure 2. Northern Peru, showing some of the Incaic provinces and modem towns mentioned in the text.

COCA PRODUCTION ON THE INCA FRONTIER: THE YUNGAS OF CHUQUIOMA

Catherine J. Julien
Western Michigan University Aspects of Inca economy can be clarified by examining documentation for specific productive regimes during the years immediately following the Spanish arrival. Administrative documents reflect the changes which had taken place due to the transformation of the political context in which these activities occurred as well as to the entrepreneurial activity of Spaniards. However, to the degree that production in place during the later years of the Inca empire had some continued rationale in the new economy, what developed was a natural evolution from the earlier period. In the caseof cocaproduction,the leaf continued to be grown in the same regions where the Incas had organized production, and by the same producers. Images of the Inca organization of coca production can be developed from' administrative documentation. The importance of coca to the Incas and to Andean people in general is well-known to scholars (Murra 1991b:565-566). The Incas organized coca production in various regions in the eastern Andean lowlands as well as in some Pacific coastal valleys (Espinoza Soriano 1963; Rostworowski 1988). One Spanish administrator noted that the Incas controlled access to coca, prohibiting its use among commoners and reserving it for those of noble birth and the army (Matienzo 1967 [1567]: Chapter 44, p. 163), but the ability of the Incas to restrict the use of such an important substance has been seriously questioned (Parkerson 1984). Indeed, coca is an important element in everyday life in Andean communities today (Allen 1988; America Indfgena 1978; Pacini and Franquemont 1986; Instituto Indigenista lnteramericano 1989), despite a long history of efforts to curtail its use (Romano 1982; Parkerson 1984; Lyon 1994b). While we may yet develop ethnohistorical approaches that trace patterns of consumption during the later Inca empire, the documentation we have lends itself more easily to recovering aspects of the organization of production. . Because of the survival of documentation for the yungas of Chuquioma, a lowland region on the eastern Andean slopes near modem Cochabamba (Figures 1 and 2), tracing Inca organization of coca production in this area is a fruitful exercise. Coca production organized by the Incas continued after the Spanish entry into the region in 1538, and expanded markedly after the silver strike at Potosi in 1545, as demand for coca grew steadily. In these years, peoples who were providing their encomenderos with coca as tribute greatly increased the supply. Some of the documents on which this study is based, delivery records for the coca harvested by the tributaries of Sacaca during the years 15481552 are reproduced here from a lawsuit with their encomendero over the excessive tribute they paid in this period (see Document 2 in the Appendix). Even though limits to the amount an encomendero could collect were not set until the tasa (tax) of 1550, it was still possi': ble to demonstrate that the Sacaca tributaries had paid an excessive amount of coca as tribute before limits were set. Early documentation is scarce, but by the 1560s notarial records demonstrate that coca chacras (fields) were being bought and sold between Indians and Spaniards and between Spaniards and other Spaniards. A settlement of Spaniards at Chuquioma, a lowland valley northeast of Pocona (Figures 1 and 2) was noted (Meruvia 1991:204-207). By 1561, a church had been built (ANB, Escrituras Publicas 2, folio lOv). In the 1560s and 1570s, Spanish merchandise was taken there for resale (Meruvia 1991, p. 207). A will, made in 1573, by a freed slave who had been married in the church to a Huanca woman, records the coca chacras this pair owned and worked (ANB, EP24, folios 161-163v). Increases in production, the establishment of Spanish set-

ANDEAN PAST 5 (1998):129-160.

ANDEAN PAST 5 (1998) tlements and trade, the entry of entrepreneurs who developed coca plantings alongside the existing fields, and the shipment of coca to markets at La Plata and Potosi all transformed the lowland production zone. Still, alongside this entrepreneurial economic development, diverse highland peoples worked coca chacras in the yungas of Chuquioma following a productive organization established by the Incas. Under this regime, small groups of people (camayos) from highland provinces resided permanently in the lowlands, being joined periodically by larger groups of temporary workers from their province of origin (mitayos) who came down for harvest. Coca was gathered three to four times a year. The harvesters transported the coca on their backs out of the lowlands to a delivery point in the highlands, from which it was transported by animals to its destination. Although the ultimate destination of the coca had clearly changed since Inca times, the' organization of production, at least to the delivery point, reflected "In~apractice. Inca practice, of course, may have incorporated ~arlier modes of organizing production. Not only the local substrate, but differences arising from the history of contact between the Incas and local populations, the goals of the individuals involved in restructuring productive activities at the provincial level, as well as locational factors, such as proximity to the frontier, may have affected the particular form of a productive regime. Using the documentation created in the early decades of Spanish presence to define earlier productive regimes is a complex ethnohistorical problem. In the yungas, people from the higWands tended and harvested coca plantations through the middle of the 17th century. An administrative survey (visita) of Pocona in 1556 with information about the coca chacras worked by the people of Pocona was published in 1970 (Ramirez 1970), but has not generated much scholarly interest. For example, no effort has been made to use the visita as the basis for archaeological survey. The documents, though they contain excellent information about the

-130 foreigners introduced into the region during the later Inca empire (mitimas, see Rowe 1982:105-107), do not mention the monumental archaeological site of Incallacta, very near to Pocona, and the central focus of archaeological attention up to the present (Ellef-, sen 1967; Nordenskiold 1915; Lee 1998). Here we have a classic case of noncorrespondence between archaeology and ethnohistory at the evidentiary level. A serious effort must be made to generate a spatial and physical analysis from the documents to guide archaeological approaches. Such analysis motivates this study. In the years following the publication of the Pocona visita, a body of documentation containing important new information about the. Inca organization of the nearby Cochabamba Valley has been found (Morales 1977; Wachtel 1982; AHMC, Expedientes Coloniales 17, folios 404-462v). Although the connection between Cochabamba and Pocona is made clear in the documents, a regional analysis, linking the two, has yet to be offered. This study frames the productive activities in Ch1;1quioma within the context of the larger regIOn. An infusion of newly identified documentary material makes it possible to make an analysis of the Inca organization of production to the regional level (see Appendix). Besides the Pocona people, other groups paid a portion of their tribute in cocaand were directproducers in the yungas of Chuquioma. These documents pertain to the coca tribute paid by the people of Sacaca in the Charcas highlands (Appendix, Documents 1 and 2), as well as Pocona (Document 3). After establishing who was involved in producing coca in Chuquioma, the focus of this article will shift to the Inca organization there. Coca production was, only part of a major Inca project in the region. The documentation presented here is earlier than the visitas of Sonqo, in the yungas north of La paz (1568-1570) (Golte 1970; Murra 1991a; Angelis 1995). Some organizational similarities have been noted here, but no systematic comparison is offered. Cusco is an obvious lacuna in our analysis (but see Lyon 1994a, 1994b, 1995). Developing regional

131analyses for the Sonqo and Cusco cases promises to be a complex ethnohistorical undertaking indeed. By expanding our ethnohistorical focus beyond a single region, we will avoid the temptation to consider a particular case as a model of Inca organization. Rather, regional analyses will eventually make meaningful generalization possible (Julien 1993). However, even now, after we have argued that we should proceed from the particular to the general, we must begin with generalization. Perhaps the only way we have of guiding our layered reading of the written sources is through the use of generalizations about Inca administration recorded by Spanish administrators of the period who knew a great deal about it, but described it in vague terms. Our entree into the subject, then, is the report of Juan de Matienzo, who wrote about the Inca-organization of coca production. Matienzo's Report
.

Julien: Coca on the Inca Frontier Matienzo had learned about coca production under the Incas and explained which sectors of the colonial industry were derived from it:
Many caciques have chacras of coca in the Andes of Tono [east of Cusco], and also in this province [Charcas], but there is a difference, because all of them who had and have to give coca as tribute, had chacras of their own dating well into the past, and this is general in all the eastern tropical valleys. These fields do little harm because they are worked at the same time as the fields from which the tasa is paid, and the same was done under the Incas, because these chacras were not affected, and they have camayos who have been there for a long time who came from the territory of the caciques, and they have houses and the equipment they need for cultivation, and although they may have planted a few more fields, it does not matter, but the chacras planted by caciques who do not pay coca as tribute and never have are a source of great harm to the Indians who work them. , ,I

Matienzo,ajudge (oidor)of the Audiencia

of Charcas, wrote' a comprehensive report to suggest improvements to colonial administration (1967 [1567]). Having served since the founding of the Audiencia in 1561, he was familiar' with the productive organization in the yungas of Chuquioma, the nearest cocagrowing region to the seat of the Audiencia in La Plata. However, at the time he wrote his report, he had just returned from conducting the administrative review (residencia) of Dr. Gregorio Gonzalez Cuenca who had completed his tenure as administrator (corregidor) of Cusco. A considerable portion of the documentation assembled by Matienzo, and by Cuenca in his own defense, had to do with coca production in the province of the Andes, in the yungas east of Cusco. Moreover, Matienzo had travelled into the lowlands as an official envoy to Titu Cusi, the Inca who resisted Spanish control from Vileabamba. Matienzo was well-informed about coca production in Cusco and makes specific references to it in his report. Most of his statements, however, are generalizations based on what he knew from experience in both Charcas and Cusco.

Clearly, fields had been cultivated for the curacas prior to the Spanish arrival. We can infer from Matienzo' s statement that the fields cultivated to pay the tribute assessed by the Spaniards (tasa) were a separate category, and that these fields had also been cultivated for tribute in Inca times. Although Matienzo does not indicate the earlier beneficiary of the tasa fields in the above passage, elsewhere he notes that the camayos assigned-to the chacras from which the tasa is paid were put there by the Incas (1967 [1567]:Chapter 50, pages 177180). We can infer that the beneficiary of these chacras was designated by the Incas.
1Matienzo 1967 [1567]:Chapter49, pages 176-177: Mucbos caciques denen chacaras de coca en los Andes de Tono, y tambien en esta provincia, pero hay diferencia, porque todos los que tuvieron y tienen coca de tasa, tuvieron sus chacaras aparte antiquisimas, 10 cual es general en todos los Andes, y estas son de poco perjuicio, porque se benefician y coxen cuando las chacaras de la tasa, y 10 mesmo se bacia en tiempo del Inga, porque a sus chacaras no se tocaba en ninguna manera, y allende de esto tienen camayos antiguos de su mesma tierra, y casas y aparexos para el beneficio; y aunque hayan plantado alguna mas, no importa. Pero las chacaras que han puesto caciques que no tienen ni tuvieron coca de tasa, son de gran perjuicio para sus indios.

ANDEAN PAST 5 (1998) The regime Matienzo describes had no Spanish precedent. Two types of labor were involved, as noted above. Camayos, who Matienzo describes as "yanaconas who continually reside in the chacras", were assigned three principal tasks: they guarded the fields; they produced the mats needed for drying the coca and the material for packing it; and they packed coca for transport (1967 [1567]: Chapter 46, p. 170; Chapter 50, pp. 177-178). The individuals who served as camayos lived produced, accompanied by their wives and families. When a camayo died, the office was inherited by a descendant or another person was designated from his territory of origin (Rowe 1982:96-97, 102-104). Each camayo had a chacra of coca, and with this coca, he was able to trade for maize, clothing, and other highland products. Some camayos also had chacras of maize in their territory of origin and arranged for their cultivation in exchange for coca (MatienZo 1967 [1567]: Chapter 50, pp. 177-180). Mitayos, on the other hand, resided in the highland territory. They came to the yungas to pick coca leaf; to work the soil around the plants after harvest, leaving it clean; and to transport the coca, usually on their own backs, to a designated point in the highlands. Three harvests a year, one every four months, were possible (Matienzo 1967 [1567]:Chapter 45, p. 167). While the camayos were permanently assigned to their tasks, the mitayos may have served by turns, as the name mitayo indicates.2 Because the task was done intermittently, the use of the term may refer not to the rotation of a particular group of laborers but to the periodicity of the task. Each harvest was called a mita (ibid.:Chapter 45, p. 167). We know very little about differences in status that were attached to particular assignments, but we can posit that the camayos had higher status than the mitayos. The responsibilities of the camayos indicate a privileged position. Also, Matienzo noted that finding people willing to serve as camayos was not a problem; these individuals had some hopes of
2 Mittani

-132

their entire lives in the zone where coca was

becoming wealthy (Matienzo 1967 [1567]: Chapter 50, pp. 177-180). Although we cannot assume that the fields they worked for themselves had been assigned to them by the Incas, their situation may parallel that of yanaconas in Yucay who worked fields for the Incas, but also had parcels for their own use (Villanueva 1970:42-43). However, even if the assets they held were the same under both Inca and Spanish rule, demands on their time or their freedom of operation may have been substantially different. Still, we can probably conclude that they enjoyed a higher status than the people who served as harvest labor and initial porters. Because we know what basic tasks were carried out by the two types of laborers, we can reconstruct in broad outline the round of activities associated with the coca harvest. Prior to harvest, the camayos had to gather the cane and other plant materials used in packing the coca. Where these materials were not readily at hand, the camayos planted chacras to supply them. When the mitayos arrived, they harvested the coca leaf. It was then laid out on mats for drying. While it dried, the mitayos worked the earth around the coca plants, leaving the soil free of debris. On occasion fields were not harvested; the leaf was left to fall, and the decayed leaf served as fertilizer. These fields produced higher yields afterward (Matienzo 1967 [1567]:Chapter 45, p. 168). When the coca was ready, the camayos packed it. Some type of weighing and counting procedure was carried out at this time. Then the coca was distributed to the mitayos for transport to the highlands.

Optimally, the coca was green and dry when packed; if it was left in the sun too long or got wet it turned brown and lost considerable value. Matienzo also noted that coca was transported to the highlands as soon as possible. In a cold climate, it could be kept for some time, although it was no longer considered fresh after a year (Matienzo 1967 [1567]: Chapter 45, p. 167). The rainy and often humid climate of the yungas may have resulted in delays in the drying of coca, postponing the departure of mitayos. Once packed, however,

guin 1952[1608]:243).

- caberle su vez de hazer algo (Gonzalez Hol-

133transport of the coca probably took place immediately. While. Matienzo does not tell us much about the nature of the yungas settlements where coca cultivation occurred, we can extrapolate an image of this type of settlement from what he wrote and from the Chuquioma documents. Settlements in the yungas had small permanent populations, swollen at intervals by a relatively larger number of harvesters. The structures required to facilitate the activities at harvest would have been relatively modest. Because of the dependably inclement weather, some type of housing was probably provided. Huts (buhfos) are sometimes mentioned in notarial documents (ANB, EP 24, folio 276v). The drying of coca also had spatial requirements, and drying areas may have been at least partially roofed. Because coca appears to have been transported soon after it was packed, we can expect little in the way of storage facilities in the yungas where it was grown. Also, there is no evidence that animals were importantin the transferof cocafrom the fields, so there may have been no corrals or other constructions rdated to their presence. Our image of these settlements derives from the documentation we have for the organization of coca production in the yungas of Chuquioma. The Yungas of Chuquioma Pocona is a modern town, located at 3100 meters above sea level (masl) about 125 km east-southeast of Cochabamba (Figures 1 and 2). The town was the center of a colonial parish, and it was the principal place of residence of the caciques who formed part of an encomienda award called Pocona. The town was on the main road from Cochabamba to the east. To reach the yungas of Chuquioma, a traveller would have continued along the same road to Totora, another colonial reduction, then on to Tiraque Chico before beginning the steep descent into the yungas (Figure 2) Everywhere, the eastern slopes of the Andes drop sharply into the lowland~, but, according to Tadeo Haenke, who travelled the

Julien: Coca on the Inca Frontier courses of several rivers in the region .in the late eighteenth century, nowhere is the drop as sharp as in the region northeast of Cochabamba:
The drop of this cordillera and its rivers to the plains is so short, so vertical and so precipitous, that ordinarily a descent along a straight line of 5 or 6 leagues [25 to 30 kilometers] covers almost the entire altitudinal distance, through a landscape of precipices and horrifying slopes; and in particular, this descent occurs over a shorter horizontal distance in the cordillera of the Yuracarees than anywhere else.3

He uses the term "cordillera of the Yuracarees" to refer the mountain chain that was also called the Serrania of Cocapata in the same period. When Haenke visited the area, a number of missions had just been established among peoples known as Yuracarees. An itinerary of one of the early Franciscan expeditions into the region traces a course from the Franciscan College at Tarata, near' Cochabamba, into the lowlands via Tiraque (not Tiraque Chico), down the San Mateo River, then along various rivers, including the Chimore and the Rio Blanco, to the east of Chuquioma, before turning west and returning via Chuquioma and Tiraque Chico (Figure 2) (RAH, Mata Linares, 9/1666, folios 264-282v).. The new missions were located in the lowlands below the valleys where coca was grown and were accessed from Tiraque (Viedma 1836:38-42). References to Yuracarees and others are found in the documentation from the late sixteenth century onward. These peoples had not formed part of the Inca empire and remained beyond the control of an Andean state until the late eighteenth century when the Franciscans initiated their incorporation, in the missioni3 RAH, Mata Linares, 9/1667, folios 267v-268r: La caida de esta cordillera, y de sus rios asia a [sic] los planes de los Andes, es tan corta, tan empinada, y tan precipitada, que ordinariamente una bajada en distancia recta de 5 0 6 leguas, comprehende casi toda esta altura en un conjunto de precipicios y laderas horrorosas; y en particular en la cordillera de Yoracarees [sic Yuracarees], es esta distancia mas corta que en ninguna parte.

ANDEAN PAST 5 (1998) zation effort noted above. From time to time there were hostilities between groups identified as Yuracarees and the residents of the upper valleys where coca was produced (AGI, Charcas 25, folio I; AGI, Lima 166, folios 20v-26v.; Viedma 1836:43-54). The yungas of Chuquioma were located on an active fron~ tier during the last years of the Inca empire and the first century of the Spanish colonial era, the period of intensive coca production. In the review which follows, information about coca producers from Pocona will be summarized; then, new material documenting the involvement of cultivators from Sacaca will be presented; afterward, the participation of other groups in the coca economy will be considered.
Pocona cultivators

-134

The tributaries of Pocona paid the bulk of their tribute in coca. Estimates, made in 1548, of what was paid prior to the time the first tasa was set (1550) indicate that between 6,000 and 12,000 cestos (packages) per year went to the encomendero (Table 1).4Here, the
4 In early documents referring to encomiendas, the awards are often identified only through the name of the encomendero. From the 1560s on, each encomienda had a proper name. I have traced the tenure of . the awards and made the identifications in the table. Several of the identifications are not straightforward; both Francisco Pizarro and Vaca de Castro altered the composition of the awards. The names of Toledo's reductions can be located on a map drawn by Thierry Saignes (1986). The more important names appear on Figures 1 and 2. Chayanta. In 1548, Hernando Pizarro held an award of the people subject to the cacique Yucura. In 1561, this grant could be described as half of Chayanta and Chichas. Hernando Pizarro's original award had been much larger. His brother Francisco had assigned him "Consara and Hurinsaya", and although we cannot be certain what that included, we know that Vaca de Castro reformed Hernando Pizarro's grant, regranting segments of it to other Spaniards. The three groups that had been subject to Consara, the principal curaca of Sacaca when the Spaniards first arrived in the area (Espinoza Soriano 1969:149), and that were transferred by Vaca de Castro, were all referred to as Charcas. Rodrigo Pantoja received a group in highland Charcas that was called half of Chayanta in later documents (Hampe Martinez 1979:82). Luis Perdomo received a group of mitimas who produced coca (Loredo 1958:166-168, 1940:52-53, 57-58, 60). The grant did not include

tributaries to pick the coca, according to one source (Loredo 1958:168). Perhaps the mitayos who came to pick the coca were part of the Chayanta grant. Gomez de Luna was also given a group of Charcas who produced coca by Vaca de Castro; they had also been subject to Consara (Loredo 1940:54). The two groups produced a combined total of 1150 costales (Loredo 1958:165, 172). Sacaca. The first recorded holder of Sacaca waS Luis Ribera. His award was described as "being in the province of Consara" (Loredo 1958:164). In 1548, prior to setting the first tasa, this group had been giving their encomendero 700 costales of coca (Loredo 1958:164-165, 1940:53). PoconaiPojo. The first known holder of this award was Pedro de Valdivia (see below, page 140), and it was probably awarded after the expedition to Charcas in 1538. Soon after, Valdivia resigned the award and it was regranted to Pedro Anzures (also called Peranzules, or other variants on this name), founder of the Villa de La Plata (1539), and his brother Gaspar Rodriguez. They also resigned the award and Vaca de Castro (1541-1544) gave it to Diego Centeno, Lope de Mendoza, and Dionisio de Bobadilla (Loredo 1958:154-155, 1940:52-53). Probably because it had been split between several encomenderos, it began to be treated as two separate elJcomiendas from the early 1550s onward: Pocona and pojo (also called Poxo, Pocojo or Pocopoco in the documents) (Hampe Martinez 1979:84). G6mez de Alvarado held Pocona until his death, after which it was transferred to Francisco de Mendoza, the holder in 1556-1557 at the time of the suit (Ramirez Valverde 1970:300). Pojo was held by Pedro Fernandez Paniagua and then by his son Gabriel Paniagua de Loaysa (Hampe Martinez 1979:84). The grant was referred to as Mizque from the 1570s onward (see Figure I). . Cochabamba. We have no information about a Pizarro-era encomendero for either of the Cochabamba grants. If Consara was head of a huno which included what later became Chayanta, Sacaca and the two Cochabamba encomiendas, and this huno corresponded to the Hurinsaya division of a province, then Hernando Pizarro got this entire division when he received "Consara y Hurinsaya" (see above). In 1548, the two Cochabamba awards appear in the records for the first time. One was held by Alonso Camargo (Loredo 1958:162-163, 1940:61), and subsequently, by Polo de Ondegardo (Hampe Martinez 1979:83). At the time of the visita general of Toledo, it was reduced to the town of Santiago del Paso in the Cochabamba Valley (Miranda 1925:141). The other award was held by Rodrigo de Orellana (Loredo 1958:164, 1940:56, 61). It was reduced to Tiquipaya in the Cochabamba Valley during the visita (Miranda 1925:141). Yamparaes, Gualparocas, Cochabi/ca. All o( these groups were located in the vicinity of the city of La Plata. The history of tenure is difficult to reconstruct except in the case ofYamparaes: Pizarro awarded them to Diego de Rojas. Rojas also received the "yngas Gualparocas", a group of orejones (high-status Incas who wore earspools) subject to a cacique named

135estimate is given for both Pocona and Pojo (later known as Mizque), because the two were accounted as a single unit in 1548, a fact that is significant in terms of earlier organization and will be considered again below. From the early 1550s, Pocona and Pojo were treated as distinct encomiendas rather than as a single unit of population. After the split, the tributaries of Pocona owed Gomez de Alvarado, who held the grant between 1550 and 1556, a total of 2,600 cestos per year (Ramirez Valverde 1970:300).5
Gualpa Roca. Vaca de Castro awarded this group to Heman NUfiezde Segura, but possession was disputed by Pablo de Meneses in 1548 (Loredo 1940:53,56-57). Meneses held the award of Yamparaes at this time (Loredo 1958:169-170, 1940:53). The Yamparaes were reduced in Yotala and Quilaquila at the time of the visita general; the Yngas Gualparocas were reduced in Huata. Diego L6pez de Zuniga, deceased, had held the award of Cochabilca, according to one of the 1548 lists (Loredo 1958:169). They were mitimas, and resided ina town by the name of Cochabilca. The award had been diminished by Chiriguana attacks. A reference to five households of "Condesuyos" of Cochavilca appears in a 1540 Pizarro encomienda award document (Del Rio and Presta 1985:235; AGI, Justicia 1125, folio 14). Cochabilca reappears in a listing of 1561, split between two encomenderos who al~o share the grant of Lipes (Hampe Martinez 1979:83). It does not appear in the listings of the 1570s, although it may be listed under the name Condes Arabates; this group was res~ttled in a town called Arabate (Miranda 1975:30, 1925:138). Tarabuco. This group was awarded by Francisco Pizarro to Francisco de Almendras; it was composed entirely of mitimas trom Cusco and the Lake Titicaca region (Del Rio and Presta 1985:224-233; Loredo 1958:170-171, 1940:56, 61). During the visita general, the people of this award were resettled in the towns of Tarabuco and Presto (Miranda 1925:134). Moyosmoyos. The indications that two groups of Moyosmoyos provided tribute are problematic, because they are inconsistently reported even in 1548 when references to them appear in the documents (Loredo 1958:171, 176-177). They are untraceable after that. In only one of these awards was an estimate given for the amount of coca being produced (Christ6bal Pizarro's grant). The documents used to trace tenure, largely trom a 1548 survey authorized by the Governor Pedro de la Gasca, include various references to people who could earn money to pay tribute by hiring themselves out in the coca fields, but only groups who had fields and camayos in the yungas, presumably since the time of the late Inca empire, are included.

Julien: Coca on the Inca Frontier Pocona had coca chacras in the valleys of Chuquioma and Chamorro and took the produce from these fields to Tiraque (Ramirez Valverde 1970:304-305). This Tiraque is the Tiraque Chico that appears on modem maps, distinguishing it from another Tiraque which stood at the head of a pass into the lowlands farther west (Figure 2). Chamorro does not appear on modem maps, but it was accessed by the same pass as Chuquioma and may refer to the Chimore Valley. In notarial documents from the 15608, the area of coca production extended from Chuquioma to the Rio Blanco. The abandonment of the Chamorro fields was ordered at the time of the visita general (AHP, Cajas Reales 18, folio 49v). These fields were

his encomienda. Because he had been awarded an encomienda that paid tribute exclusively in coca, its value fluc~ated more than awards where the tributaries paid all or a substantial part of their tribute in silver. Here is the first reference to cestos. A defmition of the cesto was given in 1597 by Balthasar Ramirez: They have prepared cestos in which to pack [the coca], which are round, of 1 1/3 varas in length and 1/4 [vara] in width; each cesto weighs an arroba [about 11.5 kilos] ; 20 pounds are of coca and five of cesto; they make these cestos of narrow, hollow reeds called' pipo and they make a meshwork with some stiff reeds called pancho; and they cover it with some thick wide leaves that they call

coxoro

...

cinco del cesto. hazen estos cestos de vnas canuelas delgadas . hendidas que llaman pipo y vanlas enrredadondo con vnos bexucos que llaman pancho . y cubrenlascon vnas hojas muy anchas y grQesas que llaman pancho [tachado] coxoro. .. (Trimborn 1936:39) The weight of the cestos was set by the Marques de Cafiete in his 1558 ordinances at 18 pounds of coca (see Murra 1991b:573). In the documents reproduced with this article, the quipocamayos who presented their records in the 1579 lawsuit, noted that the cestos of coca in the 1548-51 period were very large and weighed double what cestos delivered at the time of writing weighed (Document 2, folio 370v). If so; they would have weighed about 23 kilos or 50 pounds.

vn arroba .' las veinte libras son de coca y

Tienen aparejados cestos en que encestalla . los quales son redondos de vara y t(e)r(ci)a de largo y vna quarta de ancho . pesa cada cesto

5 The documentationwe have for Pocona is trom a visita initiated by Francisco de Mendoza, the encomenderoof Poconain 1556,to establishthe valueof

ANDEAN PAST 5 (1998) farther away from Pocona than Chuquioma, perhaps downriver from the Rio Blanco area. Fifty camayos were in charge of the chacras in Chuquioma and Chamorro where the tribute coca was produced. Each mita, 200 mitayos were sent from the highlands to the yungas to harvest coca (Ramirez Valverde 1970:303-304, 306; AHP, Cajas Reales 18, folio 43r, see note 4). The group of camayos was divided in two. Each worked turns of three months: the 25 of the first turn were assigned to drying and packing the coca while the other 25 rested. During the time the mitayos were present, a total of 160 women and 8 to 10 children were required to cook for the mitayos. In the 1550s, these individuals also worked to earn a little money to maintain the mitayos, although the documents do not inform us about what they did (Ramirez Valverde 1970: 305-306). The amount of coca owed as tribute was reduced in 1575, when a new tasa was set following the visita general conducted. by Viceroy Francisco de Toledo. Tribute was assessed in two commodities only: silver and coca. At that time Pocona was a Crown encomienda, and, perhaps for that reason, the document is more than a simple list of what was owed; it specifies how the people of Pocona were to pay. The 899 tributaries of Pocona, after the exemption of four curacas, were to provide the following:
One thousand five hundred [cestos] of coca each year by mita, five hundred cestos each of the three mitas that there are each year, green, dry and in good condition, viewed and inspected, packed and ready to be delivered, of the size and weight that has been customary until now in the payment of tribute, delivered to Tiraque [Chico] at the expense and risk [of the tributaries] using livestock that the community has and have to have in order to take the coca trom the chacras to Tiraque. And they are not to bring it by carrying it themselves. In the coca 434 tributaries are to be occupied, including the camayos who reside where the chacras are, with 128 Indians serving in each of the three mitas that occur each year for a total of 384 Indians. The remaining fifty [of the total of 434] are to serve as the camayos. Item. They are to give 2364 pesos of assayed and marked silver, half on the day of San Juan and half at Christmas, which are to be paid by

-136
364 of the tributaries who are to occupy themselves in earning these pesos, at a rate of 6 pesos per Indian, and they are to put the silver in the community chest of the town ofPocona. And the 97 Indians remaining of the total of 899 tributaries with the help of the older people who are not very incapacitated and the rest of the common people are to cultivate a community chacra of maize, potatoes, beans, and other foodstuffs that grow in their lands for the support of the Indians who cultivate and gather the coca. And anything left [shall be] for the hospitals and the poor of the encomienda. And the Indians who have to pay tasa -- some in coca and some in silver and some in cultivating the community chacra shall be occupied in that one task and are not to be occupied in the others and vice versa -- and they are to be changed according to their turn trom one year to the next to perform these [particular] services. . 6

--

6 AHP, Cajas Reales, 18, folios 43r-43v. . . . mil e quinientos ~estos de coca en cada un afto por sus mitas, quinientas cada mita, de tres que ay en el afio, verde, enjuta y bien acondi~ionada, vistos y calados, en~e5tada e de dar y rre~iuir, del tamafto y peso que hasta aqui los an acostumbrado a dar y pagar de tassa, puest05 en el asiento de Tiraque a su costa e rriesgo en el ganado de la comunidad que tienen y an de tener para traer la dicha coca de las dichas chacaras al dicho asiento de Tiraque, y no la an de traer ellos cargado; para el benefi~io de la qual se an de ocupar, con los camayos que tienen en las dichas chacaras, quatro~ientos e treinta e quatro yndios de los dichos tributarios, mudandose por sus mitas, de tres que ay en el afio, ~iento y veinte e ocho yndios cada mita, que son tre~ientos y ochenta e quatro los yndios que la an de benefi~iar; y los ~inquenta rrestantes a cumplimiento a los dichos quatro~ientos y treinta e quatro an de seruir de los dichos camayos. Yten an de dar dos mil y tre~ientos y sesenta e quatro pesos de la plata ensayada e marcada, mitad por San Juan y mitad por Nauidad, los quales an de pagar tre~ientos y sesenta y quatro yndios de los dichos tributarios que se an de ocupar en ganar los dichos pesos, que sale a rrazon de seis pesos de la dicha plata cada yndio; y los an de poner en la caxa de la comunidad del dicho pueblo prin~ipal de Pocona. Y los nouenta y siete yndios rrestantes, a cumplimiento de los dichos ocho~ientos y nouenta y nueue yndios tributarios, se an de ocupar, con ayuda de los demas uiejos que no fueren muy ympedidos y demas gente del comun, en

137The "Tiraque" in this document has been identified as Tiraque Chico on modern maps. It was the delivery point of the coca paid as tribute. Even though in the above assessment some of the tribute was paid in silver, coca production occupied 60 percent of the tribute population. The worth of 1,500 cestos of coca, estimated at 12,300 pesos (Miranda 1975:28), accounted for more than 72 percent of the tribute income from Pocona. Of the 32 groups assessed in coca in the five colonial districts for which we have data (La Paz, La Plata, Cusco, Huamanga, and Arequipa), no group paid a higher percentage of their tribute in coca (Table 2), and only one group (Sonqo) paid a larger quantity each year (1615 cestos) (Miranda 1975).7 If we rely on tribute assessment as an indicator of economic specialization, Pocona was clearly specialized in the production of coca leaf. Also evident from the tasa document cited above is the attempt to provide the people who worked in the coca with their accustomed diet. A group of 97 triDut~es were to cultivate a community chacra of maize, potatoes, and beans and other foodstuffs for the support of both the camayQSand the mitayos. The location of this chacra is not mentioned, but it was probably somewhere not far from Tiraque Chico at an elevation where the foodstuffs mentioned were commonly grown.
benefi~iar vna chacara de comunidad de maiz, papas y frisoles y otras semillas que se dan en sus tierras, para el sustento de los yndios qucanos [asi], benefi~iar y coxer la dicha coca; y 10 que JTestare,para los ospitales y pobres del dicho JTepartimiento. Y los dichos yndios an de pagar de tassa los vnos la dicha coca y los otros la dicha plata e hazer Iii dicha chacara; y los que se ocuparen vn afio en el uno no se an de ocupar en el otro ni por el contrario; y se an de mudar por sus mitas de afio a afio a ha~er 10susodicho. . . 7 The camayo/mitayo organization, so clear for Pocona, is not evident in the Sonqo documentation. One possibility is that all of the people in the Sonqo award were camayos and that the mitayos had belonged to a group assigned to another encomendero when the award was divided in 1549 (see note 20).

Julien: Coca on the Inca Frontier Cultivating food in the highlands and transporting it to the lowlands was not a requisite for maintaining the community of people who worked in the coca. A sufficiently nutritious diet could have been obtained in the territory where coca was grown. Although we do not have information for the yungas of Chuquioma, a La Paz bishop noted what was locally grown in the yungas of La Paz at the beginning of the seventeenth century :
These people have no occupation other than cultivating this plant [coca] because the ruggedness of the mountains inhibits the cultivation of maize and wheat and so they depend for their support on it [coca] with which they purchase clothing and food. In this region are grown: bananas, guayabas, hicumas, avocadoes, granadillas, papayas, and other local fruits, all without having to cultivate them, and roots that are no less delicious: yuca, sweet potatoes, potatoes, ahipa [jiquima], arracacha, yacon, and peanuts (like the pine-nuts in Castille).8

Food could be grown in the lowlands, but the people involved in coca production were highlanders and may have preferred their accustomed diet. Food may also have been imported to reduce labor expenditures in the lowlands to a minimum, given that terracing and frequent clearing of underbrush were required for cultivation. A larger permanent settlement may have been seen as unwise in view of the problematic relations between highlanders and lowlanders beyond the frontier.9

8 AGI, Charcas 138: . . . Estos indios no tienen mas ocupacion que beneficiar esta planta [la coca] porque la fragosidad del monte 1esympide la siembra de mays y trigo; y asi se balen para su sustento de ella con que compran ropa y comidas. Danse en estos paraxes platanos, guayabas, lucumas, paltas, granadillas, papayas y otras frutas de la tierra, todos de regalo, y rayses que son de no menos gusto: yucas, camotes, patatas, axipaz, racheas, aricomas y mani (a manera de pifiones de Castilla). National Research Council (1989) gives fuller information on "axipaz" or ahipa (pp. 38-45), "rachea" or arracacha (pp. 46-55), and "aricoma" or yacon (pp. 114123). 9 Gade has argued that the risk of contracting leishmeniasis, a protozoal disease earned by a sand fly, en-

ANDEAN PAST 5 (1998) The documents also contain information about the chacras that belonged to the curacas. Eleven camayos served particular curacas in the area of Chuquioma and Chamorro (Ramirez Valverde 1970:306). The curacas testified about the amounts of coca that were harvested for them in preceding years. Don Juan Xaraxuri, the principal curaca of Pocona, had received 10 cestos per mita, although the same fields had been known to produce 100 cestos in earlier years. Xaraxuri was identified as head of the Cotas, and resided in the town of Pocona itself. Another curaca of the Cotas, identified only as don Felipe, had houses in Chimboata, as well as in Pocona. He had a chacra that had produced 30 cestos each mita, although, in the mita preceding the interview, it had yielded only 10 cestos (ibid.:295-296). These curacas had fields in the valleys of Arepucho and Yunno as well (ibid.:307-308). Arepucho is in a valley west ofChuquioma (see Figure 2). Part of the Pocona grant.was composed of mitimas, people who had their origins in' the Cusco region, in the Conde territory south of Cusco, and in the Lake Titicaca region. The principal curaca of the mitimas, don Hernando Turumaya, of the town of Pocona, had formerly received 30 cestos per mita. Three other curacas, all of whom belonged to the group of mitimas, also had coca chacras. One, don Francisco Collasuyo, received 10 cestos per mita from his fields; however, only 3 cestos had been harvested in the preceding mita. Another, identified as don Tomas, curaca of the Collas and resident in Chimboata, received IS cestos. A third, don Francisco Vilcachagua of the ayllo Chinchasuyo, was receiving 3 cestos from fields that had formerly produced 7. Although this accounting may be somewhat incomplete, variability in access to coca among the different curacas is clearly evident (as is an overall drop in the amount of coca harvested). The curacas also described their fields in Tiraque Chico. There, they had chacras of hot peppers and cotton, worked by a total of nine camayos distributed as follows: Two served
demic to the eastern lowlands where coca was grown, also inhibited permanent settlement (1979).

-138 Xaraxuri, Two served Turumaya; Pedro Chirima, a Cota curaca, had one; Pedro Caya, a curaca of the ayllo (kin group) Inga, had two; and Tomas Caua, cacique of the ayllo Colla and perhaps the same don Tomas as mentioned above, had two. Although we might assume that these people were there principally to cultivate hot peppers and cotton" they were described as "camayos of the storage place for the coca". They may also have been involved in the storage and transfer of coca (Ramirez Valverde 1970:304). Sacaca cultivators Sacaca was the name of an encomienda that included some of the people of a larger group known as Charcas (see note 4). It was also the name of their principal town (Espinoza Soriano 1969:141, 148). The people of Sacaca had initiated a suit against their encomendero, don Alonso de Montemayor, for demanding excessive amounts of tribute. To substantiate their suit, they presented records for the years 1548 to 1552, two years before and two years after the first tasa was set (1550) (Documents 1 and 2, Appendix).10The tasa assessed tribute in silver, coca, textiles, maize, tubers, camelids, fat, pigs, chickens, ducks, partridges, honey, wax, camelid hides, tack, salt, and service in the fields and household of the encomendero iri La Plata.ii Much of what Sacaca owed in tribute was obtained in the highlands, but coca, honey, and wax had a lowland origin. What they owedin cocawas substantial:
Item. You will pay each mila that you harvest coca 250 ceslos of coca of the size and measure that you paid and made at the time you were visited which was on the fifth of September in the year 1549 from the chacras that you said had 10 To document what they had paid in coca from 1548 to 1552, records separate from those kept in the highlands were introduced into the testimony. These records were in the hands of quipocamcryos who resided in the yungas of Pocona. 11 In the document livestock are termed ovejas, and could as easily refer to sheep as to camelids. Given later assessments of this same encomienda (Miranda 1975:25), the ovejas here are camelids, probably, llamas.

139belonged to the Inca which are the chacras from which you are accustomed to pay [tribute in coca] and you will deliver them to the settlement of Tiraque. If the encomendero gives you livestock with which to transport it, you will deliver it to La Plata, Potosi or Porco. And for the trip both to and from [La Plata, Potosi or Porco], you, the encomendero, have to give the Indians the food that is necessary and three cestos of coca each mita which they will divide among themselves and neither you nor anyone else will occupy them in any other task in addition to carrying out the above. (Document I, folios 4v-5).

Julien: Coca on the Inca Frontier the coca that was paid, named the 11 chacras that were worked to pay tribute (Table 3) (Document 2). They also stated that these chacras had belonged to Consara. In the absence of Cusco control, he appears to have commandeered fields that had previously been worked for the Incas. Even though curacas, had individual holdings, what was harvested from these 11 fields would have been far larger than a curaca's share. The Tiraque where coca was to be delivered is again Tiraque Chico. However, the delivery of coca at Tiraque, where the people of Pocona delivered their coca, was contested. Montemayor asserted that their traditional delivery point was Totora:
.

Note that the chacras harvested for tribute were chacras that had belonged to the Inca.12 In the lawsuit against Montemayor, the officials (quipocamayos) who kept records of
12 Although it is clear in the cases of Pocona and Sacaca that the tribute was paid from specific chacras, it is more difficult to identifY these resources in the Sonqo documents. The tributaries reported that they paid their tribute from their own fields. When asked in 1568 whether they had community chacras "like the other Indians in the jurisdiction of La Paz who pay their tribute in coca, and lik~ all the others in the jurisdiction of Cusco and La Plata," .one witness said he did not know (Murra 1991a:269). A cover-up was suspected and investigated in 1570. Community chacras were found, but only a small amount of the tribute coca was still being supplied from this source, and furthermore, two of the four fields mentioned had been created within the preceding decade (ibid.:313-315, 327). One witness did provide information about fields that had served this purpose in earlier years. Antonio Llula Estaca, principal curaca of Chacapa, testified that fields that had earlier served for the payment of tribute were no longer cultivated: once did to clean the fields to avoid worm damage. And he said that the worm had gotten into the chacaras that they used to pick coca and that there are no good lands for
building chacaras as good as those were

Don Alonso de Montemayor, in the petition he presented, made a report saying that the Indians of his encomienda had and had had the custom of putting the coca that they paid in tribute in the town of Totora which is two leagues [10 km] beyond Tiraque on the ~ame road, where they had built their houses and settlements for the time when they arrived there; and they were harmed by the order that they deliver [the coca] to another town, and they would have to build their houses and settlements all over again. (Document 1, folio 9)

. . . because

they no longer took the care they

This settlement had to house approximately 150-200 mitayos who harvested the tribute coca and others who worked fields belonging to particular curacas, in addition to any resident personnel. If transport of the coca was the responsibility of others, then another 50-60 people and 100-125 llamas were present at times. The testimony indicates that the settlement included permanent constructions.13

...

. . . porque no tienen el cuidado que entonces


solian tener para limpiar las dichaschacaras para que el gusano no las come dijo que porque en las chacaras donde solian coger la coca se metio el gusano y no hay buenas tierras donde hacer otras chacaras tan buenas como
las otras

. . . (Murra

1991a:279)

A chaera formerly cultivated for the Incas that had been taken over by a curaca was also identified (Murra 1991a:496-497).

13 A delivery point for the coca (a site like Tiraque or Totora, located in the highlands at a convenient point for transfer of the coca) can also be identified for Sonqo. The Sonqo documents date to a considerably later period (1568-70) than the Pocona material, and the point of delivery had changed several times. First Sonqo had delivered its coca to the storage houses at Tohone (or Toone), 3 leagues (IS km) from Chuquiabo (modem La Paz), near the shrine that marked the mountain pass of Canavire (apacheta) (Murra 1991a:77, 515); the coca of Chacapa went to Quicheciste, half a league (2.5 km) from Chacapa; and the coca of Challana to Chiqui, 2 leagues (10 km) from Challana (ibid.: 134, 190, 190,263,267,269).

ANDEAN PAST 5 (1998) The quipocamayos provided the names of 11 chacras, but only the names Chuquioma and Laymetoro appear on modem maps (Figure 2). However, the two Laymetoros located in the highlands are not the Laymetoro of the documents. A number of references in notarial records clearly identify Laymetoro as a coca chacra in the valley of Chuquioma (ANB, Escrituras Publicas 24, folios 161163v,387v-388v). Other aspects of the organization of coca production can be recovered through an analysis of the data provided by the quipocamayos. The time required to transport the coca from Chuquioma to Totora was three days (Table 4). The distance between Chuquioma and Totora is only 30 lan, but the first 20 cover very steep and rugged terrain. Moreover, the people who harvested the coca had to transport it up this steep trail, carrying two cestos per person. From Totora, it was carried by llamas and by "peons" to Potosi or La Plata (modem Sucre),the two final destinations recorded in the documents of 1548-1552. The tenn "peons" is used in the documents, and we shall use it here, but we can identify these laborers as mitayos. In Table 4, we can see the figures for each of the mitas in these years. First~I want to call attention to the number of peons who transported the coca to Totora. With the exception of the first two entries, the numbers support the declarations recorded in an accompanying testimonial to the effect that each peon carried two cestos. In the case of the first two entries, calculation indicates that some of the peons may have carried three cestos. When the same coca was transferred to animals, fewer peons were required. The animals carried two cestos, the same number as the human carriers. Because the sum of animals and peons is in all cases approximately equal to the number of peons who brought the coca out of the yungas to Totora, we can conclude that fewer humans were involved in the next stage of transport, but that they continued to serve as bearers. The figures reveal a proportion of approximately two llamas per human.

-140
Now that we can document a difference between the number of peons who were required to transport the coca from the yungas to Totora and the number involved in taking the same coca to its final destination, the question arises as to whether the people who brought the coca out were the same as those who continued onward with it: the task of harvesting coca may well have been a separate obligation' from the service of long-distance transport.14 One indication that the two obligations were separate is that the transport of the coca to its final destination, aided by llamas, was not automatic: only when the encomendero provided the animals were the people of Sacaca required to take it farther than the point of delivery (Document 1, folio 5). There is another inference we can make from the data in Table 4: the coca that was haryested in each mita was shipped as a single lot to its final destination. If 314 cestos were harvested, the records indicate that 314 cestos were shipped together to their final destination. We cannot infer that coca from one mita was shipped with coca from another, or that the lot was divided to be shipped to different destinations. Coca was customarily shipped within a brief period, consonant with what we know about its value when fresh. At least during the period covered by our documents, Totora was not a site of long-tenn storage. Located at 2800 masl, it would not have been the best place to conserve coca leaf, if that had been a consideration. These documents clearly establish the involvement of the people of Sacaca in coca production. However, in the Toledo assessment (1575), Sacaca was no longer required to pay tribute in coca. Other cultivators Besides Pocona and Sacaca, other groups from the Charcas highlands and eastern lowland valleys paid coca as tribute in the years before the tasa was set (Table 1; see note 4).

14 Patricia J. Lyon has noted that, in the coca-growing region of Paucartambo near Cusco, they were separate obligations.

141While we cannot rule out the possibility that coca was produced in the lowland valleys east of La Plata, notarial documentation for the 1560s and the first half of the 1570s indicates that Chuquioma was the only actively producing coca region during this period. Sale of coca and coca chacras were frequentlyrecorded. One of the transactions involves the sale of chacras by the Yamparaes community (ANB, Escrituras Pliblicas 24, folios 431433). Moreover, Martin de Almendras, the encomendero of Tarabuco, collected tribute from his encomienda in the yungas of Chuquioma, as evident in the following statement by Gonzalo Martin, a witness in the suit against don Alonso de Montemayor:
To the fourth question, he answered that that which he knows about what he was asked is that, at the time referred to by the question, from 48 to 51, during this time this witness' entered the yungas of Chuquioma several times with Martin de Almendras, a vecino [Spanish townsman] of La Plata, to collect his coca; and the Indians paid in coca, and if there was not enough coca, they
paid at the rate of 4 pesos per cesto

Julien: Coca on the Inca Frontier area, the labor input was the same for all groups except for the travel of the mitayos from their province of origin. The data summarized in Table 1 indicate notable differences in the quantities delivered by the various groups who paid coca to their encomendero. The production of PoconaIPojo could be qualified as industrial in comparison with Sacaca or Chayanta, although the documents note a substantial drop from past levels in the quantities of coca supplied by Tarabuco and Cochabilca to their encomenderos in 1548.16 In the Toledo tasa of 1575, Pocona is the only group still assessed in coca. Just what brought about the end of coca production for tribute payment cannot be ascertained. Toledo may have simply required payment in silver from tributaries who continued to be active in coca production, expecting the sale of coca to supply the silver tribute. However, the sharp drop in production o( coca after the boom years of 1548-1552 suggests that supplying coca as tribute had become problematic. Even before 1548, Pocona had been unable to produce cocaat capacity. The problemsappearto have been organizational. Pocona had suffered a serious population loss prior to the visita of 1556. Many of the mitimas returned to their places of origin during the time of the Gonzalo Pizarro rebellion (1544-48). One witness noted that the field sergeant of Pizarro, Francisco,de Carvajal, had collected the people subject to a number of curacas, thereby depopulating Pocona (Ramirez Valverde 1970: 297). At the time of Carvajal's arrival, about 1546, a principal curaca was also said to have taken a large number of people with him into the interior, bringing about the collapse of the coca enterprise (Loredo 1958:155156). In 1556, the caciques of Pocona also attributed the drop in production to a lack of productivity in the fields and to a disease they called earache (Ramirez Valverde 1970:300).

. . .15

The eastern lowlartd valleys were developing as centers of agricultural production to supply the adjacent mining centers. In the abundant documentation related to the transfer of lands in these' valleys,' there are many references to maize and fruit crops, but no references at all to coca. A Spanish resident of Tiraque testified in 1556 that the coca of Pocona, Pojo, and "other repartimientos" was brought there (Document 3, folio 76r; cf. Ramirez Valverde 1970:301). Coca from' different encomiendas was marketed in both Tiraque and Pocona at that time (Document 3, folios 76v, 77r). If the delivery point for the coca was in the Tiraque/Totora
15AGI, Justicia 653, folio 418v:

. . . 10 que

saue de 10 contenido en la pregunta es

que, a la sazon que la pregunta dize de los quatro afios de quarenta y ocho hasta ~inquenta y vno, en el dicho tienpo, este testigo entr6 en los yungas de Chuquioma algunas vezes con Martin de Almendras, vezino de la <;iudad de La Plata, a cobrar coca suya, e que los yndios pagauanla en coca; y si les faltava coca, les pagauan a quatro pesos el sesto.

16 Cochavilca is referred to, by the names of its encomenderos, Segura and Tapia, in Ramirez Valverde (1970:300; see Loredo 1958:170-171)

ANDEAN PAST 5 (1998) Still, in comparison with other groups who paid coca as tribute in 1548, the marked specialization of Pocona in coca production is evident, even years after Spanish entry into the enterprise. The Incas' organization of coca production in the Pocona area, indicated by the quantity of mitima settlers and industrial levels of production, survived some decades after Spanish involvement began. Inca Organization The documents that permit us to trace the involvement of various groups in coca production in Chuquioma, though comparatively early in date, reflect a territorial organization created by the breakup of Inca political units into encomiendas. To understand Inca organization, an effort has to be made to redefine earlier political units. There are notable lacunae in the evidence available for this task (see Note 4), but what has survived for the Charcas region does permit us to trace in broad outline the shape ofInca-political organization. As noted above, Pocona and Pojo were a single entity until the early 1550s, when the two began to be treated as distinct encomiendas (see note 4). The first Spaniard to receive Pocona/Pojo was Pedro de Valdivia. Valdivia had been part of the initial expedition of the Pizarro brothers (Gonzalo and Hernando) into highland Bolivia in 1538. He had also participated in the expedition of Pedro Anzures to found La Plata (modem Sucre) in 1539 (Esteve Barba 1966:28). We do not have the award document, but Valdivia wrote in a letter to his agents at court about having to give up this award -before his expedition to Chile (Santiago de Chile 15 October 1550; Esteve Barba 1966:28-29,42). Francisco Pizarro had given him in encomienda:

-142

Centeno, Mendoza, and Bobadilla held the grant in 1548, when it was still a single entity (Loredo 1958, p. 154). Pocona/Pojo included a substantial highland territory, but the grant was referred to as the "valley of La Canela" (Loredo 1940:52-53), a reference to Chuquioma. Sacaca was also part of a larger territory: In 1598 don Juan Ayaviri Cuysara presented information collected by his father in 1583 that detailed the genealogy of the caciques of Sacaca. Cuysara's great-grandfather was Consara, who had been head of a huno, a unit of 10,000 households in Inca decimal accounting, that was divided into four separate encomiendas: Sacaca, Chayanta, Cochabamba (Santiago del Paso), and Cochabamba (San Miguel de Tiquipaya) (AGI, Charcas 45, folios 20-20v). The Inca project at Cochabamba has _generated interest among ethnohistorians since the discovery of documents that identify the restructuring of productive activities in the valley by the Inca Huayna Capac to grow maize on an industrial scale (Morales 1977; Wachtel 1982). The people granted in encomienda in Cochabamba (the two grants identified above) were composed of mitimas with origins in various highland provinces in the Inca district of Collasuyo. The testimony of Cuysara allows us to identify a larger political unit within which the Inca maize project was embedded and to link Sacaca to it. The Inca restructuring of Cochabamba affected the Pocona/Pojo area so our analysis should encompass the entire region. Native witnesses testified in an inquiry that took place in 1560 that Thupa Inca had taken some lands in the Cochabamba Valley for his own use. However, a large-scale reorganization of agricultural production had been effected in Cochabamba during the rule of his son,
17 Esteve Barba 1966:28: . . . me di6 en dep6sito y encomienda el valle todo llamado de la Canela, que despuees que yo Ie deje Ie di6 al capitan Peranzulez e a su hermano Gaspar Rodriguez e a Diego Centeno; e Vaca de Castro, cuando gobern6 aquellas provincias del Peru a S.M. di6 en el de comer a tres conquistadores, que fue a los capitanes Diego Centeno, Lope de Mendoza e Dionisio de Bobadilla...

. . . the valley of La Canela, which, after I renounced it [to go to Chile] he regranted to Pedro Anzures and to his brother, Gaspar Rodriguez, and to Diego Centeno; and Vaca de Castro, when he governed the provinces of Peru, granted awards ITom this same territory to three conquerors, namely Diego Centeno, Lope de Mendoza and Dionisio de Bobadilla,17

143Huayna Capac. This Inca removed most of the inhabitants of the valley, identified as Chuyes (or Cauis) and Cotas, and resettled them in the PoconaJPojo area. The reason given was that these people were to serve as security against attacks by Chiriguana at the eastern frontier. No mention was made of their involvement in coca production, but the witnesses admitted that they knew less about Thupa Inca's activities than about his son's "because they do not remember and it happened a long time ago" (AHMC, EC 17, folios 425-425v). Information about the Chiriguana "threat" has to be considered in terms of what was happening at the frontier at the time information was gathered. From the 1560s onward, the Spaniards had a heightened awareness of how difficult relations at the frontier could be, as the violence escalated and two wars were declared (in 1574 and 1584). While we can document a threat from beyond the frontier during the reign of Huayna Capac, there is no reason to extend this situation any further into the past (Del Rio and Presta 1985; Julien 1995). It is. with this caveat in mind, that we should examine the testimony of a group of descendants of Thupa Inca Yupanqui, taken in Cusco in 1568. In a listing of military campaigns undertaken by their forebear, very likely recorded on a quipo, Thupa Inca's reorganization of the PoconafPojo area follows the Inca annexation of Charcas. Thupa Inca passes through Charcas and continues on South "to Chiriguanas" and Tucuman. Then Thupa Inca (or his captains) crosses the mountains to Chile. From Chile, he heads north to Tarapaca and then back over the mountains to Pocona:

Julien: Coca on the Inca Frontier In 1568, defense concerns were uppermost and may have inhibited the witnesses' recollection of their forebear's involvement in reorienting the region toward industrial-level coca production. In a later inquiry, conducted in the 1570s', a collective interview of curacas and tributaries led to the copying of an earlier census of the lands that had been distributed under the Incas in the Cochabamba. One of the administrators involved in the inquiry, Francisco de Saavedra, summarized the distribution as follows:
And from [the census and] the testimony it appears that Thupa Inca annexed the valley and that the inhabitants he found there were Cotas and Cavis and Sisisipis [Sipesipes]. He removed them and sent the Cotas and Cavis to Pocona and Mizque [Pojo] where he gave them lands. The Sipesipes were given lands in the valley and some other Indians from this area [the Charcas highlands] were brought in to tend particular fields. Afterward, Huayna Capac made a general distribution of all of the lands that he assigned to himself in the valley and brought in 14,000 settlers of different nations to tend the fields. Some were permanent residents, others came from their own territories to work the lands of Huayna

Capac.. .19

18 Rowe 1985:226: . . . y asi salieron a Pocona y hi(:ieron muchas fortale(:as en el mesmo Pocona y en Sabaypata que es en los Chiriguanas y en Cuzcotuiro y pusso en todas las fortale[(:a]s muchos yndios de diuerssas partes para guardasen la d[ic]ha fortale(:a y frontera a donde dexo muchos yndios orexones y al pressente estan poblados sus hijos y de(:endientes en las d[ic]has ffortale[(:a]s y fronteras. 19 AHMC, EC 17, f.406: E que por ella e por la dicha ynforma(:ion parece que Topa Ynga conquist6 el dicho valle, y a los yndios naturales que en ella hall6, que eran cotas e cavis y sipisipis, los sac6 de su natural, y a los cotas y cavis los pas6 a Pocona y Mizque y alIf les dio tierras, y a los dichos de Sipi Sipi les seiial6 en el dicho valle tierras, e meti6 algunos yndios de esta provinl;ia para que Ie venefil;iasen 'tiertas chacaras; y despues, Guayna Capa hizo repartimiento general de todas las tierras de dicho valle para sf, y meti6 en el venefil;io de las dichas sus chacaras catorze mil yndios de muchas nasl;iones, y algunos eran perpetuos y

. . . and they went to Pocona and made many fortresses in Pocona and in Sabaypata [Samaypata] which is in the Chiriguanas and in Cuzcotuiro and put many people from different places [in them] to serve as guards of the fortress and frontier where he left many orejones. At present their children and descendents reside in the said fortresses and frontier.18

ANDEAN PAST 5 (1998) According to the witnesses, both Thupa Inca and his son shifted elements of local population from Cochabamba to the PoconaIPojo area. The Cochabamba documents, which provide us with a historical narrative about the activities of Huayna Capac in the Valley, only begin to describe the Inca restructuring of productive activity in the region. When we link Cochabamba to Pocona, the rough outline of a much larger project begins to emerge. This involved the industrial production of coca in the yungas of Chuquioma and maize in the Cochabamba Valley, as well as the maintenance of people at the frontier. Some years ago Mercedes del Rio and Ana Maria Presta noted a correspondence between the origins of mitimas assigned to produce maize in the chacra of Colchacollo in Cochabamba and the origins of mitimas who were resettled in the town of Tarabuco at the time of the visita general (1985:232). Colchacollo was organized during the rule of Huayna Capac. This Inca also responded to an emergency situation at the frontier by sending mitima settlers into the region to act as a buffer against incursions from the east (Julien 1995). In part. at least. the organization of maize production was linked to security at the frontier. We can detect another linkage. The mitimas of the encomienda of Tarabuco "had very good coca chacras" (Loredo 1958:171). Like other mitimas in the frontier region. and some groups from highland Charcas. they participated in the industrial production of coca leaf in the yungas of Chuquioma. The Inca project. in the CochabambaIPocona region was a complex web of different groups related through productive activities in the lowlands and security measures at the frontier. Certain aspects of the productive activities organized by the Incas may have been general (like the establishment of a delivery point for
otros venian de sus tierras al venefi'Yiode las chacaras del dicho Ynga. . . The citation refers to both Cotas and Cavis. The term Cavis may refer to the people also known as Chuyes (Urquidi 1970. cited in Ellefson. 1978). although the terms are not linguistic equivalents.

-144 coca), but the composition of a project like the CochabambaIPocona one may have been a unique historical development that obeyed a logic specific to the resources of the region while taking particular circumstances, like the external threat. into account. The documentation for Charcas, though incomplete, allows us to develop an ethnohistorical image of the complex restructuring of productive activities' in lowland Charcas. Creating a similar image in the larger region surrounding the cocaproducing valleys of Sonqo will be a major ethnohistorical undertaking.2o In the yungas

20 Sonqo was originally part of the encomienda of Gabriel de Rojas. Rojas arrived in Cusco with Francisco Pizarro and was present for the foundation of the Spanish city (Lohmann Villena 1986:163-167). He may have received the encomienda at that time. We know something about his encomienda from a description of it in 1548: Gabriel de Rojas receives 50,000 pesos of income from his encomienda, because he has a great deal of coca and maize. He has the coca Indians 85 leagues [425 km] from the city, more or less. Near the city he has service Indians who have a great deal of maize and wheat. He received them from the Marques [Francisco Pizarro]; I do not know if he got any from Vaca de Castro. Al capitan Gabriel de Rojas Ie renta su repartimiento cinquenta mil pesos, por tener mucha coca i maiz, tiene los Indios de la coca ochenta i cinco leguas de esta ciudad poco mas 0 menos, tiene junto a esta ciudad Indios de servicio de mucho maiz i trigo; tienelos por titulo del Marques. no se si tiene algunos por Vaca de Castro. (Loredo 1941:115) The coca referred to was from the yungas of La Paz, judging by the distance given. Rojas may not have received the entire award at one time. The Rojas award headed the list of Cusco encomiendas because it was the most valuable encomienda in the region, more valuable even than the immense encomienda held by Hernando Pizarro, which was in third place. After Rojas' death, his encomienda was split so that the holders did not have encomiendas in two city districts. Pedro Portocarrero received at least part of the Cusco portion, referred to as Quiguares, from Governor Pedro de la Gasca in 1549 (AGI. Escribania de Camara 498c, folios 590v-S92). Alonso de Alvarado appears to have received Sonqo, and maybe Oyune and Suri, from Gasca at the same time (Hampe Martinez 1979:85).

145of La Paz, the Incas organized the extraction of gold. This region was part of an Incas province called Chuquiabo which encompassed territory in the highlands as well as the yungas. We can expect to find an equally complex web of ties between groups involved in primary production in the lowlands, and others involved in food supply, transport, delivery, storage, elaboration of the finished product, and perhaps, defense. By examining these regional networks, we stand to learn a great deal about Inca economy. The answer to the question of who benefited from the production of maize or coca may even become clear. Acknowledgments Support for research in Spanish archives was provided by the Deutsche Forschungsgemeinschaft. A particular debt is owed to the Archive of the Indies in Seville, where the three documents included in the appendix are held. It was a privilege and a pleasure to 'York with Archive staff. A preliminary version of this paper was presented in a symposium organized by Marius'ZiQlkowski at the CIELAC (Congreso Intemacional de Estudios sobre Latino-America y el Caribe) conference in Warsaw. (I 992). Because the participants were archaeologists, the paper tried to define the spatial correlates of coca production. This orientation has been preserved to a great degree in the present version, despite having passed through two drafts since then. The paper was thoroughly rewritten for submission to Andean Past. It was rewritten again to respond to comments and counsel offered by Juan Jose Villarias, Patricia Jean Lyon and John Howland Rowe, as well as by anonymous reviewers. I particularly want to thank
The link between Sonqo and the Quiguares of Cusco is early and may have been forged by the Incas. The second largest award in Cusco was made to Antonio Altamirano. It was worth 30,000 pesos. The most important part of it was 90 leagues (450 km) from Cusco (Loredo 1941:115), again in the area where La paz was later founded. Like Rojas, Altamirano was also awarded Indians near Cusco, but this group cannot be identified at present. The ties between groups near Cusco, who were largely orejones or non-royal Incas, are significant, although substantiating this argument is beyond the scope of the present paper.

Julien: Coca on the Inca Frontier Villarias for sharing his transcription of the Cochabamba documents with me. I also want to thank both Villarias and Lyon for sharing with me the insights they have gained from working on large bodies of primary source material for Cochabamba and Paucartambo, respectively. They know these regions a good deal better than I do. References Cited Documents
Abbreviations of archive names AGI Archivo General de Indias, Seville, Spain AHMC Archivo Hist6rico de la Municipalidad de Cochabamba, Cochabamba, Bolivia. AHP Archivo Departamental de Potosi, Potosi, Bolivia. ANB Archivo Nacional de Bolivia, Sucre, Bolivia. RAH Real Academia de la Historia, Madrid, Spain. AGI, Charcas 25, ramo 6, numero 45, 4 folios. Carta de Diego Christoval Mesia, presidente de Charcas, a S.M., La Plata, 120ctubre 1692. AGI, Charcas 45 Don Juan Ayaviri Cuysara, cacique principal del repartimiento de Sacaca y pueblo Sant Christoual de Panacache y su prouincia, alcalde mayor de los naturales de la prouincia de los Charcas y capitan de los tres naciones della, 1598. 86 folios. AGI, Charcas 138 Relaci6n de las cosas notables del obispado de nuestra sefiora de La Paz . . " 1651, 34 folios. AGI, Escribanfa de Camara 498c Pedro Portocarrero con el Fiscal, 1579. AGI, Justicia 428, nilmero 2, ramo 2 Don Francisco de Mendoza sobre los reparos que se ofrecian para ponerle en posecion de la encomienda que tenia concedida sobre yndios en las provincias del Peru, 1557 [sic: 1556], 119 folios. AGI, Justicia 653, numero 2 EI cacique principal e yndios del pueblo de Sacaca con los herederos de don Alonso de Montemayor sobre la demasia de tributos del tiempo que tubo dichos yndios en encomienda, 1579. AGI, Justicia 1125, nilmero 5 EI capitan Cristoval Barba con el adelantado Juan Ortiz de Zarate ambos vecinos de la ciudad de La Plata sobre el derecho a yndios moyosmoyos, 155I. AGI, Lima 166 Informaci6n de Francisco Rodriguez Peinado, Villa de Salinas [Mizque], 16 enero 1644. AHMC, Expedientes Coloniales 17 Proceso de los indios carangas sobre las tierras de Colcapirhua, 1575. Transcription by Juan J. R. Villarias-Robles, folios 404-462v.

ANDEAN PAST 5 (1998)


AHP, Cajas Reales 18
.

-146
Gade, Daniel W.

Libro donde se asientanlas tasas de los yndios que


estan en la corona real que mand6 hazer el excelentisimo sefior Don Francisco de Toledo Visorrey e Capitan General en estos reinos e provincias
del PirU . . .,

1575.

ANB, Escrituras PUblicas2 Gaspar L6pez, 1562-1572. ANB, Escrituras Ptiblicas 24 Juan Bravo, 1572, 1573, 1575. RAH, Colecci6n Mata Linares, 9/1666 Diario de la entrada a las montafias havitadas de la nacion de Yndios Yuiacarees que en el afio 1796 hizo el R. P. fray Bernardo Ximenez Bejarano, prefecto de missiones del Colegio de San Josse de Tarata, con los padres fray Pedro Hernandez y fray Ylario Coche individuos de dicho colegio. Septiembre a noviembre de 1796. Real Academia de la Historia, folios 264-282v. RAH, Colecci6n Mata Linares, 9/1667 Informe de Tadeo Haenke a don Francisco de Viedma, Cochabamba, 16 febrero 1796. folios 267r-272v.

Publications
Allen, Catherine J. 1988 The Hold Life Has: CocfJ and Cultural Identity in an Andean Community. Washington, D.C.: Smithsonian Institution. America Indlgena 1978 [Issue dedicated to articles on coca.) 38(4). Mexico: Instituto Indigenista Interamericano. Angelis, Kristina 1995 Die Yungas yon La Paz als Erweiterung des Vertikalitlltsmodells anhand der Visita Yon Songo und komplementllrer Quellen. Thesis (M.A.), Bonn: Philosophy Faculty, Rheinische FriedrichWilhelms-Universit!it. Del Rio, Mercedes, and Ana Maria Presta 1985 Un estudio etnohist6rico en los corregimientos de Tomina Yamparaes: casos de multietnicidad. Runa, 14:221-246. Buenos Aires: Facultad de Filosofla y Letras, Universidad de Buenos Aires. Ellefsen, Bernard 1967 Incalllicta y su relaci6n hist6rica. Khana. Revista Municipal de Arte y Letras 11(2) 39:46-62. La Paz, Bolivia: Municipalidad de La Paz. 1978 La dominaci6n incaica en Cochabamba. Bulletin de l'Institut Franfais d'Etudes Andines 7 (12):73-86. Espinoza Soriano, Waldemar 1963 La guaranga y la reducci6n de Huancayo. Revista del Museo Nacional 32:8-80. Lima: Museo Nacional de la Cultura Peruana. 1969 EI memorial de Charcas; "cr6nica" inedita de 1582. Cantuta 4:117-152. Huancayo: Universidad Nacional de Educaci6n. Esteve Barba, Francisco 1966 Cronicas del Reino de Chile. Biblioteca de Autores Espafioles (cont.) 134. Madrid: Ediciones Atlas.

1979 Inca and Colonial Settlement, Coca Cultivation and Endemic Disease in the Tropical Forest. Journal of Historical Geography 5(3):263-279. Golte, JUrgen 1970 Algunas consideraciones acerca de la producci6n y distribuci6n de la coca en el estado Inca. Verhandlungen des XXXVIII Internationalen Amerikanistenkongresses, Stuttgart-MUnchen, 12. bis 18. August 1968,2:71-78. Munich: Kommis-, sionsverlag Klaus Renner. Gonzalez Holguin, Diego 1952 [1608) Vocabulario de la lengua general de todo el Peru //amada Qquicchua, 0 del Inca. Lima: Imprenta Santa Maria. Hampe Martinez, Teodoro 1979 Relaci6n de los encomenderos y repartimientos del Peru en 1561. Historia y Cultura 12:75117. Instituto Indigenista Interamericano 1989 La coca. . [sic) tradicion, rito, identidad. Mexico: Instituto Indigenista Interamericano. Julien, Catherine Jean 1993 Finding a Fit: Archaeology and Ethnohistory of the Incas. In Provincial Inca, edited by Michael A. Malpass, pp. 177-233. Iowa City: University of Iowa Press. 1995 Oroncota entre dos mundos. In Espacio, etnlas, frontera; atenuaciOnespollticas en el sur del Tawantinsuyu, siglos XV-XVIII, edited by Ana Maria Presta. Ediciones ASUR 4, pp. 97-160. Sucre: Antrop610gos del Sur (ASUR). Lee, VincentR. 1998 Reconstructing the Great Hall at Inkallacta. Andean Past 5:35-69. Lohmann Villena, Guillermo 1986 Francisco Pizarro. Testimonio. Documentos oficiales, cartas y escritos varios. Monumenta Hispano-Indiana, 5 Centenario del Descubrimiento de America, 3. Madrid: Consejo Superior de Investigaciones Cientfficas, Centro de Estudios Hist6ricos, Departamento de Historia de America "Fernandez de Oviedo". Loredo, Rafael 1940 Relaciones de repartimientos que existfan en el Peru al finalizar la rebeli6n de Gonzalo Pizarro. Revista de la Universidad Cat6lica del Peru 8(1): 51-62. Lima. 1941 Alardes y derramas. Documentos para la Historia del Peru. Lima: Oil, S.A. 1958 Los repartos. Bocetos para la nueva historia del Peru. Lima: D. Miranda. Lyon, Patricia J. 1994a La coca en perspectiva hist6rica. Revista del Instituto Americano de Arte 14:109-116. Cusco. 1994b The More Things Change . .. Latin American Anthropology Review 6(1):29:-32. 1995 El ocaso de los cocales de Paucartambo. Revista del Museo e Instituto de Arqueologia, Museo Inka25:171-179. Cusco.

ANDEAN PAST 5 (1998)


AHP, Cajas Reales 18
.

-146
Gade, Daniel W.

Libro donde se asientanlas tasas de los yndios que


estan en la corona real que mando hazer el excelentisimo sefior Don Francisco de Toledo Visorrey e Capitan General en estos reinos e provincias
del PirU . . ., 1575.

ANB, Escrituras PUblicas2 Gaspar Lopez, 1562-1572. ANB, Escrituras PUblicas24 Juan Bravo, 1572, 1573, 1575. RAH, Coleccion Mata Linares, 9/1666 Diario de la entrada a las montafias havitadas de la nacion de Yndios Yuracarees que en el afio 1796 hizo el R. P. fray Bernardo Ximenez Bejarano, prefecto de missiones del Colegio de San Josse de Tarata, con los padres fray Pedro Hernandez y ftay Ylario Coche individuos de dicho colegio. Septiembre a noviembre de 1796. Real Academia de la Historia, folios 264-282v. RAH, Coleccion Mata Linares, 9/1667 Informe de Tadeo Haenke a don Francisco de Viedma, Cochabamba, 16 febrero 1796. folios 267r-272v.

Publications
Allen, Catherine J. 1988 The Hold Life Has: Cocp and Cultural Identity in an Andean Community. Washington, D.C.: Smithsonian Institution. America Indfgena 1978 [Issue dedicated to articles on coca.] 38(4). Mexico: Instituto Indigenista Interamericano. Angelis, Kristina 1995 Die Yungas yon La Paz als Erweiterung des Vertikalitlitsmodells anhand der Visita yon Songo und komplementlirer Quellen. Thesis (M.A.), Bonn: Philosophy Faculty, Rheinische FriedrichWilhelms-Universitlit. Del Rio, Mercedes, and Ana Maria Presta 1985 Un estudio etnohistorico en los corregimientos de Tomina Yamparaes: casos de multietnicidad. Runa, 14:221-246. Buenos Aires: Facultad de Filosofla y Letras, Universidad de Buenos Aires. Ellefsen, Bernard 1967 Incall~cta y su relaci6n historica. Khana. Revista Municipal de Arte y Letras 11(2) 39:46-62. La Paz, Bolivia: Municipalidad de La Paz. 1978 La dominacion incaica en Cochabamba. Bulletin de I'Institut Franfais d'Etudes Andines 7 (12):73-86. Espinoza Soriano, Waldemar 1963 La guaranga y la reducci6n de Huancayo. Revista del Museo Nacional 32:8-80. Lima: Museo Nacional de la Cultura Peruana. 1969 El memorial de Charcas; "cronica" inedita de 1582. Cantuta 4:117-152. Huancayo: Universidad Nacional de Educacion. Esteve Barba, Francisco 1966 Cronicas del Reino de Chile. Biblioteca de Autores Espafioles (cont.) 134. Madrid: Ediciones Atlas.

1979 Inca and Colonial Settlement, Coca Cultivation and Endemic Disease in the Tropical Forest. Journal of Historical Geography 5(3):263-279. Golte, Jilrgen 1970 Algunas consideraciones acerca de la produccion y distribucion de la coca en el estado Inca. Verhandlungen des XXXVIII Internationalen Amerikanistenkongresses, Stuttgart-Munchen, 12. bis 18. August 1968,2:71-78. Munich: Kommis-, sionsverlag Klaus Renner. Gonzalez Holguin, Diego 1952 [1608] Vocabulario de la lengua general de todo el Peru llamada Qquicchua, 0 del Inca. Lima: Imprenta Santa Maria. Hampe Martinez, Teodoro 1979 Relacion de los encomenderos y repartimientos del Peru en 1561. Historia y Cultura 12:75117. Instituto Indigenista Interamericano 1989 La coca. . [sic) tradicion, rito, identidad. Mexico: Instituto Indigenista Interamericano. Julien, Catherine Jean 1993 Finding a Fit: Archaeology and Ethnohistory of the Incas. In Provincial Inca, edited by Michael A. Malpass, pp. 177-233. Iowa City: University of Iowa Press. 1995 Oroncota entre dos mundos. In Espacio, etnfas, frontera; atenuaciOnespolfticas en el sur del Tawantinsuyu, siglos XV-XVIII, edited by Ana Maria Presta. Ediciones ASUR 4, pp. 97-160. Sucre: Antropologos del Sur (ASUR). Lee, Vincent R. 1998 Reconstructing the Great Hall at Inkallacta. Andean Past 5:35-69. Lohmann Villena, Guillermo 1986 Francisco Pizarro. Testimonio. Documentos oficiales, cartas y escritos varios. Monumenta Hispano-Indiana, 5 Centenario del Descubrimiento de America, 3. Madrid: Consejo Superior de Investigaciones Cientificas, Centro de Estudios Historicos, Departamento de Historia de America "Fernandez de Oviedo". Loredo, Rafael 1940 Relaciones de repartimientos que existian en el Peru al finalizar la rebelion de Gonzalo Pizarro. Revista de la UniversidadCatolicadel Peru 8(1): 51-62. Lima. 1941 Alardes y derramas. Documentos para la Historia del Peru. Lima: Gil, S.A. 1958 Los repartos. Bocetos para la nueva historia del Peru. Lima: D. Miranda. Lyon, Patricia J. 1994a La coca en perspectiva historica. Revista del Instituto Americano de Arte 14:109-116. Cusco. I994b The More Things Change. .. Latin American Anthropology Review 6(1):29:-32. 1995 El ocaso de los cocales de Paucartambo. Revista del Museo e Instituto de Arqueologfa, Museo Inka 25:171-179. Cusco.

147Matienzo, Juan de 1967 Gobierno del Peru (1567). Edition et etude preliminare par Guillenno Lphmann Villena. Travaux de l'Institut Franfais d'Etudes Andines 11. Meruvia B., Fanor 1991 La coca en los yungas de Pocona. Historia y Cultura 20: 195-2.10. La Paz, Bolivia. Miranda, Crist6bal de 1925 Relacion hecha por el Virrey D. Martin Enriquez de los oficios que se proveen en la gobernacion de los reinos y provincias del Peru, 1583. Gobernantes del Peru, cartas y papeles, Siglo XVI, 9: 114-230. Publicacion dirigida por D. Roberto Levillier. Madrid. 1975 Tasa de la visita general de Francisco de Toledo [1583]. Introduccion y versi6n paleografica de Noble David Cook. Lima: Universidad Nacional Mayor de San Marcos. Morales, Adolfo de 1977 Repartimiento de tierras por el Inca Huayna Capac (Testimonio de un documento de 1556). Cochabamba, Bolivia: Museo Arqueol6gico, Departamento de Arqueologla, Universidad Boliviano Mayor de San Simon. Murra, John Victor 1991a Visita de los valles de Sonqo en los yunka de coca de la Paz [1568-1570J. Madrid: Instituto de Cooperaci6n Iberoamericana. 1991b Introduccion al estudio' de la hoja de .coca [Erythroxylon coca] en los Andes. In Visita de los valles de Sonqo en los yunka de coca de la Paz [1568-1570J, pp. '565-581. Madrid: Instituto de Cooperaci6n Iberoamericana. National Research Council 1989 Lost Crops of the Incas: Little-Known Plants of the Andes with Promise for Worldwide Cultivation. 'Washington, D.C.: National Academy Press. Nordenskiold, Erland von 1915 1ncallacta, eine befestigte und von Inca Tupac Yupanqui angelegte Stadt. Ymer 2: 169-185. Pacini, Deborah, and Christine Franquemont, editors 1986 Coca and Cocaine; Effects on People and Policy in Latin America. Proceedings of the Conference "The Coca Leaf and its Derivatives-Biology, Society and Policy", Sponsored by the Latin American Studies Program (LASP), Cornell UniversitjJ,April 25-26, 1985. Peterborough, New Hampshire: Cultural Survival Report 23. Parkerson, Phillip T. 1984 El monopolio incaico de la coca: l,Realidad 0 ficci6n legal? Historia y Cultura 5:1-27. La Paz, Bolivia. Ramirez Valverde, Maria 1970 Visita a Pocona (1557). Historia y Cultura 4:269-309. Lima: Museo Nacional de Historia. Romano, Ruggiero 1982 Alrededor de dos falsas ecuaciones: coca buena=cocaina buena; cocafna mala=coca mala. AI/panchis 19:237-252. Cusco. Rostworowski de Diez Canseco, Maria 1988 Conflicts over Coca Fields in XVlth-Century Peru. Memoirs of the Museum of Anthropology, University of Michigan 21. Ann Arbor.
.

Julien: Coca on the Inca Frontier


Rowe, John Howland 1982 Inca Policies and Institutions Relating to the Cultural Unification of the Empire. In The Inca and Aztec States, 1400-1800; Anthropology and History, Studies in Anthropology, edited by George A. Collier, Renato I. Rosaldo and John D. Wirth, pp. 93-118. New York: Academic Press. 1985 Probanza de 10s incas nietos de conquistadores. Historica 9(2): 193-245. Lima: Pontificia Universidad Cat6lica del Peru. Saignes, Thierry 1986 En busca del poblamiento etnico de los Andes bolivianos (Siglos XV y XVI). Avances de Investigacion 3. La Paz, Bolivia: Museo Nacional de Etnografla y Folkore. Trimbom, Hennann 1936 Quellen zur Kulturgeschichte des prakolumbischen Amerilca, edited by Hennann Trimbom. Stuttgart: Strecker und SchrOderVerlag. Urquidi, Jose M. 1970 EI origen de la noble Villa de Oropesa. Cochabamba, Bolivia: Ed. Municipalidad de Cochabamba. Viedma, Francisco de 1836 Descripci6n geografica y estadistica de la provincia de Santa Cruz de la Sierra [1793]. Colecci6n de obras y documentos relativos a la historia antigua y modema de las provincias del Rio de la Plata 3, edited by Pedro de Angelis. Buenos Aires: Imprenta del Estado. Villanueva Urteaga, Horacio 1970 Memoria de los yndios que dieron los caciques y viejos de este valle de Yucay que andaban fuera de ella con nombre de yanaconas. Revista del Archivo del Cuzco 13:85-148. Wachtel, Nathan 1982 The Mitimaes of the Cochabamba Valley: The Colonization Policy of Huayna Capac. In The Inca and Aztec States 1400-1800: Anthropology and History, edited by George A. Collier, Renato I. RosaIdo, and John D. Wirth, pp. 199-235. New York: Academic Press.

ANDEAN PAST 5 (1998) Table 1. Tribute in coca paid in Charcas before 1548 (approximate).
Region Serranfa de Mojos or Cocapata Cestos Group Chayanta Sacaca Sacaca PoconatPoJo Cochabamba (El Paso) Cochabamba (Tlquipaya) Caracara Yamparaes Tarabuco Gualparocas Cochabilca Moyomoyos

-148

1500 700 450 12000/6000 100 50 500 600 600 100 1000 150

Source: Loredo 1940; 1958: 149-172.

Table 2. Encomiendas where more than 20% of total tribute was paid in coca in the Toledo assessment. .
Encomienda Pocona Majes I Gualla Copacopa MaJes II St>DO Ayapata Tarabuco Ollachea Source: Miranda 1975.

Coca
(Cestos)
..

Net
. Tributaries 527 162 190 120 138 362 72 90 52

Cestos
per Trib. 2.85 0.49 6.00 4.00 0.36 4.46 4.00 5.33 5.00

Tribute in coca 72.46% 61.54% 50.29% 39.54% 38.46% 30.47% .23.68% 23.19% 21.3 8%

1500 80 1140 480 50 1615 288 480 260

Table 3. Chacras of coca worked by the people of Sacaca. Payromani Chuquioma Laymitoro Sipsipampa otra chacra en Chuquioma donde agora estan poblados los espafioles Cataquila Liquiliqui Apachita Tontoni Catani Charupampa
Source: AGI, Justicia 653, pieza 1,folio 370r.

149-

Julien: Coca on the Inca Frontier

Table 4. Tribute in coca paid by the people ofSacaca (1548-1551).


Transport to Totora: Time 3 days 3 days 3 days 3 days 3 days 3 days 3 days 3 days 3 days 3 days 3 days 3 days Transport to Final Destination: Peons Llamas 57 63 63 46 51 72 41 54 51 31 31 41 100 116 126 100 120 110 70 82 100 70 80 90

Ano 1548 Total 1549 Total 1550 Total 1551 Total

Mita 1 2 3 1 2 3 1 2 3 1
2'

Cestos 314 355 377 1046 292 341 363 996 222 274 302 798 203 242 262 707

Peons 150 179 189 146 171 182 111 136 151 102 121 131

Source: AGI, Justicia 653, pieza l,folios 370r-372v.

ANDEAN PAST 5 (1998) Documentary Appendix

-150

1. A.G.l, Jus/ida 653, pieza 1, folios 4-9v. Tasa de Sacaca hecha par la audiencia, Los Reyes, 1 octubre 1550. Nos, don frai Geronimo de Loaisa, por la gra~ia de Dios y de la santa sede apostolica de Roma primero obispo y ar~obispo desta <;ibdadde Los Reyes y del consejo de Su Magestad, y elli~enciado Fernando de Santillan, oydor en la Avdien~ia y Chan~illeria Real que por mandado de Su; Magestad rreside en esta dicha ~ibdad, y fray Domingo de Santo Thomas de la Horeden de los Predicadores, por el nombramiento y comision a nos dada por el muy yllustre sefior elli~enciado Pedro Gasca del Consejo de Su Magestad de la Santa y General Ynquisi~ion y su presydente en estos Reynos y prouin~ias del Piru para entender en hazer la tasa de los tributos que los rrepartimientos de los dichos rreinos han de dar a sus encomenderos, etc., avos, el capitan don Alonso de Montemayor, vezino de la Villa de Plata, Prouin~ia de los Charcas, y avos, Ayabuiri y Achacacha,ca~iques, y a los demas prin~ipales e yndios vuestros subjetos que al presente soys y despues de vos sub~edieren en el rrepartimiento de Sacaca questAencomendado en vos, el susodicho, y a cada vno y qualquier de vos, saued que en cunplimyento de 10 que Su Magestad tiene proueydo y mandado a~erca de la tasa que se a de hazer de los tributos que los naturales destos dichos rreinos han de dar a sus encomenderos, assi para que los susodichos sepan 10que les han de pedir y les dar como para que los dichos naturales sean bien tratados y se conserven y aumenten, se nombraron visytadores que visytasen el. dicho vuestro rrepartimiento, los quales, como sabeys, hizieron la visyta del y la presentaron ante nos; y visto y comunicado con los visytadores y otras [f. 4v] personas que paresce que podian thener noti~ia de la dispusy~ion y posibilidad del dicho rrepartimiento e yndios, del por virtud del dicho nombramyento, tasamos y declaramos dever dar el dicho rrepartimieIito, en tanto que Su Magestad 0 la persona que en su rreal nombre otra cosa a~erca de la dicha tasa dispone y manda, los tributos que de yuso yran declarados por la forma y.horden que se sigue. Primeramente dareys vos, el dicho ca~ique e yndios del dicho rrepartimiento, al dicho vuestro encomendero en cada un ano quatro mill y quinientos pesos, de valor cada vno de a quatro~ientos y ~inquenta maravedies, en oro 0 en plata, como vos los dichos yndios mas quisieredes y mejor pudierdes, cada quatro meses mil y quinientos pesos, puestos en la Villa de Plata en casa del encomendero.

Yten. Dareis cada mita, de las que cogeis la coca, dozientosy ~inquentacestos de coca, del
tamano y medida que los dauades y haziades al tiempo que os visitaron, que fue a ~inco dias del mes de septiembre del ano pasado de mil e quinientos y quarenta y nuebe afios, de las chacaras que diz que heran del Ynga, que son las chacaras de que las soleys dar; y los porneys en el asyento [f. 5] de Tiraque, y si el encomendero os diere ganado para en que Ilevallo, se la porneis en la Villa de Plata 0 Potosi 0 Porco; y a la yda y a la buelta vos, el dicho encomendero, aveis de dar a los dichos yndios la comida que fuere menester y cada mita tres ~estos de coca que rrepartan entre sf; y no los ocupareys vos ni otra persona en otra cosa alguna mas de en hazer 10dicho. Yten. Dareys cada seys meses quinze vestidos de avasca, la mitad de hombre y la otra mitad de muger, que se entiende vn vestido manta y camiseta y anaco y liquida, la manta del yndio y el anaco de la yndia de dos baras en ancho y otras dos baras en largo y otras dos baras en largo [tachado], y la camiseta de bara y ochava en largo y en el ancho del rruedo dos baras menos ochaua, y la liquida de bara y ter~ia en largo y en el ancho de vna bara, puestos en casa del encomendero en la dicha Villa. Yten. Dareys cada seys meses tres fra~adas, del tamafio que 10soleys dar, y quatro mantas para cauallos y otros quatro mandiles, todo de lana, puesto en casa del encomendero.

151-

Julien: Coca on the Inca Frontier

Yten. Dareys cada aiio seis~ientas [f. 5v] fanegas de mayz, puestas en el asyento de Potosi 0 Porco 0 Villa de Plata, en vuestros propios ganados. Yten. Dareys cada aiio ochenta fanegas de papas y chufio, puestas en vuestras tierras. Yten. Dareis cada seys meses ~inquenta ovejas, y dando machos, no se os pidan hembras porque puedan criar; y en cada vna Pasqua, de las tres del aiio, vna oueja y vn cordero, puesto en casa del encomendero. Yten. Dareis cada seys meses tres arrouas de seuo y ocho cantarillos de manteca de oveja, del tamaiio que los soleys dar, puesto en casa del encomendero. Puercos Yten. Dareys cada vn aiio, pasado este primer aiio que se cuente desde el dia questa tasa se notificare avos, el dicho ca~ique, en adelante, veynte puercos de aiio y medio 0 dende arriba; y no los dando, dareys por cada oveja [tachaco] puerco vna oveja. Yten. Dareys cada seys meses veynte y ~inco gallinas, la mitad hembras, y ~ient patos y patas y setenta y ~inco pares de perdizes, puestos en casa del encomendero. Yten. Dareis cada semana treinta huevos y algund pescado, si 10oviere, en sus tierras, puesto en ellas mesmas. [f. 6] Yten. Dareis cada seys meeses [asi] media arroua de miel y otra media de ~era, puesto en casa del encomendero. . Yten. Dareys cada seis meses quinze pescue~os de ovejas adovados, puestos en casa de vuestro .. encomendero. Yten. Dareys cada seys meses seis cargas de sal, puestas en casa del encomendero, y assimesmo, tres cargas de sal, en vuestras tierras. Yten. Dareys cada aiio de xaquimas con sus cabiestos y ... con sus latigos, y sueltas, de cada cosa destas diez, de cabuya 0 de lana, puesto en casa del encomendero. Yten. Senbrareys, benefi~iareys y cogereyes en la Villa de Plata en las chacaras del encomendero dos fanegas de mayz y trigo, y en vuestras tierras ocho fanegas de mayz y trigo; y para el trigo vos, el encomendero, les aveys de dar la semilla para ello; y 10 que se cogierre y pro~ediere destas dichas sementeras aveys de dar el mayz desgranado y el trigo en~errado en espiga; y vos, el encomendero, 10aveys de trillar a vuestra costa y ayudaros ... a ello algunosde los dichos yndios; y todo 10que se cogiere y pro~ediere de la dicha sementera que hizierdes en la Villa 10aveys de poner en casa del encomendero; y 10que se cogiere de la sementera de vuestras [f. 6v] tierras la aveis de poner en casa de el encomendero [tachado] y 10 que se cogiere de la sementera de vuestras tierras [asi, repetido] 10dareys en ellas mesmas; y si el dicho encomendero quisiere senbrar en vuestras tierras con bueyes, como sea syn perjuicio dellas, Ie dareys al tiempo del senbrar,para que ayudenallabrador y para rregar, doze yndios, y al tiempo del deseruary coger, cada vez treynta yndios e yndias; y dandole los dichos yndios para que ayuden allabrador, como dicho es, en tal caso no aveys de senbrar las dichas ocho fanegas de mayz y trigo que os mandamos senbrar en vuestras tierras; y de todo 10que pro~ediere destas dichas sementeras, senbrandolas vos, los dichos yndios, 0 ayudandolas a senbrar, 10dareys puesto en vuestras tierras.

ANDEAN PAST 5 (1998)

-152

Yten. Dareys para serui~io gordinario [asi, para h] de la casa del encomendero en la dicha Villa de Plata quinze yndios e yndias que se muden por sus mitas, de los quales sean las tres ofi~iales; y si el encomendero fuere a vuestras tierras, Ie dareys todo el tiempo que en ellas estuuiere para que Ie siruan de serui~io hordinario ocho yndios e yndias de los dichos quinze que os mandamos dar en la Villa. Yten. Dareys para guarda de los ganados 0 benefi~io de guertas del encomendero, si los tuuiere, seys yndios, [f. 7] en vuestras tierras; y en la dicha Villa, otros dos para 10mesmo. Yten. Si el encomendero quisiere llevar comida al asyento de las minas de Potosi 0 Porco 0 Villa de Plata en su ganado, Ie dareis vos, el dicho ca~ique e rrepartimiento, dos vezes en el afio, cada vez treynta, yndios para que ayuden a cargar y descargar y guardar el dicho ganado; y vos, el dicho encomendero, ni otra persona no los ocupareys en otra cosa; y a la yda y a la buelta les aveis de dar la comida que fuere menester. Y porque con menos cargo y escrupulo de con~iencia vos, el dicho encomendero, podays llevar los dichos tributos, os encargamos y mandamos que hagais doctrinar a los dichos naturales en las cosas de nueStra santa fe catholica y a thener y guardar ley natural y buen poliyia; y no auiendo clerigo 0 rreligioso que 10haga, pomeys vn espafiol de buena vida y exenplo que los doctrine en 10susodicho. Y porque al clerigo 0 rreligioso que doctrinare a los dichos naturales es justo que se Ie provea de comoda sustenta~ion, en tanto que no ay diezmos de' que se pueda sustentar, vos, el dicho ca~ique e yndios del dicho rrepartimiento, dareis para su sustentacion carla mes quatrO hanegas de mayz y dos ouejas y cada [f.0 7v] tres meses un puerco 0 una oveja en su l\Jgary una carga de sal y cada quatro meses vn cantarillo de manteca de ovejas y cada semana ocho aves, gallinas y patos y perdizes, y los dias de pescado, cada dia diez huevos y algund pescado, si 10 ouiere, en vuestras tierras, y cada dia vn cantarillo de chicha y lena para quemar y yerua para su caualgadura; y el salario de dineros y otras cosas mas, si fuere menester para la sustentayion del dicho clerigo 0 rreligioso, 10pagareis vos, el dicho encomendero, 0 la parte que os cupiere. Por tanto, por el presente mandamos avos, el dicho capitan don Alonso de Montemayor, y avos, Ayavire y Achacata, cayiques, y a los demas prinyipales e yndios vueStrOS subjetos, y a cada vno, qualquier de vos que al presente soys y despues de vos subyediere en la dicha encomienda y rrepartimiento, que guardeys y tengais la tasa de susocontenida, y quedeys en cada un afio, que se cuente desde [e]1dia que os fuere notificada avos, el dicho ca~ique, en adelante, por sus mitas los tributos y cosas en ella contenidas, so pena que si, pasado el dicho tennino en que asy 10 aueis de dar, dentro de veynte dias mas, primeros siguientes, no los dierdes y pagardes y ouierdes dado y entregado al dicho vuestro encomendero, conforme a la dicha tasa, que Ie deys y pagueis los tributos y cosas que asy Ie deuierdes y rrestardes por dar y entregar de cada mita con el doblo y costas que sobrello se les siguieren y rrecreyieren; en la qual [f. 8] dicha pena vos condenamos y avimos por condenados en ellas, desde agora para entonces y de entonyes para agora; y mandamos a la justiyia mayor y hordinaria de la dicha Villa de Plata hagan y manden hazer entrega y execuyion en vuestras personas y bienes por el dicho prin~ipal y pena del doblo y costas confonne a derecho, y assimesmo, que vos, el dicho capitan don Alonso de Montemayor, y los que despues de vos subyedieren en la dicha encomienda, no podais rrecibir ni llevar ni rre~ibais ni lleveis mas por vos ni por interpuesta persona publica ni secretamente, direte ni yndirete, Otra cosa alguna del dicho rrepartimiento, saluo 10 ayaye [tachado] contenido en la dicha tassa sobre penas en la dicha prouision rreal de Su Magestad contenidas, que es que, por la primera vez que paresyiere que ayays rreyibido mas, como dicho es, demas de boluer a los dichos yndios 10que asi les ouierdes llevado, pagueys de pena el quatro tanto del valor dello para la camara de Su Magestad, y por la segunda vez, rrestituyais assimesmo a los dichos yndios 10 que asy les

153-

Julien: Coca on the Inca Frontier

ouierdes llevado y seais priuado de la dicha encomienda dellos y perdais otro qualquier derecho que tengais 0 podais tener a los dichos tributos, y mas, la mitad de todos vuestros bienes para la camara de Su Magestad; en las quales dichas penas yncurrais vos, el dicho encomendero, y qualquier persona que despues de vos subyediere en la dicha encomienda, si eyedierdes de 10en la dicha tassa contenido; y vos condenamos y avemos por condenados en elIas, desde agora para entonyes y de entonyes para agora, aplicados segun dicho es; y porque vos, el dicho en,;, comendero, no pretendais ynoranyia y sepais 10que aveys de rreyibir y los dichos cayiques e yndios 10 que han de dar, mandamos que cada vno [f. 8v] de vos tenga en su poder este proveymiento de vn..., rreseruando, como rreseruamos en nos y en la persona que en nombre de Su Magestad 10 ouiere de hazer, facultad de afiadir 0 quitar en la dicha tassa todas las vezes que paresyiere deverse quitar 0 anadir en ella, conforme a 10que [e]1tienpo y posibilidad de los dichos cayiques e yndios pidiere y rrequiriere. Fecho en Los Reyes, a primero de otubre de mil y quinientos e yinquenta aiios. Fray Gieronimo, archip[reste] de Los Reyes. Ellicenciado Hernando de Santillan. Fray Domingo de Santo Thomas. Por mandado de su Sacra Real Magestad y Mercedes, Pedro de Avendaiio. En la <;ibdad de Los Reyes, a diez dias del mes de abril de mil y quinientos e yinquenta e vn afios, el muy yllustre sefior don Geronimo de Loayssa; primero aryobispo de esta dicha <;ibdad,y los muy magnificos senores ellicenciado Andres de <;ianca, oydor en la Avdien~ia y Chanc;:il1eria Real, que por mandado de Su Magestad rreside en esta dicha <;ibdad, y frai Domingo de Santho Thomas de la Horden de los Predjcadores, a quien esUicometida la continuayion de la tasa de los tributos que los rrepartimientos destos dichos rreynos del Piru an de dar a sus encomenderos por el muy yllustre sefior elIiyenciado Pedro Gasca del Consejo de Su Magestad de la Santa y General"Yl)quisiyion y su presydente en estos dichos rreynos, etc., dixeron que, por quanta, en la tassa que hizieron en primero de otubre del afio pasado de mil y quinientos y yinquenta afios de los tributos que los cayiques e yndios del rrepartimiento de Sacaca, que al presente es~aencomendado en don Alonso [f. 9] de Montemayor, vezino de la Villa de Plata, an de dar a su encomendero, mandaron que la segunda partida della, que diesen cada mita de las que cogiesen coca dozientos y yinquenta yestos de la medida y tamafio que los dauan en el tienpo que los visitaron, puestos en el pueblo de Tiraque, y por parte del dicho don Alonso de Montemayor, por su peticion que present6, hizo rrelayion, diziendo que 105yndios del dicho rrepartimiento abian tenido y tenian por costumbre de poner la coca que dauan de tributo en el pueblo de Totora, que es dos leguas adelante de Tiraque en el mismo camino, donde tenian fechas sus casas y assientos para quando alii llegaron, y rrezibian agrauio en mandar que 10 diesen en otro pueblo, e que seria nesyesitarlos a hazer casas y assiento de nuebo, de que les siguiria dafio; por tanto, que se proveyesey mandaseque el dicho tributo de coca10diesenpuesto en el pueblo de Totora,10 que, visto por Su Seiiono y Meryedes, y la ynformayion que de personas particulares sobrello ouieron, dixeron que mandavan y mandaron que los cayiques prinyipales e yndios del dicho rrepartimiento de Sacaca, desde el dia queste auto se les notificare en adelante, pongan a su encomendero los dichos dozientos y yinquenta yestos, que por la segunda partida de la dicha tassa se les manda dar en cada vna mita de las que la cogieren, puestos en el dicho pueblo de Totora, y enbargante que la dicha partida diga en Tiraque; y que, con esta dedaracion guarden la dicha partida, segund y como en ella se..., dexando como dexauan en... fuerca y vigor todo 10demas en la dicha tassayion contenido como en ella se ... para que se guarde,so las penas en [f. 9v] ella contenidas; y 10 firmaron de sus nombres. Frai Gieronimo, archep[reste] de Los Reyes. EI Iiyenciado <;ianya. Frai Domingo de Santo Thomas. Por mandado de Su Sefiorio y Meryedes, Pedro de Avendaiio. Y Francisco Lopez, [e]scriuano de Su Magestad y de camara en la Audienyia e Chanyilleria Real desta <;ivdad de Los Reyes, fize sacar este traslado del original quesUien el ofiyio del secretario

ANDEAN PAST 5 (1998)

-154

Pedro de Auendano, en diez e seis dias del mes de henero de mil e quinientos e sesenta e tres anos. En testimonio de verdad, fize aqui mi signo. Francisco Lopez [firmado y rubricado]. 2. AGI, Justicia 653, pieza 1, folios 370-372v. Re/acion de /os quipocamayos de /0 que dieron a don Alonso de Montemayor. Relacion de los quipocamayos de las chacaras de coca de los quatro anos que pagaron sin tasa a don Alonso de Montemayor, que fue encomendero del rrepartimiento de Sacaca desde el ano de' xlviii fasta el ano de Ii, que fueron 105quatro anos que pagaron sin tasa, por fuerya y contra su voluntad, gran numero de yestos de coca y por ellos fueron vejados y molestados todo el rrepartimiento de Sacaca en los dichos quatro anos por el dicho don Alonso de Montemayor; y todos los quipocamayos que rresidian en los dichos valles de los yungas dedaran, conforme a sus quipos; y quando ponemos la demanda a los herederos del dicho don Alonso de Montemayor no se hallaron estos dichos quipocamayos por estar, como estauan lejos, y los otros quipocamayos no supieron dar rrazon por no tener los quipos y por no auer rresidido en los dichos valles de los yungas, como estos dichos quipocamayos. Las chacaras que en aquel tiempo auia y eran de Consara, cayique prinyipal del rrepartmiento de Sacaca, porque en aquel tienpo era senor, y son estas que se.siguen. Payromani Chuquioma Laymitoro Sipsipampa otra chacara en Chuquioma donde agora estan poblados los espanoles Cataquila Liquiliqui Apachita Tontoni Catani Charupampa Y estas dichas onze characaras [sic: chacaras] de coca eran de Consara, senor de todos los charcas y del rrepartimiento de Sacaca EI Primero ano de 48 En cada vn ano se cojian tres vezes en el ano de las dichas chacaras, de quatro a quatro meses, como es publico y notorio. Y asi la primera mita cojieron de las dichas chacaras trezientos y catorze yestos de coca; y para sacar estos dichos yestos de coca al pueblo de [ en blanco] [f. 370v] dieron por manado de su mayordomo, que a la sazon era Anton Gomez que siruio en las dichas chacaras de los yungas dos anos, y sacaron ciento y yinquenta peones ados cestos yestos [asi, repetido] cada vno porque los yestos de aquel tiempo eran muy grandes y pesa dos de los del dia de oy y tartaclaron [sic: tardaron] tres dias en poner en el dicho pueblo' de Totora; y desde el pueblo de Totora hizieron lleuar los dichos yestos de coca a la <;iudad de La Plata con el ganado de los del dicho rrepartimiento de Sacaca; y parte dello lleuaron 105yndios en esta manera: fueron yien cameros cargados, dos yestos cada vno, porque en aquel tienpo 105cestos eran muy grandes, y mas, fueron yinquenta y siete yndios, cargados cada vno ados yestos; y los propios yndios cargaron con el dicho ganado hasta la <;iudad con vn criado del dicho don Alonso de Montemayor que se llamaua Quintero 0 Tintiro.

155-

Julien: Coca on the Inca Frontier

La segunda mita de los quatro meses cojieron de las dichas chacaras trezientos. y ~inquenta y ~inco cestos de coca; y para sacar estos dichos cestos de coca al dicho pueblo de Totora fueron ~iento y setenta y nueve peones, a lTazon de a dos ~estos de coca; y tardaron tres dias hasta el pueblode Totora;y desdeel dichopueblode TotorahizieronIleuarlos dichoscestos de cocaala Villa de Potosi con ~ientoy dies y seis cameros de los yndios y sesenta y tres yndios, a lTazonde ados cestos, como esta lTeferidoaITiba;y los dichos yndios cargaron y fueron con el dicho ganado hasta la Villa de Potosi con el dicho Quintero, criado del dicho don Alonso de Montemayor. La tercera mita de los quatro meses cojieron de las dichas chacaras trezientos y setenta y siete ~estos de coca, que a la sazon era su criado Anton Gomez; y para sacar estos dichos ~estos de coca al dicho pueblo de Totora fueron ~iento o. ochenta y nueve peones y tardaron tres dias; [f. 371] y desde el pueblo de Totora hizieron Ileuar los dichos ~estos de coca a la <;iudadde La Plata con el ganado de los yndios; y fueron ~iento y veinte y seis cameros y sesenta y tres yndios cargados, y ellos propios cargaron con el dicho ganado; y fueron con Quintero criado del dicho don Alonso de Montemayor. EI segundo afio de 49 En la primera mita de los quatro meses cojieron de las dichas chacaras de coca dozientos y noventa y dos cestos de coca; para sacar estos dichos cestos de coca al dicho pueblo de Totora sacaron ~iento y quarenta y seis yndios y tardaron tres dias; y desde el pueblo de Totora hizieron Ileuar los dichos ~estos de coca a la <;iudad de La Plata con el dicho ganado de los yndios que fueron; [y fueron] ~ien cameros y quarenta y seis yndios cargados, y elIos propios cargaron el dicho ganado; [y fueron] con el criado llamado Quintero hasta la ~iudad de La Plata. . . En la segunda mita de los quatro meses coxieron de las dichas chacaras trezientos y quarenta y un cestos de coca; para sacar estos dichos cestos de coca al dicho pueblo de Totora sacaron ~iento y setenta y.vn peones y tardaron tres dias; y desde el dicho pueblo de Totora hizieron Ileuar los dichos ~estos de coca a la ~iudad de La Plata con el ganado de los yndios que fueron; [y fueron] ~iento y veinte cameros y cinquenta y vn yndios cargados, y eIlos propios cargaron el dicho ganado; [y fueron] con Quintero, criado de don Alonso de Montemayor. En la ter~era mita de los quatro meses cojieron de las dichas chacaras trezientos y sesenta y tres ~estos de coca, siendo su mayordomo el dicho Anton Gomez; para sacar estos dichos cestos de coca al dicho pueblo de Totora sacaron ~iento y ochenta y dos peones [f. 371v] y tardaron 3 dias en Ilegar al pueblo de Totora; y desde el dicho pueblo de Totora hizieron Ileuar los cestos de coca a la Villa de Potosi con el ganado de los yndios que fueron; [y fueron] ~iento y diez cameros y setenta y dos yndios todos cargados, y eIlos propios cargaron el dicho ganado; [y fueron] con Quintero, el criado del dicho don Alonso de Montemayor, hasta la Villa de Potosi. El ter~ero afio de 50 En la primera mita de los quatro meses cojieron de las dichas chacaras de coca, siendo su mayordomo en ellas vn espafioillamado Neto 0 Nieto, dozientas y veinte y dos ~estos de coca; para sacar estos dichos ~estos de coca al dicho pueblo de Totora, sacaron ~iento y onze peones y tardaron tres dias; desde el dicho pueblo de Totora hizieron Ileuar los dichos ~estos de coca a la villa de Potosi con el ganado de los yndios que fueron; [y fueron] setenta cameros y quarenta y vn yndios todos cargados, y eIlos propios cargaron el dicho ganado; [y fueron] con los dichos ~estos de coca con el dicho Quintero. En la segunda mita de los quatro meses cojieron de las dichas chacaras dozientos y setenta y quatro cestos de coca y a la sazon era su mayordomo Neto 0 Nieto; para sacar estos cestos de coca al dicho pueblo de Totora sacaron ~iento y treynta y seis peones y tardaron tres dias; y

ANDEAN PAST 5 (1998)

-156

desde el dicho pueblo de Totoral hizieron lIeuar los dichos ~estos de coca a la ~iudad de La Plata con el ganado de los yndios que fueron; [y fueron] ochenta y dos cameros y ~inquenta y quatro yndios cargados, y ellos propios cargaron el dicho ganado; [y fueron] con el dicho su criado lIamado Quintero. En la ter~era mita de los quatro meses cojie- [f. 372] ron de las dichas chacaras trezientos y dos ~estos de coca que su mayordomo era Neto; para sacar estos dichos ~estos de coca al dicho pueblo de Totora sacaron ~iento y ~inquenta y vn peones y tardaron tres dias; y desde el dicho pueblo de Totora hizieron lIeuar los dichos ~estos de coca a la ~iudad de La Plata con el ganadO' de los yndios que fueron; [y fueron] ~ient cameros y c;inquenta y vn yndios, todos cargados, y ellos propios cargaron el dicho ganado; [y fueron] con el dicho Quintero, su mayordomo. El quarto ano de 51 En la primera mita de los quatro meses cojieron de las dichas chacaras dizientos y tres ~estos de coca, siendo su mayordomo en los dichos valles de los yungas Pero Garcia; para sacar estos dichos cestos de coca al dicho pueblo de Totora sacaron c;ientoy dos peones y tardaron tres dias; y desdeel dichopueblode TotorahizieronlIeuarlos dichosc;estosde coca a la Uilla de Potosicon el ganado de los yndios que fueron; [y fueron] setenta cameros y treynta y vn yndios cargados, y ellos propios cargaron el dicho ganado; [y fueron] con vn espanol, criado del dicho don Alonso de Montemayor, del qual no se les acuerda el nombre. En la segunda mita de los quatro meses cojieron de las dichas chacaras dozientos y quarenta y dos cestosde coca, siendosu mayordomoPero Garcia;para sacar estos dichoscestosde cocaal dicho pueblo de Totora fueron c;ientoy veinte y vn peones y tardaron tres dias; y desde el dicho pueblo de Totora hizieron lIeuar los dichos ~estos de coca ala <;iudad de'[f. 372v] L~ Plata con el ganado de los yndios que fueron; [y fueron] ochenta cameros y treinta y vn yndios, todos cargados, y ellos propios cargaron el dicho ganado; [y fueron] con vn espanol, criado del dicho don Alonso de Montemayor, del qual no se sabe su nombre. En la tercera mita de los quatro meses cojieron de las dichas chacaras de coca dozientos y sesenta y dos cestos de coca, siendo su mayordomo Pero Garc;ia;para sacar estos dichos cestos de coca al dicho pueblo de Totora sacaron ~iento y treynta y vn peones y tardaron tres dias; y desde el dicho pueblo de Totora hizieron lleuar los dichos cestos de coca a la Uilla de Potosi con el ganado de los yndios que fueron; [y fueron] noventa cameros y quarenta y vn yndios, todos cargados, yellos propios cargaron el dicho ganado; [y fueron] con vn espanol, criado del dicho don Alonso de Montemayor que fue vn ano entero, y no se sabe su nombre. Todas las quaIes dichas partidas contenidas en esta declarac;ion y quenta de los dichos quipocamayos entregaron al dicho don Alonso de Montemayor y a sus mayordomos, asi en la Villa de Potosi como en la <;iudad de La Plata, como pares~en claramente que montan tres mil y quinientos y quarenta y siete c;estosde coca, todos cestos muy grandes como se usauan en aquel tiempo, que eran muy mayores que la de agora; y en aquel tiempo valian los c;estosde coca, puestas en esta uilla, a catorze pesos, y en Chuquisaca, a doze pesos. Ellicenciado Juan de Arevalo [firmado y rubricado]. Joan de Banos [firmado y rubricado] 3. AGl, Justicia 428, no. 2, folios 75-78. Informacion, Pocona, 16 de mayo de 1556. En el pueblo de Pocona, termino e juridic;ion de la <;iudad de La Plata, sabado, diez y seis dias [f. 75v] del mes de mayo de mil e quinientos y c;inquentay seis anos, <;ipio Ferrara, en nonbre de don Francisco de Mendoc;a,present6 ante mi esta comision desta otra parte y me pidio cunpla 10 en ella contenido, y en cunplimiento della, tome los testigos que presentare ~erca del valor de la coca en el asiento de Tiraque; e por mi vista Ie dixe traxese los testigos de quien se entendia en

157-

Julien: Coca on the Inca Frontier

ello aprouechar e yo los tomaria. Testigos Pero Gomez de Cayeres e Juan de Leon. Escreui, Juan Leon, escriuano. E despues de 10susodicho, en el dicho pueblo de Pocona el dicho dia, lunes, diez y ocho de mayo del dicho ano, el dicho <;ipioFerrara, en el dicho nonbre, present6 por testigo en la dicha rrazon a Diego Lopez de Aguilera, que dixo rresidir vna legua de Tiraque, y a Gaspar de Collayos e a Diego Valera e a Pablos [sic] de Baluerde e a Niculas Ortiz, estantes en el dicho pueblo, de los quales e de cada vno dellos tome e rresyibi juramento en (orma de derecho por Dios e por Santa Maria y por las palabras de los santos evangeIios; e por la senal de la cruz depusieron, sus manos derechas, so cargo del qual prometieron de dezir verdad de 10 que supiesen e les fuese preguntado en este caso que son presentados por testigos; e 10que dixeron e depusieron por SIe sobre sl, secreta e apartadamente, es 10siguiente. [f. 76] El dicho Diego Lopez de Aguilera, testigo, jurado en forma de derecho e preguntado por el tenor del dicho mandamiento, dixo que 10 que sabe es que este testigo a que rreside [debe leer: este testigo rreside] e abita en el dicho asiento de Tiraque, contratando alIi e viendo contratar en coca de los yungas que se trae a Tiraque, que es el deposito donde se trae la dicha coca a Pocona e a Poxo e a los demas rrepartimientos de quatro meses a esta parte, poco mas 0 menos; e que a bisto vender e se vende e conpra cada yesto hordinariamente del dicho tienpo a esta parte a ocho pesos en corriente por menudo; e que no se vende por junto porque los encomenderos e por parte de los ofiyiales rreales se lleua a vender toda a la <;ibdad de Potosi; e que, si por junto se oviese de bender, por ser en cantidad Ie paresye a este testigo que se benderia a siete pesos e medio 0 a siete pesos y ducado en corriente, pero que no se uende por junto sino de la manera que dicho tiene, llebandose en cameros a la <;ibdad de la Plata e a Potosi donde los ofyyiales rreales la suelen vender en almoneda e los encomenderos; e que esta es la uerdad para el juramento que hizo; e firm610. D~egoLopez de Aguilera. [En el margen: Testigo] Gaspar de Collayos, estante al presente en este pueblo de Pocona, testigo presentado en [f. 76v] la dicha rrazon, aviendo jurado en forma de derecho e siendo preguntado por el tC?nor del dicho mandamiento, dixo que 10 que sabe es que este testigo rreside en este pueblo de Pocona hordinariamente de seys anos a esta parte, poco mas 0 menos, tratando en coca, conprandola e vendiendola aqui y en el asiento de Tiraque, que es de Iii coca de los yungas que los yndios deste rrepartimiento dan en tributo; e que la a visto vender, e vendi[en]dola e conprandola aqui y en el asiento de Tiraque que es de la coca de los yungas que los yndios deste rrepartimiento dan en tributo e que la a visto vender e vendidola e conpradola [asl, repetido] este testigo, a diuersos preyios en Pocona y en Tiraque; y en el dicho asiento de Tiraque vale puesto alli en partidas a ocho pesos y medio cada yesto, y en este pueblo de Pocona, a ocho pesos en plata corriente; e que los encomenderos destos pueblos de Pocona e Poxo e Totora 10hazen lleuar a la villa de Potosi a uenderlo por ser en cantidad e no auer quien 10conpre por ser tanta; e a vezes se bende en la dicha Villa [de La Plata] e Potosi a menos, e pagandose los fletes a como se conyierta; e que como dicho tiene los offiyiales rreales y encomenderos no 10venden aqui sino en la dicha Villa e asiento, como 10a visto acosti.mbrardel dicho tienpo a esta parte; e que esta es la verdad para el juramento que hizo; e firm610. Gaspar Collayo. [f. 77] [En el margen: Testigo] Pablo del Valuerde, yndio ladino, natural que dixo ser de Tangara, que es en los terminos de la prouinyia de Piura, y criado de Lorenyo de Aldana, testigo presentado en la dicha rrazon, aviendo jurado en forma de derecho e seyendo [sic: siendo] preguntado por el tenor susodicho, dixo que 10que sabe es que al presente a bisto venderse y conprarse coca en este pueblo de Pocona y en el asiento de Tiraque de 10 de los yungas donde este rrepartimiento 10 suele dar e poner; y 10a visto conprar e vender por yestos a ocho pesos y medio en plata corriente, esto de yinco 0 seys meses a estaparte; y este testigo 10 a conprado a este preyio; e por junto e partidas no 10 a bisto vender porque por parte de los ofYyiales del rrey y los en-

ANDEAN PAST 5 (1998)

-158

comenderos destos pueblos 10hazen lleuar a vender a la Billa [de La Plata] e a Potosi e no 10 venden aqui y esto se a acostumbrado e acostunbrado [asl]; e que esta es la uerdad para el juramento que hizo, e firm610. Pablo de Valuerde. [En el margen: Testigo] Diego Valera, estante en este pueblo de Pocona al presente, testigo presentado en la dicha rrazon, aviendo jurado en forma de derecho e seyendo [sic: siendo] preguntado por el tenor susodicho, dixo que sabe e a bisto que en el asiento de Tiraque el pre~io [f. 77v] hordinario venderse partidas de coca a pres~io cada sesto de seys pesos ensayado de la que sale' de los yungas de los valles de Chuquioma y Chamorro, esto de vn afio a esta parte; y a este testigo se Ie dio para en quenta de ~ierto salario al mismo pres~io, puesto en el dicho Tiraque, del tienpo que siruio en la administra~ion de Pocona; e que sabe que la coca que desto a dado Pocona a los officiales rreales la an lleuado sienpre al asiento de Potosi a venderse por ser partidas en cantidad e por rrazon desto no hallar quien 10conpre, e 10mismo encomenderos de otros pueblos que dan de Ia dicha coca, porque este testigo 10a uisto; e que esta es la verdad para el juramento que fizo; e firm610. Diego de Balera. [En el margen: Testigo] Nicolas Ortiz, estante al presente en este pueblo de Pocona, testigo presentado en la dicha rrazon, aviendo jurado en forma de derecho e siendo preguntado por el tenor susodicho, dixo que este testigo a que rreside en el asiento de Tiraque, ado[nde] viene la coca de los yungas deste rrepartimiento de Pocona e otros por. Su Magestad para rres~ebir1ay enbiarla a los ofic;ialesrreales, esto de dos meses a esta parte; e que en el tienpo sabe e a bisto que se bende y conpra la coca en el dicho Tiraque a presc;iode ocho pesos e dos tomines [1 tomin = .125 pesos] e a ocho pesos [f. 78] Ymedio en corriente cada c;esto,e que esto, algunos ~estos particulares que a bisto vender, e que, si en cantidad fuese, Ie paresc;eque baxaria en cada c;estovn peso y mas porque la misma coca que del dicho asiento de Tiraque se lleua a Potosi, como es cantidad, la a bisto vender por los ofyc;ialesrreales en almoneda a siete pesos ensayados e a mas, y esto por ser en cantidad; e aca.D.O se abria quien en tanta conprase sy no fuese en gran baja e que no sabe ni a bisto que los dichos ofic;ialesrreales la ayan vendido en el dicho Tiraque; e ansi es publico e notorio y ellos y los encomenderos de los pueblos desta comarca 10venden alli en la Billa la que no se vehderia sino en baxa grande por ser en tanta cantidad, como es; e que esta es la uerdad para el juramento que hizo; e firm610. Niculas Ortiz. Juan Leon, escriuano. ' E de todo esto, en como pas6 de pedimiento del dicho <;ipio Ferrara, en el dicho nonbre, Ie di de 10susodicho testimonio como ante mi pas6; que es fecho e pas6 en el dicho dia e mes e afio susodichos. Testigos Pero Gomez de Cac;erese Juan de Leon, escriviente. Ante mi, Juan de Leon, escriuano

159-

Julien: Coca on the Inca Frontier

Pacific Ocean
E2I Mesothennic Valleys
I

200km

Figure 1. Mesothermic valleys east of the eastern Andes (after Saignes 1986). Note: solid line shown to the east of the valley region notes boundary between the foothills and the true lowlands.

ANDEAN PAST 5 (1998)

-160

. Tiraque
3200m

.Arepucho
1300m

Chuquioma .'

Estancia
Monte. Puncu
~

Canela
1400m

Chimboata. 2600m

Tiraque
2600m

Pocona 3100m

Chico. Laymitoro

Laymitoro

1 20km

Figure 2. Pocona and the Yungas of Chuquioma. Note: The road shown approximates the course of the main road in the 16th century.

CREATING A RUIN IN COLONIAL CUSCO: SACSAHUAMAN AND WHAT WAS MADE OF IT

Carolyn S. Dean
University of California, Santa Cruz

In 1534, Pedro Sancho de la Hoz (1968 [1534]:329), one of the first Europeans to see the capital city of the Inca (Inka), praised a certain structure saying, "neither the bridge of Segovia nor any of the buildings that Hercules or the Romans built are so worthy of being seen as this."1 The monument he admired was Sacsahuaman (Saqsaywaman), an Inca architectural complex perched atop a hill overlooking the city of Cusco in the southern Andean highlands (Figure 1). Sancho's hyperbole would later reverberate in the writings of dozens of Spaniards.2 Yet, despite their admiration of Sacsahuaman, the Spaniards dismantled most of the complex in the sixteenth and seventeenth centuries. In s~ doing, they created a ruin. While their physical control of Sacsahuaman was unquestioned after the failed indigenous uprising of 1536, the colonists could not maintain interpretive control over what remained of the site and its history. During the colonial period the ruin hosted rival histories, and around its monumental skeleton competing elements of colonial society fleshed out differing versions of the "conquest" of the Inca's capital. Thus, in the colonial period the physical ruination of Sacsahuaman was accompanied by the creation of histories which made those ruins meaningful. In examining several of these histories, it will become apparent that Sacsahuaman, as a ruin, was a metonym for the control of Cusco, if not all of Peru. While it is not my intention to recapitulate all that is known of and speculated about Inca Sacsahuaman, some general discussion of the
1 Sancho (1968 [1534]:329) wrote, "Los espafioles que las ven dicen que ni el puente de Segovia, ni otro de los edificios que hicieron Hercules ni los romanos, no son cosa tan digna de verse como esto." 2 Here "Spaniard" is used to refer both to those who were born in Spain and individuals of Spanish descent.

monument provides the requisite foundation for an analysis of Sacsahuaman in the colonial period -- that is, Sacsahuaman as a ruin. Sacsahuaman, bounded by a steep incline on one side and a flat plain on the other, consisted of 'numerous structures. Most prominent in its plan are three terraces which restricted access to edifaces on the upper level (Figure 2). The zig-zagging terrace walls are characterized by polygonal masonry featuring dressed megaliths of irregular shape, each of which has been worked so as to fit precisely in its place (Figure 3). While much of these megalithic walls can still be seen, only the foundations of the rest of the complex remain. There were numerous rooms and a~ many as three towers in the elevated central portion.3 The largest tower, called Muyuc Marca, was circular in plan.4 Sancho (1968 [1534]), in describing Sacsahuman before it was "ruined," reported that it was,
. . . of earth and very handsome stone with large windows that look upon the city and make it appear more handsome. Inside of it are many rooms and [there is] a principal tower in the center made in the manner of a round tower with four or five guardrooms one atop the other; the lodgings and rooms inside are smaH.5

3 Both Pedro Pizarro (1986 [1571]:104) and Cieza (1959 [1553]:154), however, report that Sacsahuaman had just two towers. 4 According to Gonzalez Holguin (1901 [1608}:211, 236) muyu means round and marca can be translated as protector, as well as city, town or region. Garcilaso (1966 [1609}:468-469) translates Muyuc Marca [M6yoc Marca] as "the round fortress" and names the other two square towers as Sacllac Marca and Paucar Marca. 5 Sancho (1968 [1534}:329) wrote, It.. . hay una fortaleza de tierra y de piedra muy hermosa; con sus ventanas grandes que miran a la ciudad y la hacen parecer mas hermosa. Hay dentro de ella muchos aposentos y una torre principal en medio, hecha a modo de cubo

ANDEAN PAST 5 (1998):161-183.

ANDEAN PAST 5 (1998) More~ though perhaps less accurate, information about Sacsahuaman is provided by Garcilaso de la Vega who was the son of an important Spaniard and a woman of royal Incan ancestry. After moving to Spain~he wrote a version of Inca history based~in part~on accounts of prehispanic Peru told to him by his Inca relatives when he was a boy in Cusco. Dubbing himself "EI Inca," he paints a rather glorious picture of his Andean ancestors. As a preeminent Inca monument~ Sacsahuaman accounts for considerable descriptive prose in his chronicle. Of the decoration of the tower of Muyuc Marca he wrote~
All its walls were adorned with gold and silver, with animals, birds, and plants imitated from life and fitted into the wall, serving as a kind of tapestry. (Garcilaso 1966 [1609]:469)

-162
shields, feathers, skins of animals and birds, coca, bags of wool, a thousand kinds of jewels; in a word, everything anyone had ever

heard of was in it.

Cieza thus suggests that it was both a temple and a storage facility or treasury. Both Sancho (1968 [1534]:329-330) and Pedro Pizarro (1986 [1571]:104) describe Sacsahuaman as a'
depository. All three functions

-- temple,

de-

pository~ and stronghold -- may have been combined. Garcilaso (1966 [1609]:469) explains that " [Sacsahuaman] was considered a house of the Sun for arms and warfare" and that in it was stored clothing and footwear for the Inca military. From these accounts, Sacsahuaman was a military complex dedicated to the imperial solar cult and the concomitant agenda of an aggressively expanding state. It has been suggested that Sacsahuaman was' built after 1440 and completed in the first decades of the sixteenth century (Gasparini and Margolies 1980:19~). Nearly all the sixteenth- and seventeenth-century documents that discuss Sacsahuaman, report that it was begun during the reign of the ninth Inca ruler (pachacuti/Pachakuteq) or his son, the tenth ruler (Inca Yupanqui or Tupac Inca Yupanqui).? Archaeology has been unable to determine the age of Sacsayhuaman~however, and so these historical accounts should be regarded as tentative at best.8 Many chroniclers report
7 While most chroniclers maintain that Pachacuti was another name for Inca Yupanqui, Garcilaso (1966 [1609]:463,471) holds that they were distinct individuals: Pachacuti was the ninth Inca ruler, Inca Yupanqui was the tenth, and Tupac Inca Yupanqui was the eleventh. He claims that the construction of Sacsahuaman began under Inca Yupanqui although he concedes that Pachacuti may have planned the work and collected much of the stone. A number of chroniclers agree with Garcilaso that the structure was begun by Pachacuti's successer (e.g., Collapifia, Supno y otros Quipucamayos [1974 {l542-75}:40] and Esquivel y Navia [1980 {ca. 1749}:40-43]). See Angles Vargas (1990:16-24) for a summary of references to the construction of Sacsahuaman in colonial chronicles. 8 Zuidema (1983:89) argues that the building of Sacsahuaman was credited to Tupac Inca Yupanqui by Pedro Sarmiento de Gamboa who was writing for Viceroy Toledo in 1572. According to Zuidema, Sarmiento wanted to characterize Cusco under Pachacuti as a city without a "head" (i.e., a legitimate ruler). Upon the

In addition to this elaborate and intricate decoration~ Garcilaso also described extensive subterranean passages beneath Sacsahuaman; neither feature is evident in the ruins. Spaniards underscored the use of Sacsahuaman as an Inca.stronghold by consistently referring to it as "the fortress" (la jortaleza).6 Several chroniclers maintain, however, that Sacsahuaman was a temple of the Inca patron, the sun (Herreray Tordesillas 1945 [1615]:73; Cieza 1959 [1553]:153; Garcilaso 1966 [1609]:469; Contreras 1982 [1649]:4). Pedro de Cieza de Leon (1959 [1553]:153, 1985 [1553]:147)~one of the earliest chroniclers of Peru, and one of the most reliable, calls it a royal house of the sun ("Casa real del Sol") and says it was conceived as a
house of the sun which should surpass everything done until then, and that it should house everything imaginable, such as gold and silver, precious stones, fine garments, arms of all the types they used, materials of war, sandals, con cuatro 0 cinco cuerpos, uno encima de otro; los aposentos y estancias de adentro son pequefios." 6 Gasparini and Margolies (1980:280-281) suggest that the Spaniards identified Sacsahuaman as a fortress because its masonry reminded them of medieval European strongholds. They conclude that many of the Inca structures called fortresses by the Spaniards were actually religious edifices designed to protect sacred space.

163that the work was continued by succeeding rulers being finished, or nearing completion, at the time Francisco Pizarro and his company arrived in the Andean area in 1532. Most agree that it had been in construction for over fifty years.9 The period of construction was lengthy and may have continued until the time of the conquest. Nevertheless, the destruction of Sacsahuaman began soon after the Spaniards settled in Cusco. Officially "founding" their city of Cusco in 1534, the Spaniards followed well known precedents including, of course, the founding of Mexico City on the ruins of the Mexica (Aztec) capital, Tenochtitlan. Such Spanish municipalities made use of the native labor and materials close at hand. In the case of Cusco, the Spaniards used the Inca's central plaza (the Haucaypata) as their own plaza' mayor and positioned their cathedral on this square. All around the central precinct Spanish edifices were erected on the foundations ofinca structures. While they had no immediate plans for the site of Sacsahuaman, its masonry was used in the building of Spanish Cusco.' H~nce, in the colonial period, Sacsahuaman, the Inca's multipurpose fortress, depository, and temple, served a new function.-- that of quarry. Apparently rumors of hidden treasure buried in Sacsahuaman also encouraged destructive excavations (Cieza 1959 [1553]:155; Garcilaso 1966 [1609]:471). Thus, within a few decades, the ruins of Sacsahuaman were literally created. In Garcilaso's words,
The Spaniards. . . demolished [the fortress] to build private houses in Cusco. And to save themselves the expense, effort and delay with which the Indians worked the stone, they pulled down all the smooth masonry in the walls. There is indeed not a house in the city

Dean: Sacsahuamiin
that has not been made of this stone, or at least the houses built by the Spaniards. The large slabs that formed the roof of the underground passages were taken out to serve as lintels and doorways. The smaller stones were used for foundations and walls, and for the steps of the staircases they sought slabs of stone of the size they needed, pulling down all the stones above the ones they wanted in the process, even though there might be ten or twelve rows and many more.

accession o( Tupac lnca Yupanqui, Cusco gained a metaphoric head; Sarmiento, who followed Juan de Betanzos in likening Cusco to a puma's body, then dubbed Sacsahuaman its topographical head. 9 For varying estimates of the construction period see Garcilaso (1966 [1609]:471-72), Esquivel y Navia (1980 [ca. 1749]:40-43), Collapifia, Supno, y otros Quipucamayos (1974 [1542-75]:64).

While Cusco's cabildo prohibited the removal of stones from Sacsahuaman in 1561 (Angles Vargas 1990:23), their edict had little apparent effect. That the city's corregidor commented in 1571 that Sacsahuaman could supply enough dressed stones to build four churches like those of Seville, suggests that Sacsahuaman continued to be an exploited resource for dressed stones (Polo de Ondegardo 1916 [1571]:107). Other officials and chroniclers affirm extensive usage of stones from Sacsahuaman in the building of Cusco (see e.g., Sarmiento 1943 [i572]:137 and Toledo 1904 [1571]). The friar Murua (1986 [160011]:500) claimed that all Spanish structures built in Cusco used stone from Sacsahuaman. He also explained that the only reason any stones were left standing at the site was because the large boulders would have been too expensive to move and would have required the labor of too many natives.lO Contracts from the sixteenth and seventeenth centuries provide concrete evidence for the usage of stones from Sacsahuaman in the building of Spanish Cusco's symbolic center; documents dated October 6, 1559 and February 19, 1646 indicate that Sacsahuaman was the source of stone used in building the cathedral of Cusco (ValcarceI1934:176; Cornejo 1960:147). An eye witness account (Cobo 1990 [1653]:229230) of the construction of Cusco's cathedral details the Andean method of lithic construction employed by native laborers -- the natives this chronicler watched were undoubtedly us10 Murila writes, "Es de suerte que todos los edificios modemos que despues se han hecho en la ciudad por los espafioles, han alido de la piedra de alH, aunque a las piedras grandes y toscas no han llegado, por no poder lIevarlas a otro lugar sin costa excesiva e infmito trabajo de los indios."

ANDEAN PAST 5 (1998) ing some of the same stones that their ancestors had used to build Sacsahuaman.11 Although much of Sacsahuaman was dismantled in the first half of the colonial period, the most striking and impressive part of the complex remained -- its polygonal megalithic walls which feature stones measuring over fifteen feet in height and weighing over two hundred tons. Both Andeans and Spaniards esteemed the great size of these stones and the labor they represented. There are many Andean legends concerning boulders coming to life and of people being turned into rocks. The large boulders at Sacsahuaman also inspired legends. There was, in particular, one megalith --possibly part of a nearby rock outcrop -- which has been associated with Sacsahuaman in native myths. This boulder was called Saicusa [Saykuska] which can be translated as "weary" or "tired" (Garcilaso 1966 [1609]:464, Gonzalez Holguin 1901 [1608]:330). According to ~ost accounts, this megalith was being dragged by thousands of laborers into position when it grew so weary that it began to weep tears of blood. The Inca left the distressed stone where it lay. The native chronicler Felipe Guaman Poma de Ayala (1988 [1615]:159 [161]) illustrates this boulder being moved by an Inca captain (Inga Urcon) who supervises the effort while the stone sheds tears of blood (the gloss reads "the stone cried blood", "l1or6sangre la piedra") (Figure 4). Maarten Van de Guchte (1984) has analyzed six versions of the weary stone myth recorded in colonial chronicles; he argues convincingly that the stone symbolized the completion 9f work on Sacsahuaman.12 This
II Cobo (1990 [1653]:229-230) describes the method of dragging large stones to construction sites and then the use of earthem ramps to place them, saying, "I saw this method used for the Cathedral of Cusco which is under construction. Since the laborers who work on this job are Indians, the Spanish masons and architects let them use their own methods of doing the work, and in order to raise up the stones, they made the ramps mentioned above, piling earth next to the wall until the ramp was as high as the wall." See Lee (1986) and Protzen (1985, ]986) for a more detailed analysis of Inca building methods. 12 Van de Guchte (1984:548) identifies the Weary Stone as that which is located at the place referred to as

-164

legend also directs .attention to the immensity of the task entailed in transporting such rocks from both local and distant quarries, especially because they could be unwilling transients. On one level, then, Sacsahuaman's megalithic walls surely underscored the power of the Inca state. The complex was built by la-' borers from throughout the empire who worked periodically for the state as a form of tribute. Sacsahuaman manifested the extensive resources of a state that could mobilize the labor required to cut, dress, transport, and build a structure of these (apparently sometimes recalcitrant and always potentially animate) Herculean rocks. Garcilaso (1966 [1609]:465, 469), calling Sacsahuaman the Inca "trophy of trophies," concludes that its primary purpose was to impress:
The immensity and majesty of this monument seems to show that the Incas intended to demonstrate the greatness of their power. It was indeed made to impress rather than for another reason. They also wished to exhibit the skill of their masons and builders, not only in dressing the smooth masonry (which the Spaniards are never tired of praising) but also in rough stonework, which they executed with no less mastery. They sought also to exhibit their knowledge of military science in the plan of the fortress, including everything necessary for its defence against the e1)emy (emphasis mine).

While we may doubt Garcilaso's assertion, we should note that the putative objective -- to impress and amaze -- continued to be accomplished even after much of the structure was dismantled, for the "ruins" of Sacsahuaman consist primarily .0 its megalithic terrace walls. There is no doubt that the Spaniards were truly impressed by the Inca masonry of Sacsahuaman. One after another, their descriptions echo the words of Sancho (1968 [1534]:329) who wrote,

la Chingana Grande today. It is found to the north and east of Sacsahuaman. Angles Vargas (1990:100-103, 179) disagrees, however, noting that the stone is in situ and has never been moved.

165The most beautiful thing that can be seen in the edifices of this land are these walls because they are of rocks so large that no one that sees them can say that they were placed here by human hands; they are so large, like pieces of mountains and cliffs. They are thirty palms high and as many wide and others 25 and others 15, but none is so small that they could be carried in three carts. These are not rocks of an even size but well fitted and joined one against another,13

Dean: Saciahuaman
That which is most admirable [about Sacsahuaman] is that these [rocks] are not cut regularly but are very irregular in size and in shape. They fit one against another with incredible joins and no mortar. All this was done with many people and great suffering in the labor because to fit one rock against another, as these are placed, probably necessitated repeated [fittings], most of them being neither equal [in size] nor complete [in shape]. 14
.

Like Sancho, later writers consistently comment on the enormous size of the stones, the amount of labor required, and the quality of masonry which needed no mortar. Two examples of similar prose--one from midcentury and one from the last decade of the sixteenth century--will suffice to make the point. Twenty years after Sancho, Ci~za de Leon (1959 [1553]:154-5) wrote:
There are stones of such size and magnificence in these walls that it baffles the mind to think how they could be brought up and set in place; and who could have cut them, for they had so few tools. Some of these stones are about twelve feet wide and over twenty long; others are as thick as an ox, and all so exactly set that a coin could not be inserted between them.

While Garcilaso (1966 [1609]:471)called

Sacsahuaman an "inadequately portrayed and insufficiently praised fortress," it consistently generated Spanish stupefaction. All manner of men, including a magistrate of Cusco and a viceroy of Peru, recorded words of amazement at the size of the stones and at the labor such a structure obviously required.ls Their com14 " La labor es extrafta y para espantar; y no usaban de mezcla ni tenian hierro ni acero para cortar y labrar las piedras, ni maquinas ni instrumentos para traellas, y con todo eso estan tan pulidamente labradas, que en muchas partes apenas se ve la juntura de unas con otras; y son tan grandes muchas piedras de estas, como estA dicho, que seria cosa increible si no se viese . . ." and "Y 10 que mas admira es que no siendo cortadas estas que digo de la muralla, por regia, sino entre si muy desiguales en el tamafto y en la facci6n, encajan unas con otras con increible juntura, sin mezcla. Todo esto se hacia a poder de mucha gente y con gran sufrimiento en el labrar, porque para encajar una piedra con otra, segun estan ajustadas, era forzoso proballa muchas veces, no estando las mas de ellas iguales ni Ilenas." 15 Polo de Ondegardo (1916 [1571]:106), who was first named corregidor of Cusco in 1558, estimated that some of the larger stones would have required twenty masons an entire year to dress a single one. He says, "cierto cosa maraviIIosa de ver en la fortaleza del Cuzco, que ay piedras tan grandes y ajuntadas, que yo e estado presente delante de canteros y se espantan como se podian suvir sin artificio, avn que el edificio no es muy alto, a labrallas e ponellas de suerte que bynyesen vien; entieendese cierto aue ay piedras de aquellas que vna sola seria menester trauajo de veynte personas un afto entero para desvastarla, y vistas las obra (sic) que alH ay' debaxo de la tierra y encima al derredor del Cuzco. . ." (ibid.). Viceroy Toledo (1904 [1571]:174) also found it difficult to believe that Sacsahuaman was made by the "strength and industry of men." ("parece imposible haberlo hecho fuerza ni industria de hombres. . .") as did Pedro Pizarro ("Auia piedras . . . tan grandes y tan gruesas, que parecfa cosa ymposible habellas puesto manos . . .") (1986:104). For similar views dating to the seventeenth century see Murua (1986 [1600-11]:500); Mogrovejo de la Cerda (1983 [1690{l649}J:43); and Cobo (1990 [1653]:229).

Forty years, after Cieza, the Jesuit Joseph de Acosta (1962 [1590]:297) expressed similar wonderment:
[T]he labor is extraordinary and amazing and they did not use mortar nor did they have iron nor steel to cut and dress the stones, nor machines nor instruments to transport them, and notwithstanding all this [the rocks] are smoothly. worked so that in many parts one can scarcely see the joining of one [rock] with another. Many of the rocks are so large, as I have said, that it would be unbelievable if one had not seen it . . .

13 "La mas linda cosa que puede verse de edificios en aquella tierra, son estas cercas, porque son de piedras tan grandes, que nadie, que las vea no dira que hayan sido puestas alli por manos de hombres humanos, que son tan grandes como trozos de montafias y peftascos, que las hay de una altura de treinta palmos, y otros tantos de largo, y otras de veinticinco, y otras de quince, pero no hay ninguna de ellas tan pequefta que la puedan llevar tres carretas; estas no son piedras lisas, pero harto bien encajadas y trabadas unas con otras."

ANDEAN PAST 5 (1998) mentary underscores European technological superiority, however, as the authors wonder at what was achieved despite the Inca's lack of (iron) tools. Thus, their effusive praise is implicitly mitigated, and it should be noted that few Spaniards made any serious attempts at documenting Inca architectural methods or priorities (i.e. why they built the structure in the way, shape, and location they did). Clearly, most Europeans were content to marvel at the Andean "mystery" and leave virtually uninvestigated the Inca's achievement.16 Yet, the ruins of Sacsahuaman rested uneasily as an inexplicable relic of a bygone era; perhaps this is because so many had witnessed the process of ruination. Some Spaniards expressed distress, or at least ambivalence, at the destruction of the site. Cieza (1959 [1553]: 155-6), for example, argued that Sacsahuaman "should be preserved in memory 'of the greatness of this land." Then, perhaps knowing that that reason was insufficient to impress most of his countrymen, he adopted a practical tact adding that Spaniards could use it as a fort at very little cost to themselves (seeing that it was already built). "L*ewise, Pedro Sarmiento de Gamboa (1943 [1572]: 137) said the ruins inspired "great pity" ("Hace gran lastima a los que agor~ [sic] ven las ruinas della.") In the late eighteenth century, it was noted that the wonderous ruins would be more famous if Spanish "ambition" (ambici6n) had not led to the extraction of so many stones for use in new Cusquefio edifaces (Castro 1978 [1788]:39-40). While "ambition" is an ambiguous tenn (yet one used frequently in chronicles which address conquest activities), Garcilaso (1966 [1609]:471) is more straightforward when he writes
The Spaniards . . . were as if envious of the remarkable achievements of the Incas, and not only failed to preserve the fortress, but even demolished it.

-166 Thus Garcilaso suggests that, while the stones of Sacsahuaman were certainly convenient for the Spaniards of colonial Cusco, the majestic complex was more than a local quarry. It was the target of envy-engendered spite. While Garcilaso's accusation is not easily supported, it is significant that the Spaniards, of Cusco did more than partially dismantle "the fortress". They adopted Sacsahuaman as their symbol of the conquest of Cusco and the Inca. In 1540, King Charles V awarded the city a blazon featuring a castle of gold on a red background surrounded by eight condors (Figure 5). In Spanish heraldry, the castle not only represented the kingdom of Castille, but was a generic reference to a fortified city, and was, therefore, a symbol of warfare. The grant itself, however, suggests that the castle on Cusco's coat of arms referred specifically to the . Inca architectural complex of Sacsahuaman and to a specific moment in the history of the conquest. The grant reads,
In Madrid on July 19, 1540, a grant of arms is awarded to the city of Cuzco in which the arms that are given consist of a shield inside of which is a golden castle on a field of red to remember that this city and its castle were conquered by the might of arms in our service. It shall have a golden border of eight condors that are great birds, resembling vultures, that they have in the province. of Peru, as a reminder that at the time this city was won, these birds descended to eat the dead that died there 17

Referred to in this grant is the battle for Sacsahuaman which was fought in 1536 between Spaniards and rebellious Incas under the command of Manco Inea (sometimes called ManeoII), the son of the last undisputed ruler
17 The grant reads, "En Madrid XIX dias del mes de julio de MDXL afios se despacho un previllegio de armas para la ciudad del cuzco en que se Ie dieron por armas un escudo que dentro del este un castillo de oro en campo colorado en memoria que la dha ciudad y el castillo della fueron conquistados por fuerza de armas en nro. servicio e por orla ocho condures que son unas aves grandes a manera de buytres que ay en la provincia del peru en memoria que al tiempo que la dha ciudad se gano abaxaron las dichas aves a comer los muertos que en el1amurieron los quales esten en campo de oro." (Montoto de Sedas 1928:75).

16 The reluctance to recognize Inca achievement has continued into the present century; both Fawcett (1953) and von Diiniken (1970) have authored fanciful theories to account for the megaliths of Sacsahuaman. For a plausible method of masonry construction, see Lee (1986).

167of the Inca empire, Huayna Capac.18 The retaking of Sacsahuaman by the Spaniards and their Andean allies was held to be a crucial success during this rebe11ionin which much of the city was destroyed. The rebels who held Sacsahuaman were well-protected behind its walls. Eyewitness Pedro Pizarro (1986 [1571]:127-134) described in his memoirs how the rebels were able to inflict great injury on the besieged from Sacsahuaman due to its position above the city: using slings they threw heated stones onto the highly combustible straw roofs of Cusco and rocks onto those in the plaza below. Because the imperiled Spaniards recognized Sacsahuaman as a strategic vantage point, they exerted considerable effort recapturing it. In the process, the Spaniards and their native allies suffered great losses. Among the casualties was Juan- Pizarro, the brother of Francisco, who had led the assault on Sacsahuaman. While Juan was killed and the casualties were heavy for both sides, the rebels were eventually overcome. Although the siege continued for many months, this victory was held to be the turning of the tide. Possession of Sacsahuaman thus became a metaphor for the possession of Cusco; and the ruination of Sacsahuaman referred metonymically to the conquest of Peru. From the grant of 1540 we know that the generic tower on Cusco's coat of arms refers specifically to Sacsahuaman and the battle of 1536. While Sacsahuaman was literally dismantled during the next century, figuratively the "fortress" faced reconstruction. On Cusco's blazon the Inca's "fortress" has been Europeanized. The pictorial Sacsahuaman constructed by the Spaniards is a stereotypic medieval European fortress similar to the castle representing Castille on the royal Spanish blazon and those adorning hundreds of other coats of arms. Thus Cusco is pictorially characterized as just another fortified city in the expanding Spanish domain. But Cusco was not a medieval European town and its "for-

Dean: Sacsahuaman tress" was not constructed by Europeans. The blazon's stereotypic tower, which denies the Andean-ness of Cusco, figuratively converts the city from Incaic to Spanish and so participates in ongoing hispanicizing efforts. The tower on Cusco's coat of arms is the articulation of the Spanish desire to convert Cusco into a Spanish city. The juxtaposition of the empirical reality of Sacsahuaman (an Inca structure in the first phases of ruination) and the imagery on Cusco's coat of arms (a whole and perfect Europeanized tower) advanced the colonizer's cultural agenda. To the Spaniards Sacsahuaman represented more than just military conquest and control, however, for the so-called fortress itself was more than just an Inca stronghold. Considering its associations with both the Inca's military might and its solar cult, Sacsahuaman symbolized, as no other structure in Cusco could do, both military and religious triumph. Not long after the battle of 1536 was won, the legend spre~d that Santiago (St. James the Greater), the patron saint of Spain, had ridden to the aid of the Christians astride his great white charger. The Mercedarian Mun'la provides one version of the legend. He records what he heard "Spaniards and Indians tell as a thing known for certain and true" that during the battle "there appeared someone on a white horse, fighting on the side of the Spaniards and killing many Indians so that they all fled from him" (MurUa 1986 [160011]:235). The Spaniards later realized that this had been Santiago. This same chronicler also relates how the Virgin Mary came to the aid of the besieged Spaniards by flinging sand into the eyes of the rebel Andeans. He writes,
Also the Indians say that when they were below fighting and had cornered the Spaniards, a woman blinded them with grains of sand and they could not stand before her but fled from her and it is presumed to have been Our Lady, Intercessor and Mother of Sinners. . .19

18 Manco Inca had been named ruler with the blessings of the Spaniards in 1533.

19 Murna (1986 [1600-11]:235-6) writes, ". . . quiero referir 10 que he ofdo contar a espafioles e indios por cosa constante y verda<.lera, y.es que dicen que andanzo en el mayor conflicto de la pelea apareci6 uno de un caballo blanco, peleando en favor de los espafioles y haciendo en los indios gran matanza, y que todos hufan del. Muchos espafioles tuvieron por cierto que era

ANDEAN PAST 5 (1998) Pierre Duviols (1962) has analyzed the variations in the numerous accounts of these two miraculous apparitions as has Henrique Urbano (1992). Duviols demonstrates that these legends were codified in the late sixteenth century based on rather oblique references in writings of Cieza and Acosta.2o The legends are particularly confused on the location and date of Mary's miraculous appearance.21 By the late sixteenth century, it was generally believed that while Santiago appeared at Sacsahuaman, Mary descended at the suntur wasi, a circular tower that was a focus of Inca ritual and where the Spaniards had taken refuge. According to legend, the Spaniards decided to build Cusco's cathedral on this site in memory of Mary's descent and she was elected as its patron. When a new and larger cathedral was built along side the original (finished in the seventeenth century), the earlier edifice was converted into a chapel. This chapel was called EI Triunfo ("the Triumph") in memory of the victory won with Mary's aid.22
Mansio Sierra [Mancio Sierra de Legfzamo], conquistador principal del Cuico,.y que despues, averiguando el caso, hallaron que Mansio Sierra no habia peleado alii sino en otra parte y no habia otro que tuviese caballo blanco, sino el, y asf se entiende haber sido el Ap6stol Santiago, singular patr6n y defensor de Espatla el aue allf apareci6, por 10 cualla ciudad del Cuzco Ie tiene por abogado. Tambien se refiere por los indios que, estando abajo peleando y teniendo apretados en gran manera a los espatloles, una mujer les cegaba con putlados de arena y no podfan parar delante della, sino todos Ie hufan, la cual se presume haber sido Nuestra Setlora Abogada y Madre de los pecadores, que querrfa en aquel trance favorecer a los espafloles, y asf la Santa Iglesia del Cuzco la tiene por patrona y titular suya." 20 Urbano (1992:241) notes that while some chroniclers not only recounted, but amplified the apparition legend, others (whom he does not name) tried to explain away the miracle by suggesting that images of the Virgin Mary and Santiago, which adorned banners that were raised by Hernando Pizarro at Sacsahuman after their victory, were misinterpreted as the real thing. 21 Writing in the mid-eighteenth century, Cusco historian Esquivel y Navia (1980 [ca. 1749]:99), however, is very specific when he reports that "Ia Reina de los Cielos" appeared at night on May 21, 1536, the seventeenth day of the eight month siege. 22 Despite the popular legend, it is doubtful that EI Triunfo is actually located on the spot formerly occu-

-168

It may be significant that both Mary and Santiago were said to have appeared at the site of well known circular Inca towers (Muyuc Marca, at Sacsahuaman, and the suntur wasi in the Haucaypata). Both towers, being tall and circular, were identified in the Spanish mind with castle fortresses and, therefore, with armed resistence and military triumph. Both, these Inca towers were destroyed in the decades following the broken siege and failed uprising. Significantly, however, while both monuments were ruined, only Sacsahuaman became a ruin. The cathedral and Sacsahuaman thus generate a dialogue between the Spanish and Christian center and the Andean and "pagan" periphery. As the cathedral was erected, Sacsahuaman was dismantled; stones from the latter were used to edify the former. The conversion of stones from Incaic to European architecture not only paralleled the religious conversion effort, but seemed to illustrate and actuate it. As a ruin, Sacsahuaman hosted memories of a victorious battle against paganism. Many Spaniards described Sacsahuaman as the product of satanic handicraft. Such a characterization not only denied Andean ingenuity and skilled masonry, but positioned the structure itself as an "enemy." Viceroy Francisco de Toledo (1904 [1571]:174), for example, writing to the king from Cusco on March 25, 1571, describes Sacsahuaman as "a thing in which the power of the devil and his subjects is well demonstrated. . ." ("es cosa en que se muestra bien el poderio del diablo.") Diego de C6rdova y Salinas (1957 [1651]:45-6), chronicler of the Franciscan Order in Peru, similarly suggests that Sacsahuaman represents devil's
pied by the suntur wasi. Garcilaso (1966 [1609]: 701) describes the suntur wasi and locates it as having been in the Haucaypata in front of the Amarukancha; further, he remarks that La Compatlfa in Cusco stands on the site. While Rowe (1990:92) is skeptical of Garcilaso's accuracy, Victor Angles Vargas (1978-83, v. 1:48-49, 96-97) suggests that there were two suntur wasis, one where Garcilaso places it and one where EI Triunfo is now. What concerns me here is not whether the legend is accurate in this regard, but rather what was believed to be true in late sixteenth- and early seventeenthcentury Cusco, who advanced particular histories, and what purposes those histories might have served.

169work. He wrote, "it makes one imagine that the devil helped in [building] that ediface" ("Ie hace imaginar que el demonio ayudo a aquel edificio"). Later, in the seventeenth century, one of the officials of Cusco's ecclesiastic council summed up the view of the ruins as a monument to the destroyed pagan past when he wrote, " [The fortress] was once dedicated as a house of the sun and in this time only serves as a witness of its ruin" ("Fue dedicada al principio para casa del Sol y en este tiempo, solo sirve de testigo de su ruina.") (Vasco de Contreras y Valverde 1982 [1649]:4). As Sacsahuaman was a product of devilish powers, its ruins were a testament to the Christian god's ultimate triumph. Sacsahuaman -- as a ruin -- thus came to symbolize Christian triumph in the Andes. Sacsahuaman was escorted by European historians into the great global ring where the Christian god meets and defeats all opponents. According to Co~dovay Salinas (1957 [1651]: 141) .
Satan was in this city [of Cuzco] as in Rome or Jerusalem until to his sorrow he was cast out and exorcised bY-the evangelizing ministers who destroyed his altars, demolished his temples and burned his huacas [shrines].23

Dean: Sacsahuaman nians, Egyptians, and even the Chinese.24 Using the tried and true recipe through' which Spaniards understood their relation to nonEuropeans, the procedure seemed simple: add Inca and stir. In the resulting confection, Andean ethnicities were implicitly and aggressively denied. Significantly, Spaniards equated the victory of Christianity with Span" ish victory. Catholicism and nationality were inextricably linked in their understanding of conquest history. Indigenous Cusquefios not only witnessed Sacsahuaman being reduced to ruin in order to raise the Christian temple, they were the laborers who, under Spanish supervision, dislodged the stones of Sacsahuaman and built the cathedral. If Sacsahuaman was built by the Incas (at least in part) in order to impress other Andeans, what impression must its reduction have made? What memories were invested in Sacsahuaman by native Andeans and what histories did they generate? While Spaniards left relatively full acounts of their notions about Sacsahuaman, Andean views are less recoverable. From the scant records that remain, it becomes clear that, in the complex, multiethnic society of colonial Cusco, Sacsahuaman became the focal point of rival histories. Like the Spaniards, Andeans invested the "ruins" of Sacsahuaman with memories of 1536. But their ways of remembering affirmed native identities and countered Spanish claims to cultural superiority. The native chronicler, Felipe Guaman Poma, who compiled and illustrated his chronicle during the first decades of the seventeenth century, emphasizes the actions of supernatural beings in the decisive battle for Sacsahuaman. Signifi24 Ignacio de Castro (1978 [1788]:23), priest and resident of Cusco (rector of the Colegio Real de San Bernardo and curate of San Jeronimo), for example, compares the roads built by the Inca favorably with the Great Wall of China and roads of Rome ("no son obra inferior a la gran muralla de la China; y aun hay Estrangeros, que dicen ser empresa superior a las mas celebradas de los Romanos. He goes on to discuss Sacsahuaman which he says merits the greatest celebrity of all Inca constructions ("Quisa merece mayor celebridad la Fortaleza del CUZCo.").
It)

Numerous chroniclers compared the Inca with ancient Rome. Just as Roman ruins occupy the peripheries of Spanish cities recalling a grand (but defunct), non-Christian empire so, too, does Sacsahuaman. In this way, the Andean landscape with its unique monuments was inscribed in the familiar pattern of European history, and the Inca were added to the standard stock of non-Christian or preChristian players. In addition to Romans, the Inca were compared (favorably) by various Spaniards to other non-Christians, including the Jews, Moors, Greeks, Chaldeans, Babylo-

23 "Estaba Satamis en esta ciudad como en su Roma 0 Jersalen, hasta que a su pesar fue echado y lanzado por los ministros evangelicos, y por ellos destruidas sus aras, derribados sus templos y quemadas sus huacas." C6rdova y Salinas goes on to say, "Yen su lugar colocaron la santa cruz y el reino de Cristo nuestro Dios y Sefior ocup6 10que el tirano tenia usurpado."

ANDEAN PAST 5 (1998) .cantly, Guaman Poma was not Inca and his chronicle contains considerable anti-Inca sentiment. For him, and very likely for other non-Inca Andeans, Sacsahuaman-as-ruin symbolized the demise of the oppressive Inca empire as well as their state religion. Guaman Poma was also critical of the Spanish occupation of Peru. It is thus not surprising that he minimizes both Spanish and Inca efforts during the battle and emphasizes the activity of Christian supernatural beings in that conflict. This focus on supernatural forces manifested in the battle deflects the question of cultural superiority and separates Christianity from any specific ethnic or national group. This distinction is essential; while Spaniards linked Spanish and Christian triumph (and, in fact, recognized no distinction), Andean groups emphasized the separateness of God and Spaniards. Such a separation was critical to Andean memories of the conquest period. Recall, for example, that when relating the legend of Mary's miraculous activities during the siege of.Cusco, the Spanish chronicler Mun'ia specified that the source of his account of her appearance was natives, not Spaniards. Clearly, natives had as much if not more at stake in "these legends than did Europeans. Indigenous Andeans may even have authored the account of Mary's descent. Guaman Poma (1988 [1615]:402 [404], 404 [406]) illustrated both Mary and Santiago coming to the aid of the besieged (Figures 6 and 7). The Spaniards are nowhere in evidence: we see only the deity and the defeated rebels. Guaman Poma (ibid.:403 [405]) gives Mary credit for breaking the siege by throwing dust in the eyes of the "infidel Indians" ("yndios ynfieles"). He writes
God and his blessed mother made the miracle to show mercy to the Spanish Christians, but it is better to say that the mother of God wanted more to show mercy to the Indians so that they be Christians and the souls of the Indians saved. . .25

-170 Of Santiago, Guaman Poma (ibid.:405 [407]) says,


Like a lightning bolt [Santiago] hurled down from heaven to the fortress of the Inca called Sacsahuaman . .. And as he fell to the ground
the Indians were frightened

lord Santiago descended to defend the Christians. . . [He] came with great destruction and many Indians died and he broke the siege of the Christians by the Indians that Manco Inca had ordered and the saint was accompanied by such a racket [of thunder] that came from him that the Indians were scared away.26

. ..

And so the

Guaman Poma goes on to say that the natives were eye witnesses to the miracle of Santiago and, as a result, they became his devoted followers. Thus, in Guaman Poma's eyes, the most significant result of these miraculous appearances was the religious conversion of Andeans by Christian supernatural powers. According to Guaman Poma (1988 [1615]:406 [408]), the victory at Sacsahuaman was the end of Inca soyereignty. He relates how the miraculous appearance at Sacsahuaman caused Manco Inca to flee:
And he left the kingdom and the crown, the mascapaycha [royal fringe] and chanbi [war club], to the lord emperor and king, our lord don Carlos of glorious memory who is in heaven and to his son don Felipe the second who is in heaven and to his'son don Felipe the third the king our 10rd.27

26 "Como rrayo cay6 del cielo a la fortalesa del Ynga


lIamado Sacsa Guaman

espantaron los yndios . .. Y anciauaj6el sellorSanctiago a defender a los cristianos. . " Venfa con gran destruyci6n y muerto muy muchos yndios y desbarat6 todo el serco de los yndios a los cristianos aue aufa ordenado Mango Inga y que lIeuaua el santo mucho rruydo y de ello se espantaron los yndios." 27 "Y dex6 el rreyno y corona, masca paycha [borIa real] y chanbi [porra de pelear] al senor enperador y rrey, nuestro senor don Carlos de la gloriosa memoria guesta en el cielo y a su hijo don Phelipe el segundo questa en el cielo y a su hijo don Phelipe el tersero rrey nuestro senor."

. ..

Y como cay6 en tierra se

25 "C6mo hizo Dios milagro para hazelle merced y su madre bendita a los espalloles cristianos, por mejor dezir que mas quizo hazer merced la Madre de Dios a los yndios porque fuesen cristianos y saluasen las animas de los yndios . . ."

171Thus political control of the Andean area is characterized as the result of divine intervention. Furthermore, mundane power shifted from Inca emperor to Spanish ruler. No mention is made of the local Spanish authorities whose presence in the Andes Guaman Poma opposed. If Guaman Poma speaks for other non-Inca Andeans, we may speculate that the ruins of Sacsahuaman were both a symbol of the Inca's defeat in 1536 and a memorial to the power of the Christian god. If the non-Incas of colonial Cusco held Sacsahuaman as an emblem of the conquest of the Incas, how did the Incas themselves understand the ruins? Speaking for the Incas who descended from the rebels is Titu Cusi Yupanqui (1973 [1570]:82-87), the son of Manco Inca (the leader of the rebellious Incas who organized the siege).28 Titu Cusi'was the ruler Vilcabamba. In his account of the battle for Sacsahuaman, he underscores the valient defense of the "fortress" by rebel troops (a 'fact confirmed by Spanish eyewitnesses including Pedro Pizarro) and'states,
This battle was bloody for both sides because of the many Indians that sided with the Spaniards.. Among. them were two of my father's brothers, one called Inguill and the other Vaipai and their many followers. . .29

Dean: Sacsahuamim damage" ("ffue el primero que entro en la ffrotaleza al tiempo que se tomo les hacia
mucho dafio . . ."). Thus the battle of Sacsa-

huaman is characterized as primarily Inca against Inca and as rebel troops against divine forces. In this light, the important outcome of 1536 was religious conversion. The dire political consequences of the loss are minimized: The ethnic Inca who descended from the allies of the Spaniards also viewed Sacsahuaman as the locus of divine activity and as the place where the Incas and the Christian god formed an important and lasting alliance. The Quipucamayocs (Collapifia, Supno, y otros Quipucamayos 1974 [1542-75]:64), Inca "historians" who relied on mnemonic devices to keep oral histories, reported, within a decade of the event itself, the version of the siege that would be held by those Incas who did not join Manco's rebellion.3o The Quipucamayocs were partisan to the Inca allies of the Spaniards; in fact, the second part of their chronicle is an exultation of Paullu Inca, the Inca leader of royal blood, who remained loyal to the Spaniards during the rebellion. They emphasize the aid rendered to Spaniards by natives of Inca and other ethnicities. Deserving special mention in their .account were four rebel Incas who changed sides at a crucial moment. The Quipucamayocs report
At this desperate time, God was served by the defection to the Christians of four Incas who were among the most important that Manco Inca had. They were Cayo Topa, don Felipe Cari Topa, Inca Paccac, and Uallpa Roca. With each one of them were great Indian squadrons. They gave great comfort to the Christians and later were seen with them and saw the need and hunger that existed. They ordered that a great quantity of food be smuggled into the city for the succor and sustenance of the Christians and Indians that were helping and aiding them.31 30 The Quipucamayocs were assembled at the orders of Governor Vaca de Castro and their reports were recorded with the aid of interpreters. 31 "A este tiempo tan trabajoso, fue Dios servido se pasaron a los cristianos cuatro ingas de los mas principales que tuvo Mango Inga, los cuales fueron Cayo Topa y don Felipe Cari Topa e Inga Paccac y Uallpa
.

of the exiled Incas(the followersof Manco)in


,

The battle, as he characterizes it, was lost due to the lack of Andean, and especially Inca, solidarity. More than any other factor, however, Titu Cusi Yupanqui (1973 [1570]:86) credits the successful storming of Sacsahuaman by-the Spaniards and their native allies to the appearance of Santiago who "was the first to enter the fortress and did much

28 When the siege failed, Manco retreated to ViIcabamba where he and his sons ruled an independant Inca state until 1572 when the Spaniards captured and later executed his successor, Tupac Amaro. 29 Titu Cusi Yupanqui (1973 [1570]:84) writes, "Fue esta batalla de vna parte y de otra, ensangrentada, por la mucha gente de indios que ffavoreszian a los espanoles, entre los quales estaban dos hermanos de mi padre, lIamados el vno Inguill y el otro Vaipai con mucha gente de su bando . . ."

Roca, cada uno de ellos con grandes cuadrillasde indios, los cuales dieron gran conseulo a los cristianos,

ANDEAN PAST 5 (1998) The Inca descendants of Spanish allies thus (logically) emphasized, in their recollections, the aid rendered the Spaniards by their ancestors. This version of the battle bolstered their positions in colonial society, for the Spaniards rewarded those who had befriended them. Their consistent use of the word "Christians" rather than "Spaniards" likewise highlights the new alliance between Incas and the Christian god. For these Incas, then, the ruins of Sacsahuaman were a monument to their own triumphant emergence from the confusions of the conquest to prominent positions in colonial Cusquefio society. Ethnic Incas of royal descent adopted the castle from Cusco's coat of arms as one of their heraldic emblems (Dean 1990:244-5). Along with Incaic symbols, a miniature castle was frequently worn in the headdr.essesofInca nobility in the middle of the colonial period (Figure 8). It also appears on the blazons belonging to postconquest native nobility, native textiles, and ceremonial wooden drinking vessels called keros (Rowe 1951:263). For public appearances on f~stive occasions, postconquest Incas would dress in the manner of their prehispanic royal ancestors and don these spectacular headdresses. Interestingly, the castle is the only major icon not of Incaic origin included in this postconquest headgear. It sits in the center of the headdress, encircled, perhaps even besieged, by emblems of Inca origin. In away, these postconquest Inca leaders have reclaimed Sacsahuaman. By adopting the stereotypic, anonymous European castle icon and planting it in a garden of Inca emblems, its Inca origins and identification are affirmed. And, by re-possessing the "fortress," these noble Incas assert their own interpretive control over the events of 1536. In their version it is the Inca alliance with Spanish Christians and Christian supernatural beings that emerges as the salient phenomenon.
que despues que se vieron conellos e vista la necesidad y hambre que pasaban, dieron orden de meter en la ciudad gran cantidad de comida para el socorro y mantenimiento de los cristianos e indios que estaban en ayuda e socorro dellos . . ," (Collapifia, Supno y otros Quipucamayos 1974 [1542-75]:65).

-172 For the numerous, different social elements sharing colonial Cusco, the monumental ruins of Sacsahuaman engendered specific histories. For each of the groups -- Spaniards,
non-Inca Andeans, and the Incas

-- the

ruins

did not hearken back to the period of use, however. Rather, the moment recalled is the one in which Sacsahuaman ceased to function, as intended -- that is, as a temple, depository, and stronghold. For each group examined, the critical moment was the battle of 1536, the moment when possession of Sacsahuaman and the futures of these groups were held in the balance. In 1536 Sacsahuaman had been a contested space; its ruins remained the space for contentious histories. In their ways of thinking about and representing Sacsahuaman, each of these groups asserted control over that moment in history. Clearly, the distinct ethnicities comprising colonial Cusquefio society had "aninvestment, each in a different way, in what remained of Sacsahuaman. As a postscript, Sacsahuaman, part of the Sacsayhuaman National Archaeological Park, is a major tourist attraction in Cusco today. Visible from its magnificent terraced walls, where visitors, dwarfed by the megaliths, congregate to take photographs, is a monumental white sculpture of Christ (Figure 9). This image now joins Sacsahuaman in watching over Cusco; its presence intensifies the dialogue which formerly occurred between the cathedral and Sacsahuaman, the past and the present, the Christian and the non-Christian. Yet, memories of 1536 no longer resonate among the ruins. Rather, the site summons memories of the Incas at the height of their power. The first modem excavations at Sacsahuaman, conducted by Luis Valcclrcel in 1932, commemorated the 400th anniversary of Pizarro's advent. Ironically, these investigations engendered renewed interest in the prehispanic history of Cusco and fed the rising tide of indigenismo. Thus, Sacsahuaman, perhaps more than any other Inca monument in and around Cusco, now serves as a sign of the preconquest past. While the single constant in colonial period understandings of Sacsahuaman had been its witness to the intervention of Christian su-

171Thus political control of the Andean area is characterized as the result of divine intervention. Furthermore, mundane power shifted from Inca emperor to Spanish ruler. No mention is made of the local Spanish authorities whose presence in the Andes Guaman Poma opposed. If Guaman Poma speaks for other non-Inca Andeans, we may speculate that the ruins of Sacsahuaman were both a symbol of the Inca's defeat in 1536 and a memorial to the power of the Christian god. If the non-Incas of colonial Cusco held Sacsahuaman as an emblem of the conquest of the Incas, how did the Incas themselves understand the ruins? Speaking for the Incas who descended from the rebels is Titu Cusi Yupanqui (1973 [1570]:82-87), the son of Manco Inca (the leader of the rebellious Incas who organized the siege).28 Titu Cusfwas the ruler

Dean: Sacsahuaman damage" ("ffue el primero que entro en la ffrotaleza al tiempo que se tomo les hacia mucho daiio . . ."). Thus the battle of Sacsahuaman is characterized as primarily Inca against Inca and as rebel troops against divine forces. In this light, the important outcome of 1536 was religious conversion. The dire po,;, litical consequences of the loss are minimized. The ethnic Inca who descended from the allies of the Spaniards also viewed Sacsahuaman as the locus of divine activity and as the place where the Incas and the Christian god formed an important and lasting alliance. The Quipucamayocs (Collapifia, Supno, y otros Quipucamayos 1974 [1542-75]:64), Inca "historians" who relied on mnemonic devices to keep oral histories, reported, within a decade of the event itself, the version of the siege that would be held by those Incas who did not join Manco's rebellion.30 The Quipucamayocs were partisan to the Inca allies of the Spaniards; in fact, the second part of their chronicle is an exultation of Paullu Inca, the Inca leader of royal blood, who remained loyal to the Spaniards during the rebellion. They emphasize the aid rendered to Spaniards by natives of Inca and other ethnicities. Deserving special mention in their account were four rebel Incas who changed sides at a crucial moment. The Quipucamayocs report
At this desperate time, God was served by the defection to the Christians of four Incas who were among the most important that Manco Inca had. They were Cayo Topa, don Felipe Cari Topa, Inca Paccac, and Uallpa Roca. With each one of them were great Indian squadrons. They gave great comfort to the Christians and later were seen with them and saw the need and hunger that existed. They ordered that a great quantity of food be smuggled into the city for the succor and sustenance of the Christians and Indians that were helping and aiding them.31 30 The Quipucamayocs were assembled at the orders of Governor Vaca de Castro and their reports were recorded with the aid of interpreters. 31 "A este tiempo tan trabajoso, fue Dios servido se pasaron a los cristianos cuatro ingas de los mas principales que tuvo Mango Inga, los cuales fueron Cayo Topa y don Felipe Cari Topa e Inga Paccac y Uallpa Roca, cada uno de ellos con grandes cuadrillas de indios, los cuales dieron gran conseulo a los cristianos,

of the exiled Incas(the followersof Manco)in


,

Vileabamba. In his account of the battle for Sacsahuaman, he underscores the valient defense of the "fortress" by rebel troops (a "fact confirmed by Spanish eyewitnesses including Pedro Pizarro) and'states,
This battle was bloody for both sides because of the many Indians that sided with the Spaniards.. Among them were two of my father's brothers, one called Inguill and the other Vaipai and their many followers . . .29

The battle, as he characterizes it, was lost due to the lack of Andean, and especially Inca, solidarity. More than any other factor, however, Titu Cusi Yupanqui (1973 [1570]:86) credits the successful storming of Sacsahuaman by'the Spaniards and their native allies to the appearance of Santiago who "was the first to enter the fortress and did much

28 When the siege failed, Manco retreated to Vileabamba where he and his sons ruled an independant Inca state until 1572 when the Spaniards captured and later executed his successor, Tupac Amaro. 29 Titu Cusi Yupanqui (1973 [1570]:84) writes, "Fue esta batalla de vna parte y de otra, ensangrentada, por la mucha gente de indios que ffavoreszfan a los espanoles, entre los quales estaban dos hermanos de mi padre, llamados el vno Inguill y el otro Vaipai con mucha gente de su bando . . ."

ANDEAN PAST 5 (1998) The Inca descendants of Spanish allies thus (logically) emphasized, in their recollections, the aid rendered the Spaniards by their ancestors. This version of the battle bolstered their positions in colonial society, for the Spaniards rewarded those who had befriended them. Their consistent use of the word "Christians" rather than "Spaniards" likewise highlights the new alliance between Incas and the Christian god. For these Incas, then, the ruins of Sacsahuaman were a monument to their own triumphant emergence from the confusions of the conquest to prominent positions in colonial Cusqueiio society. Ethnic Incas of royal descent adopted the castle from Cusco's coat of arms as one of their heraldic emblems (Dean 1990:244-5). Along with Incaic symbols, a miniature castle was frequently worn in the headdresses of Inca nobility in the middle of the colonial period (Figure 8). It also appears on the blazons belonging to postconquest native nobility, native textiles, and ceremonial wooden drinking vessels called keros (Rowe 1951:263). For public appearances on festive occasions, postconquest Incas would dress in the manner of their prehispanic royal ancestors and don these spectacular headdresses. Interestingly, the castle is the only major icon not of Incaic origin included in this postconquest headgear. It sits in the center of the headdress, encircled, perhaps even besieged, by emblems of Inca origin. In away, these postconquest Inca leaders have reclaimed Sacsahuaman. By adopting the stereotypic, anonymous European castle icon and planting it in a garden of Inca emblems, its Inca origins and identification are affirmed. And, by re-possessing the "fortress," these noble Incas assert their own interpretive control over the events of 1536. In their version it is the Inca alliance with Spanish Christians and Christian supernatural beings that emerges as the salient phenomenon.
que despues que se vieron conellos e vista la necesidad y hambre que pasaban, dieron orden de meter en la ciudad gran cantidad de comida para el socorro y mantenimiento de los cristianos e indios que estaban en ayuda e socorro dellos . . ." (Collapina, Supno y otros Quipucamayos 1974 [1542-75]:65).

-172 For the numerous, different social elements sharing colonial Cusco, the monumental ruins of Sacsahuaman engendered specific histories. For each of the groups -- Spaniards, non-Inca Andeans, and the Incas -- the ruins did not hearken back to the period of use, however. Rather, the moment recalled is the one in which Sacsahuaman ceased to function' as intended -- that is, as a temple, depository, and stronghold. For each group examined, the critical moment was the battle of 1536, the moment when possession of Sacsahuaman and the futures of these groups were held in the balance. In 1536 Sacsahuaman had been a contested space; its ruins remained the space for contentious histories. In their ways of thinking about and representing Sacsahuaman, each of these groups asserted control over that moment in history. Clearly, the distinct ethnicities comprising colonial Cusqueiio society had .an investment, each in a different way, in what remained of Sacsahuaman. As a postscript, Sacsahuaman, part of the Sacsayhuaman National Archaeological Park, is a major tourist attraction in Cusco today. Visible from its magnificent terraced walls, where visitors, dwarfed by the megaliths, congregate to take photographs, is a monumental white sculpture of Christ (Figure 9). This image now joins Sacsahuaman in watching over Cusco; its presence intensifies the dialogue which formerly occurred between the cathedral and Sacsahuaman, the past and the present, the Christian and the non-Christian. Yet, memories of 1536 no longer resonate among the ruins. Rather, the site summons memories of the Incas at the height of their power. The first modern excavations at Sacsahuaman, conducted by Luis Valcarcel in 1932, commemorated the 400th anniversary of Pizarro's advent. Ironically, these investigations engendered renewed interest in the prehispanic history of Cusco and fed the rising tide of indigenismo. Thus, Sacsahuaman, perhaps more than any other Inca monument in and around Cusco, now serves as a sign of the preconquest past. While the single constant in colonial period understandings of Sacsahuaman had been its witness to the intervention of Christian su-

173pernatural beings in the course of Andean history, today the ruin is primarily identified with its Incaic origins. Since mid-century Sacsahuaman has provided the setting for a revived and largely invented celebration of Inti Raymi, the Inca festival of the June solstice.32 The Inti Raymi performance is the culmination of Cusco's Jubilee week (Semana Jubilar del Cusco) in the month of June. As was intended, the festival has become a profitable event in terms of tourism, as well as an occasion for celebrating the city's and the nation's indigenous ancestry. Thus the ruin has become a stage for an annual commemoration of the city's prehispanic heritage, and residents of Cusco continue to rework, at least figuratively, what remains of Sacsahuaman. Acknowledgments This article is an expanded version of a paper presented at The Culture of Ruins: An Interdisciplinary Conference sponsored by The Center for Cultural Studies in collaboration with the Board of Studies in Anthropology, University of California, Santa Cruz (January 24-26, 1992). I am grateful to those attending the conference for their insights. Further research was aided by grants from the faculty senate committee on research and the Division of the Arts, UCSC. Many thanks also to Brian Bauer, Tom Zuidema, and the anonymous reviewer for the close attention they gave to this paper and their very helpful suggestions. References Cited
Acosta, Joseph de 1962 [1590] Historia natural y moral de las Indias, 2nd edition, edited by Edmundo O'Gonnan. Mexico: Fondo de Cultura Economica. Angles Vargas, Victor 1978-1983 Historia del Cusco. 2 volumes. Lima: Industrialgrafica S.A. 1990 Sacsayhuamim: portento arquitectonico. Lima: Industrialgrafica. 32 See Fiedler (1985:338-354) for an excellent discussion of the new Inti Raymi (a theatrical perfonnance loosely based on Garcilaso's descriptions). This modern Inti Raymi was the brainchild of Cuzquefio journalist Humberto Vidal Unda who organized the first festivities in 1944 as part of the period's new indigenismo.

Dean: Sacsahuaman
Castro, Ignacio de 1978 [1788] RelaciOndel Cuzco. Lima: Universidad Nacional Mayor de San Marcos. Cieza de Leon, Pedro de 1959 [1553] The Incas, edited by Victor Wolfgang Von Hagen, translated by Harriet de Onis. Norman: University of Oklahoma Press. 1985 [1553] Cronica del Peru, segunda parte. Lima: Pontificia Universidad Catolica del Peru, Academia Nacional de la Historia. Cobo, Bernabe 1990 [1653] Inca Religion and Customs, translated and edited by Roland Hamilton. Austin: University of Texas Press. Collapifia, Supno y otros Quipucamayos 1974 [1542-75] Relacion de la descendencia, gobierno, y conquista de los Incas (al Gobernador Vaca de Castro). Lima: Editorial Jurfdica, Ediciones de la Biblioteca Universitaria. Contreras y Valverde, Vasco de 1982 [1649] Relacion de la ciudad del Cusco, 1649, edited by Maria del Carmen Martin Rubio. Cusco: Imprenta Amauta. Cordova y Salinas, Diego de 1957 [1651] Cronica de la religiosfsima provincia de los doce apostoles. In Cronica franciscana de las provincias del Peru, edited by Lino G. Canedo. Washington, D.C.: Academy of American Franciscan History. . Cornejo Bouroncle, Jorge 1960 Derroteros de arte cuzqueno: datos para una historia del arte en el Peru. Cusco: Ediciones Inca. Dean, Carolyn S. 1990 Painted Images ofCuzco's Corpus Christi: Social Conflict and Cultural Strategy in Viceregal Peru. Ph.D. dissertation, University of California, Los Angeles. ' Duviols, Pierre 1962 Les traditions miraculeuses de siege du Cuzco (1536) et leur fortune litteraire. Bulletin de la Faculre des Lettres de Strasbourg 41:393-400. Esquivel y Navia, Diego de 1980 [ca. 1749] Noticias cronologicas de la gran ciudad del Cuzco, edited by Felix Denegri Luna, Horacio Villanueva Urteaga, and Cesar Gutierrez Mufioz. 2 volumes. Lima: Fundacion Augusto N. Wiese, Banco Wiese Ltdo. Fawcett, Percy Harrison 1953 Lost Trails, Lost Cities, edited by Brian Fawcett. New York: Funk and WagnaJlsCo. Fiedler, Carol Ann 1985 Corpus Christi in Cuzco: Festival and Ethnic Identity in the Peruvian Andes. Ph.D. dissertation, Tulane University. Garcilaso de la Vega, el Inca 1966 [1609] Royal Commentaries of the Incas and General History of Peru, Part One, translated by Harold V. Livennore. Austin: University of Texas Press. Gasparini, Graziano and Luise Margolies
1980

Inca Architecture, translated by Patricia J.

Lyon. Bloomington: Indiana University Press.

ANDEAN PAST 5 (1998)


Gonzalez Holguin, Diego 1901 [1608] Arte y diccionario Qquechua-Espanol correjido y aumentado por los RR. PP. Redentoristas al que en 1608 publico el rvdo. p. Diego Gonzalez de Holguin, S. J., en esta ciudad de los Reyes. Lima: Imprenta del Estado. Guaman Poma de Ayala, Felipe 1988 [1615] El primer nueva coronica y buen gobierno, 2nd edition, 3 volumes, edited by John V. Murra and Rolena Adorno, translation and analysis of the Quechua by Jorge L. Urioste. Mexico: Siglo Veintiuno. Herrera y Tordesillas, Antonio de 1945 [1615] Historia general de los hechos de los castellanos en las islas y tierra jirme de el mar oceano. Asunci6n del Paraguay: Editorial Guarania. Lee, Vincent R. 24:49-60. Mogrovejo de la Cerda, Juan 1983 [1690{1649}] Memorias de la Gran Ciudad del Cusco, 1690, edited by Maria Del Carmen Martin Rubio. Cusco: Imprenta Amauta. Montoto de Sedas, Santiago 1928 Nobiliario de reinos, ciudades, y villas de la America espanola. Colecci6n de documentos ineditos para la historia de Hispano-America, V01ume 3. Madr~d: Compafiia Ibero-Americana de Publicaciones, S.A.. . MUIita,Martin de 1986 [1600-11] Historia general del Peru, edited by Manuel Ballestero~. Madrid: Historia 16. Pizarro, Pedro 1986 [1571] Relacion del descubrimiento y conquista de los reinos del Peru, 2nd edition, edited by Guillermo Lohmann Villena. Lima: Pontificia Universidad Cat6lica del Peru. Polo de Ondegardo, Juan 1916 [1571] Informaciones acerca de la religion y gobierno de /os Incas. In Colecci6n de libros y documentos referentes a la historia del Peru, V01ume 3, edited by Horacio H. Urteaga. Lima: Imprenta y Libreria Sanmartf. Protzen, Jean-Pierre 1985 Inca Quarrying and Stonecutting. Nawpa Pacha 21.:183-214. 1986 Inca Stonemasonry. Scientific American 254(2):94-105. Rowe, John Howland 1951 Colonial Portraits of Inca Nobles. In The Civilizations of Ancient America. Selected Papers of the XXIX International Congress of Americanists, edited by Sol Tax, pp. 258-268. Chicago: University of Chicago Press. 1990 Los monumentos perdidos en la Plaza Mayor del Cuzco Incaico. Saqsaywaman 3:81-109. Cusco: Direcci6n de Actividades Culturales, Instituto Departamental de Cultura. Sancho de la Hoz, Pedro 1968 [1534] Relaci6n para su Majestad. Biblioteca Peruana, Series 1, Volume 1, pp. 275-343. Lima: Editores Tecnicos Asociados.

-174
Sarmiento de Gamboa, Pedro 1943 [1572] Historia de los Incas. Buenos Aires: Emece Edit. Titu Cusi Yupanqui, Inga Diego de Castro 1973 [1570] Relacion de la conquista del Peru. Lima: Ediciones de la Biblioteca Universitaria. Toledo, Francisco de 1904 [1571] Carta al Rey. In La Imprenta en Lima (1584-1824), by Jose Toribio Medina, p. 174. 4 volumes. Santiago de Chile: Impreso y grabado en casa del autor. Urbano, Henrique 1992 Sincretismo y sentimiento religioso en los Andes. Apuntes sobre los orfgenes y desarrollo. In Los conquistados: 1492 y la poblaci6n indlgena de las Americas, edited by Heraclio Bonilla, pp. 223261. Bogota: Tercer Mundo, Libri Mundi, and FLACSO, sede Ecuador. Valcarcel, Luis E. 1934-1935. Sajsawaman redescubierto. Revista del Museo Nacional 3(1-2):3-36, 3(3):211-233, 4(2): 161-203. Van de Guchte, Maarten 1984 El cicio mftico de la piedra cansada. Revista Andina 4(2):539-556.

1986. The Buildingof Sacsahuaman.NawpaPacha

VonDliniken,Erich

1970 Chariots of the Gods? Unsolved Mysteries of the Past, translated by Michael Heron. New York: G.P. Putnam's Sons. Zuidema, R. Tom 1983 The Lion in the City: Royal Symbols of Transition in Cuzco. Journal of Latin American Lore 9(1):39-100.

175-

Dean: Sacsahuam(m

Figure 1. Aerial view of Cusco. Sacsahuaman is in the upper left. Photo courtesy of Servicio AerofotogrMico Nacional (SAN), Peru.

ANDEAN PAST 5 (1998)

-176

SACSAHUAHAN <XJMPLEX I Esplanade 2 Terraces


3 Huyuc 4 Harca

Storehouses

Figure 2: Plan of Sacsahuaman (after Gasparini and Margolies 1980).

177-

Dean: Sacsahuaman

Figure 3. Sacsahuaman, detail of megalithic terraces.

ANDEAN PAST 5 (1998)

-178

Figure 4. Felipe Guaman Poma de Ayala, 1988 [1615]:159 [161]. The weary stone.

179-

Dean: Sacsahuaman

Figure 5. One version of the coat of arms of Cusco.

ANDEAN PAST 5 (1998)

-180

Figure 6. Felipe Guaman Poma de Ayala, 1988 [1615]:402 [404]. The miracle 0[81. Mary.

181-

Dean: Sacsahuaman

COll1aViSTA~

~~.UyR<ID~L5 I

Figure 7. Felipe Guaman Poma de Ayala, 1988 [1615]:404 [406]. The miracle of Santiago.

ANDEAN PAST 5 (1998)

-182

Figure 8. Drawing of the colonial period Inca headdress worn by Don Carlos Huayna Capac Inca in an anonymous painting of Cusco's Corpus Christi procession ("The Parish of San Cristobal,"
ca. 1674-1680).

183-

Dean: Sacsahuaman

Figure 9. The statue of Christ seen from the ruins of Sacsahuaman.

THE ALCA OBSIDIAN SOURCE: THE ORIGIN OF RAW MATERIAL FOR CUZCO TYPE OBSIDIAN ARTIFACTS

Richard L. Burger
Yale University

Frank Asaro
Lawrence Berkeley National Laboratory

Paul Trawick
University of Kentucky Fred Stross Lawrence Berkeley National Laboratory

Background A collaborative study of the provenience Central Andes was initiated at the Ernest Orlando Lawrence Berkeley National Laboratory (LBNL) in 1974 by archaeologist Richard L. Burger and nuclear chemists frank Asaro and Helen V. Michel. During the first stage of this joint research (19:74-1975), obsidian samples from 141 archaeological sites in Peru and Bolivia were studied. Thanks to the support of the US Energy Research and Development Administration, it was possible to analyze 812 artifacts by X-ray fluorescence (XRF) and 141 by instrumental neutron activation (INAA); these results constituted the first published consideration of obsidian source provenience and exchange in the Central Andes (Burger and Asaro 1977, 1978, 1979). INAA was utilized because of its high precision and accuracy for a large number of elements. The XRF technique was of lesser accuracy but acted to minimize the expense of running large numbers of samples from sites spanning the gamut of space and time. The source assignments by XRF were considered provisional until assignments of each representative source were confirmed by INAA; where ambiguity in XRF results existed, samples were rerun for definitive assignment using INAA. In this initial study, eight major chemical types of Peruvian obsidian exploited in antiquity were identified on the basis of these analyses. Ninety-eight percent of the artifacts analyzed belonged to eight chemical groups, ANDEAN PAST 5 (1998):185-202.

of . prehispanic obsidian artifacts from the

but at that time, only one group could be linked with samples said to come from a geological deposit of obsidian: the Quispisisa Source. These specimens were provided by Rogger Ravines from a quarry said to be near San Genaro in the south-central highlands of Peru. Five of the other types of obsidian, although not linked to' specific deposits, had distributions that were limited to particular regions in our sample; this patterning suggested possible areas in which the sources might be located and provided the basis for the provisional names of the major chemical types (i.e., Acari Type, Andahuaylas A Type, Andahuaylas B Type, Ayacucho Type, Pampas Type). Two types of obsidian with unknown sources were present at archaeological sites in

more than one region, and the provisional


.

names of Cuzco Type and Titicaca Basin Type were assigned to them based on the areas where these chemical types most commonly occurred judging from our preliminary results. Of the obsidian artifacts analyzed in the initial study, the vast majority (81%) of the artifacts belonged to one of three groups: the Quispisisa Source, the Cuzco Type, and the Titicaca Basin Type. Artifacts of each of these three kinds of obsidian were distributed over a relatively broad geographical area, suggesting that obsidian from these sources was widely exchanged in pre-Hispanic times. Despite explorations by Burger in the summer of 1974, the collaboration of Peruvian and U.S. archaeologists, and the dissemination of the findings in Spanish and English, little progress was made in identifying the unknown

ANDEAN PAST 5 (1998) source areas. Considering the state at that time of non-economic geology in highland Peru and the limited vehicular access to the high altitude zones where obsidian deposits occur, this was not surprising. Obsidian de'posits were discovered at Sachaca and Yuri near the city of Arequipa by Jose Antonio Chavez Chavez and near Tukumachay, Ayacucho by the Ayacucho Basin Project directed by Richard S. MacNeish, but analyses of samples from the three deposits by INAA at LBNL showed that they were chemically dissimilar from the types of obsidian utilized for pre-Hispanic artifacts in these and other regions (Burger and Asaro 1977:65, 1979:315). Subsequently, a fourth obsidian deposit on Mt. Chachani in Arequipa was sampled by Chavez but it likewise did not match any of the chemical compositions of artifacts. A second set of Peruvian obsidian was studied at LBNL in 1977/1978 with the support of the University of California Archaeological Research Facility, US Department of Energy, and NSF. Approximately 500 additional artifacts were analyzed using XRF and INAA, including samples from northern and

.-186 nally, it appeared after the second stage of research that most of the major chemical types exploited by prehistoric populations in Peru probably have now been delineated in the LBNL studies. Locating the Geological Deposits of Cuzco Type Obsidian The 1974/1975 and 1977/1978 studies at LBNL described above found that artifacts of the Cuzco Type constituted the overwhelming majority of the obsidian utilized in the Cuzco Valley (over 90% in most periods); obsidian of this chemical type also appears in low quantities in the Lake Titicaca drainage. Cuzco Type obsidian artifacts were rarely found in areas outside of the southern highlands of Peru. These findings suggested that the geological source for Cuzco Type obsidian should be sought in the Cuzco basin or in adjacent regions from which samples of archaeological artifacts of obsidian were not available. In 1981, the mother of Peruvian archaeologist Isaac Perez Angulo visited a possible primary source of geological obsidian located near the village of Yumasca on the northern slopes of Cerro Santa Rosa above the Cotahuasi River (also known as the Huarcaya River) in the northern portion of the Department of Arequipa (Figures 1 and 2). Sra. Angulo de Perez collected a sample from an unmodified "obsidian boulder" for her son who then provided it to one of us (Richard L. Burger) for analysis. Preliminary studies at LBNL of two flakes from the boulder using XRF in 1984 suggested a chemical match with the Cuzco Type. These results pointed to the need for a more reliable sample at the putative obsidian deposit. In July of 1984, Paul Trawick, at R.L. Burger's suggestion, was able to visit the obsidian source area during his long-term anthropological research in the Cotahuasi region. Trawick's visit confirmed the existence of a geological deposit of obsidian in the area near the town of Alca and he provided samples to Burger for analysis. The source samples were analyzed by INAA at LBNL in 1994. As re-

southern Peru which- had not been sampled


previously. Thus far, only a portion of these results have been published (cf Burger 1980, 1981, 1984; Burger and Asaro 1993; Burger et al. 1984). In this second stage of research, 99% of the obsidian artifacts sampled came from the chemical types defined in the earlier study. The Rare 1 chemical type defined in the original study proved to be common at sites near the Peruvian-Bolivian border and it is now referred to as the Tumuku Type. One additional obsidian chemical group, provisionally dubbed the Chumbivilcas Type, was identified for the first time in the 1977/1978 $tudy. The Tumuku Type and the Chumbivilcas Type are now considered as the ninth and
.

the tenth chemical types of obsidian intensively utilized in pre-Hispanic Peru. This second round of research confirmed our earlier impression that high quality obsidian sources are uncommon in Peru. Nevertheless, the number of obsidian sources utilized in preHispanic Peru far exceeds those exploited in neighboring Andean nations such as Ecuador (Burger et al. 1994; Asaro et al. 1994). Fi-

187ported here, the results of these INAA analyses confinned the match with the Cuzco Type artifacts. Although the discovery of the deposit in the Yumasca area has been mentioned previously (Burger and Asaro 1993:219-220), this article provides the first extended treatment of the geological source and the analytical results that pennit a definitive identification of the Alca deposit as the source for the Cuzco Type obsidian. The Alca Source: Location And Geology The Alca source (2850 meters above sea level, masl) is located about 1 Ian west of the village of Alca, Cotahuasi Province, Department of Arequipa (Figure 2). The obsidian deposit at Alca has a general location, of 158'31" S latitude, 7245'30" W longitude. It is found 16 Ian to the northeast of the provincial capital at Cotahuasi and about 190 Ian northwest of the departmental capital, Arequipa. The city of Cuzco is located about. 195 Ian to the northeast of Alca (Figure I). The headwaters of the Cotahuasi River lie on the upper slopes of the western face of the continental divide and its waters ultimately join with those. of the Chicha River and the Moran River to fonn the Ocoiia River before draining into the Pacific. Midway along its course, the Cotahuasi enters a chasm believed to be the world's deepest canyon (approximately 3,000 meters from rim to bottom, excluding the adjacent mountains, about twice the depth of the Grand Canyon). The ensuing arid gorge cuts off the densely settled highland drainage in which Cotahuasi and Alca are located from the distant lower valley of the Ocoiia (Trawick 1994:24-27). The Cotahuasi area is located in a region of ancient volcanic eruption and tectonic uplift which extends far to the south, through the Departments of Arequipa and Tacna. In the Cotahuasi area, almost the entire Tertiary was dominated by volcanic processes which continued into the Quaternary. These processes were responsible for most of the rocks and topography of the area, although the landscape was modified by subsequent glacial activity (Olchauski and Davila 1994). The modern village of Alca (2750 masl), like that of Cotahuasi (2683

Burger et al.: Alca Obsidian Source masl), sits atop post-Pleistocene alluvial deposits which fill the narrow valley that had been eroded into earlier volcanic deposits by the Cotahuasi River. Located on the southern bank of the Cotahuasi River, Alca is overshadowed by cliffs and steep slopes of the older volcanic deposits which culminate 1~ Ian to the east in the glaciated volcano Nevado Firuro (5498 masl). Large chunks of obsidian are found near Alca among the terraces along the valley bottom; they are the result of ero.sion from the source area above. Trawick recovered the in situ obsidian source samples to the west of Alca, immediately below the annex of Cahuana. In this area, erosion had exposed Tertiary deposits, including a thick band full of rather densely concentrated obsidian nodules. This layer was exposed in a cliff face of a plateau that rises 200 m above the valley bottom. Immediately above the plateau and above the village of Cahuana is a pre-Hispariic ruin believed to be the Inca administrative capital of the area. The cliffs are composed of poorly-consolidated volcanic tuff, interspersed with,nodules of obsidian; the largest observed by Trawick measured about 60 em on a side, but many measured only 20 em. The obsidian becomes evident as soon as one approaches the cliffs from the agricultural terraces on the left-hand side of the road. Many nodules lie among the eroded rocks and debris at the base of the escarpment. Looking up the cliff face, one can see that obsidian nodules are distributed rather evenly up to a point about 100 meters above the valley bottom, where a darker layer of rock and sediment is found. The obsidian samples tested at LBNL were taken from the layer of obsidian nodules itself (rather than the talus), 40 m to the left of a cave or a large hole where the material may once have been mined. This area contained a dense concentration of larger obsidian nodules, some black, others brown or red, and some multicolored. It appears that the obsidian deposit extends all around the pueblo of Alca and up the' Chococco River, a tributary of the Cotahuasi River. Trawick estimates that the deposits probably run for at least 1-2 Ian. Future investigation will be necessary to define more

ANDEAN PAST 5 (1998) precisely the extent of the deposit and to locate areas of pre-Hispanic quarrying and workshop activity. The obsidian deposits occur within a Tertiary formation known as the Tacaza Group, dating to the Miocene. The Tacaza Group generally has a varied lithology, but it mainly corresponds to volcanic rocks, particularly pyroclastics, mixed with sediments; these deposits can reach 1500 meters in thickness. In the area around Alca, strata of the Tacaza Group also include ignimbrites and dacites. Studies of the Tacaza Group in the Orcopampa area to the east of the Cotahuasi quadrangle suggest a date of approximately 18.9-19.1 mya for this formation (Olchauski and Davila 1994:24-26). It is worth noting that although the geology of this area is the subject of a recent monograph by Peru's Instituto Geologico Minero y Metalfugico, this volume makes no mention of the obsidian deposits. This omission illustrates the ongoing need for problem-oriented fieldwork to locate the obsidian sources utilized in pre-Hispanic times. Neutron Activation Analyses At LBNL Two of the samples from the Alca Source collected by Trawick were analyzed at LBNL using INAA. These were obtained in situ from an exposed deposit of obsidian nodules referred to previously. The two samples, given the code names Rosa-l and Rosa-2, were collected from the same band about 15 meters apart. The samples were measured utilizing the Luis W. Alvarez Iridium Coincidence Spectrometer (LWAlCS) which produces more precise results for a number of elements than could be obtained by the INAA methodologies used at LBNL in the 1970s and early 1980s. Because of unusual laboratory problems, including unavoidable changes in the preparation room and ventilation system previous to sample preparation, it was necessary to go to special lengths to check levels of contamination that might arise in our samples, as well as to determine what precision was attainable. Sample Prepqration

-188

Chips of obsidian were struck off each obsidian chunk with a plastic hammer with the sample completely shielded with plastic. The obsidian chips (from the interior of the chunks when possible) were ground to a rough powder with an agate mortar and pestle. The rough grinding was probably not sufficient to' homogenize the samples. Samples of 50 or 100 mg weight were encapsulated in 0.020 cm thick Al foil, which was rated by its manufacturer as 99.9999% pure. The Al capsules were swabbed clean twice (three times if the samples were expected to contain Ir) with a cotton Q-tip wet with 200 proof ethyl alcohol. The purpose was to remove loose extraneous dust from the capsule surfaces which could become radioactive when irradiated and contaminate the LWAlCS. The cleaned capsules were stacked in a quartz tube along with standards of Standard Pottery (Perlman and Asaro 1969), used for most elements, and DINO-l (Alvarez et al. 1980) U1ied for Ir, Se, and Zn. There were 9 quartz tubes in the irradiation, each containing 42 samples, and there were standards of about 24 and 50 mg weight at the bottom, center, and top of the tube containing the Alca source obsidian. The tubes were heated to about 190C for several hours to drive off excess water from the samples. Later the tubes were heat-sealed under vacuum to reduce pressure build-up during irradiation in case residual water vaporized or C02 was emitted from carbonates. Each sealed tube was tightly wrapped with Al kitchen foil to facilitate heat transfer away from the tubes and the entire cluster of 9 tubes was wrapped with Al kitchen foil for the same reason. The samples were irradiated for 48 hours in the 10m W Triga reactor of the University of Missouri at Research Park in Columbia, Missouri at a flux of 2.5 x 1013neutrons per second per cm2. Samples cooled for about one month, in which radioactive isotopes with half-lives much shorter than one month decayed away, and then the samples were returned to LBNL. One month was spent measuring standards, calibrating, checking interferences, and evaluating precisions, and then measurements on the samples were begun. Before measurement, each Al capsule was swabbed clean again

189twice (three times if they were expected to contain significant Ir). The purpose of the cleaning was to remove any contamination which might have occurred if one of the sample capsules leaked, as well as any residual dust not removed in the cleaning before irradiation. Methodology The LWAICS is a gamma ray analyzer designed especially to measure abundances of the element iridium with high sensitivity (Alvarez et al. 1988; Michel et al. 1990, 1991). It also has been adapted to the measurement of other elements simultaneously with the Ir determination. It consists of two intrinsic 5 cm long by 5 cm diameter germanium detectors whose faces are 20 mm apart, and in that space are two high voltage shields, two vacuum seals, two temperature shields, a sample track, and the sample capsule which is being counted. Measurements generally can be made by gamma ray singles determinations or by coincidence measurements with .two gamma rays with or without an anti-Compton shield (which can' reduce the background of scattered radiation wider the peaks). We used one detector, arbitrarily designated the B detector, for all of the singles measurements because its resolution was significantly better than that of the A detector. In the present experiments the anti-Compton shield was not used. The position of the capsule between detectors A and B can vary sufficiently to change the counting geometry in detector B for single measurements by about 2% and create that much uncertainty in the singles measurements. To remove this uncertainty, the intensity of the 1121 keV gamma ray of 46SC was measured with both detectors A and B for each sample, and from geometric considerations, the ratio of the intensities indicated the exact relative position of the capsule between the two detectors. For coincidence measurements with both detectors A and B, the geometry effects compensated and no corrections were necessary. Also small variations in position in the radial direction do not necessitate any corrections.

Burger et al.:Alca Obsidian Source Pulsers feed into each detector and the ratio of the B detector output to the pulser input rate is roughly the singles detector efficiency. This ratio has an exponent parameter (0.929) that was evaluated empirically by comparing results with samples of known abundances but half or less of the counting rate of the standards. The coincidence efficiency is roughly the product of the A and B detector ratios'. The B detector ratio for coincidence measurements also has an exponent parameter which was adjusted empirically to give the same answers for coincidence measurements as single measurements on 46SCradiations. This parameter, however, was dependent on the sample counting rate and in this work var- . ied between 0.426 for samples counting less than 25,000 counts per second (cps) to 0.445 for samples counting 36,000 cps. Results Table 1 shows the results of our measurements for the 10 diagno.sticelements measured in the two samples of Alca Source obsidian. The precisions of measurement, accuracies, and reasons for deciding that these particular elements were diagnostic are given later in this paper. The average difference in abundance between the two source samples was 1.7% for the 9 elements with counting errors of about 1% or less and 1.6% for the 7 of those elements that were previously (Burger and Asaro 1977) well-measured in Peruvian artifacts belonging to the Cuzco Type chemical group. The average differences in abundance for the Rosa-1 and Rosa-2 source samples from Alca with the Cuzco Type artifact group are 1.7 and 2.1%, respectively. These are excellent agreements and demonstrate that the obsidian deposit in the AIca area is the source of the Cuzco Type chemical group. The extent to which we can distinguish between the samples of obsidian collected fifteen meters apart is discussed in the following sections. Backgrounds And Contamination We measured the background radiation in the LWAlCS and calculated the background

ANDEAN PAST 5 (1998) for each sample from the sample weight. Table 2 Sh9WS the background levels for a nominal sample weight of 55.4 mg, the average of our two obsidian samples from Alca. In order to determine the impurities in the 99.9999% pure Al foils and any we might have introduced in the laboratory, we ran 10 blank Al capsules with no samples inside them. In addition we unwrapped the two measured irradiated samples of Alca Source obsidian, split each one, repackaged them in non-irradiated AI, and remeasured each split. Table 2 also shows the mean and root-meansquare deviation for 14 elements in the 10 Al foils and the values for the Al capsules which contained samples Rosa-1 and Rosa-2 during the irradiation and initial measurements. The effect of residual obsidian powder in the Al capsules was determined from the abundances of those elements not expected to be impurities in AI, e.g., Cs, Th, and Hf. The weight of Al was assumed to be constant for each sample, and the constancy of the measured weights of tbe blank Al foils suggest that this assumption is accurate to 1%. Likewise the Sc abundance (which has a very small cQunting error) in the Al foils was con~tant to 1%, which suggests the Sc is indeed inherent within the Al foil from which the capsules are made. The data suggest that the Al foils also contain impurities of Zn, Co, Fe, and Sb with possible laboratory contaminations of some Sb and some Zn. The Se, however, appears in large part to be due to laboratory contamination. Precision of Measurement The gamma rays used are shown in Table 2. Tests on other obsidian have shown our overall results to have a precision for the best measured elements of the larger of 1% or the counting error. The present work discussed below is consistent with that determination. Precisions can be an indicator of how closely artifacts should agree with very homogeneous sources if: 1) the artifacts originated from those sources; 2) both sets of measurements were calibrated with the same stan-

-190 dards; 3) the same abundances were used for the standards; and 4) both sets of measurements were made in the same laboratory by the same techniques and methods. The highest quality comparisons can be made when artifacts and source samples are run in the same irradiation against exactly the sam~ standard measurements. If different values for standard abundances are used, but the standards were the same, it is a simple matter to recalibrate the results so that the same abundances were used for all standard measurements. If the same standards are used by two laboratories and the same abundances in the standards are used, the poorer of the precisions of measurements of the two laboratories for any element may be used as rough indices of what type of agreement one should expect. But measurement methodologies are usually not exactly the same, and differences can introduce discrepancies. Although we speak of average precision in this work, in reality each element has independent problems and should be considered independently. In comparing the present work with the LWAICS to previous INAA studies made at LBNL with conventional equipment, the methodology is different and the precisions are different. Generally, with the exception of those elements measured in the previous LBNL work with a very small high-resolution detector, e.g., Eu, Hf, Yb, and Ta, the precisions of the LWAlCS measurements should be better than those of the earlier LBNL work. In the past we have used Ce abundances for determinations of provenience with the LWAlCS. These abundances are measured with radiations from l41Ce,which contain a component from the fission of 235U.As we do not currently measure U abundances with the LWAICS, we cannot evaluate and remove the fission correction, which usually is about 0-3 parts-per-million (ppm). We can compare results quantitatively without correcting for fission if they are measured with the same reactor. Otherwise, as different reactors can produce slightly different proportions of 141Ce due to 235Ufission, we would not expect exact agreement between uncorrected Ce abun-

191dances of artifacts and the sources (irradiated in different reactors) from which they originated. In the present work, we are trying to measure with higher precision than previously and do not use uncorrected Ce abundances in the provenience determination. Accuracy When comparing results between two laboratories using different standards, the uncertainties in the abundances in the standards must be considered as well as the precisions of measurement. The last column of Table 1 shows the abundances of the elements in the standards used in the present work and also the one sigma estimates of the uncertainties in those abundances. In order to determine the percentage accuracy of our measurements, the square root should be taken of the sum of the squares of the precision (in percent) and the uncertainty in the standard (also in percent). Intercalibration between laboratories, which can greatly reduce these differences, becomes a necessity for precise work, but the intercalibration error must be carefully evalUated (Stross et al. 1983). The samples Rosa-1 and Rosa-2 were each removed from their Al encapsulating foils, split into two portions and each portion reencapsulated in non-irradiated AI and remeasured. (As indicated earlier, the irradiated Al foils were also measured.) The data are shown in Table 3. Selenium and Sb, which had previously been shown to be inordinately high in the AI foils from Rosa-1 and Rosa-2, are also variable in abundance in the splits of Rosa-1 and probably Rosa-2. We therefore consider that Se and Sb were probably contaminants in the samples introduced in the sample preparation procedure. Hafnium, which did not appear as a large contamination on the Al foils, definitely has different abundances in the splits of Rosa-1 and of Rosa-2. To demonstrate that this deduction is correct, we included in Tables 2 and 3 the independent determination of Hf abundances by both singles measurements with the 482 keV gamma ray and coincidence measurements with the 482 and 133 keY gamma rays. We concluded that part of the Hf was probably also a laboratory

Burger et al.: Alca Obsidian Source

contaminant in these experiments and did not use Hf for the provenience determination. The abundances measured in split irradiated samples can be higher than the abundance before splitting, if volatile material, such as water, escaped from the samples during irradiation. On the other hand abundances can be lower if any extraneous bits and pieces of Al foil became mixed with the sample powder during splitting. However, all elements will be affected the same way by either of these two processes. By taking ratios of abundances .both problems can be removed, and any error associated with weighing samples is removed. Table 4 shows the abundances of the parent and the split samples, all normalized to the abundance ofSc, 1.95 ppm, in the Cuzco Type chemical group of artifacts. For each element in Rosa-1 and Rosa-2 the mean value from the three measurements is calculated and also the root-mean-square deviation (RMSD). The square of the average counting error from the three measurements is subtracted from the square of the RMSD and the square root of this value divided by the mean abundance is an index of some of the residual errors of measurement. The residual error could reflect errors in sample positioning, improper subtraction of interfering elements which decay at different rates than the element of interest during the course of the measurement regime, heterogeneity in a 50 mg sample due to contamination or (infrequently) natural geochemical variations in obsidian, or measurement problems not yet understood at the level of precisions of a few tenths of a percent. Because the residual errors are so small, <0.7% for the best measured selected elements, we deduce that we probably do not have contamination problems for these elements. As Rb and Co, with larger counting errors than 1%, have root-mean-square deviations comparable (actually fortuitously smaller) than the counting errors, they also are unlikely to be contaminants at th~ levels at which they are measured. There are other errors in addition to the residual error that affect the precision, such as errors in weights, in evaluations of gamma-ray efficiency as a function of radiation gross

ANDEAN PAST 5 (1998) count rate, and in evaluations of gamma-ray interferences. We estimate our precision to be about 1% on the average for the best measured elements. For nearly ideal comparisons, such as those made in Table 4, however, the rootmean-square deviations divided by the means average 0.50%, and this value can be used instead of 1%. The differences in element abundances are not significant at the 0.5% level [(Rosa-2 - Rosa-l)/Rosa-l] except for those of Fe and possibly Zn which differ by 2.4 and +1.6%, respectively. Cobalt and Rb (with counting errors greater than 1%) also appear to have larger deviations, 8.6 and 3.2% respectively, than expected from the counting errors. These deviations suggest a difference in element abundance pattern between the two samples collected 15 meters apart. Three more samples of Yumasca source obsidian, Rosa-3, Rosa-4, and Rosa-5, were measured following an irradiation made about 11 months after Rosa-l and Rosa-2 were irradiated. For the ,8 elements.with counting errors of less than 1% shown in Table 1, the' average coefficient of variation for Rosa-3, Rosa-4, and Rosa':5~as only 0.6%, and the average difference from the Rosa-2 abundances was only 1.2%. Archaeological Considerations From the outset of our research over two decades ago, our approach to obsidian exchange has employed a pan-regional framework. This has generated a broad data base that permits the inference of shifting patterns of interaction which crosscut traditional cultural boundaries and ecological zones. This approach has also facilitated the location of the geological sources of raw material for obsidian artifacts. The determination that the Alca Source provided the obsidian for those communities using Cuzco Type obsidian artifacts has many ramifications for our understanding of cultural contact and exchange in pre-Hispanic Peru. A full diachronic treatment of these ramifications within a broad pan-regional framework is beyond the objectives of this article; however, this has been attempted elsewhere, in collaboration with Karen Mohr-Chavez and Sergio Chavez (Bur-

-192 ger et al. n.d.). In this paper, we will focus more narrowly on the significance of this discovery from a more local perspective--that of the Cotahuasi area. Despite its importance in Colonial, Republican, and contemporary Peru, the Department of Arequipa has received surprisingly little attention from most Peruvian and foreign archaeologists, and syntheses of Central Andean prehistory give scant attention to preHispanic developments in Arequipa. Fortunately, a small number of regional archaeologists, most notably Mciximo Neira and Eloy Linares Malaga, have spent decades laying the foundations of the archaeology of Arequipa, and a younger generation of local archaeologists, such as Jose Chavez Chavez, Augusto Belan Franco, and Pablo de la Vera Cruz, also have done much to advance research there. The publication of syntheses by Linares MaIaga (1990) and Neira (1994) should make the prehistory of Arequipa more accessible, but knowledge of the region's archaeology still remains highly tentative (MciximoNeira, personal communication 1995). The obsidian results reported in this article serve to focus our attention on the oftenoverlooked region of Cotahuasi, because one of the three most important obsidian sources exploited in pre-Hispanic Peru is located there. A pioneering archaeological reconnaissance of the Cotahuasi watershed by Chavez Chavez (1982) documented twenty-seven preHispanic sites. Nine additional sites were subsequently located by Trawick (1994). Little evidence has been found thus far of an early occupation, but because obsidian from A1ca was widely used at Marcavalle, Chanapata, and other sites in Cuzco during the late Initial Period and Early Horizon (Burger and Asaro 1979),it is likely that the Alca area may have been already inhabited. Without supplementary excavation, site surveys in the highlands frequently have difficulty documenting the period of early settlement, particularly when sites are occupied continuously for long periods. At the present time, the first unambiguous evidence for permanent settlements in the

193-

Burger et at.: Alca Obsidian Source

Cotahuasi area dates to the Middle Horizon. Significantly, two sites in the area, Collota near Cotahuasi and Mawka Llaqta in the upper end of the watershed have the distinctive architecture, known as orthogonal cellular, that is often viewed as the hallmark of Wari administrative centers (Isbell and McEwan 1991). In addition, smaller villages tentatively dated to the Middle Horizon by Trawick overlook substantial terracing and are associated with lengthy canal systems. Given the apparent ties of the Cotahuasi area to the Wari state, we should not be surprised to find that obsidian from the Alca Source appears at the Wari provincial centers of Jincamocco in Lucanas (Schreiber 1992) and Cerro Balli in Moquegua (Moseley et aZ. 1991), as well as at the site of Wari itself. Even more striking were. numerous projectile points of Alca Source obsidian recovered by Uhle and McCown in Huamachuco, far to the north of .Wari (Burger and Asaro 1979). Considering the importance of the Alca Source prior to the Middle Horizon, it is possible that this raw material as well as the agricultural and other resources of the area were responsible for Wari interest in the watershed. The "cWef town" in the Cotahuasi area immediately prior to the Inca invasion was located at Alca [Allca] and the critical battle to conquer this region took place there, according to Garcilaso de la Vega (Garcilaso 1966[1609]: 152-3). Judging from earlier provenience studies, exploitation of the Alca Source continued into the Late Horizon: residents of Inca sites in the Cuzco Valley utilized tools of Alca Source obsidian even .after the expansion of Tawantinsuyu. It is probably not coincidental that the Incas chose to locate their own center for administering the Cotahuasi watershed in the Alca area, on a steep fortified ridgetop that lies directly above the obsidian deposit. This site (known as Qhawana or Qhaqha), first documented archaeologically by Trawick (1994), features dwellings with trapezoidal doors and niches, as well as high gabled roofs. It has two plazas, each of which is flanked by large rectangular buildings. A portion of the paved stone Inca highway system passes through the site and then climbs up to the puna.

The sparse information summarized here suggests that the Cotahuasi watershed was an important locus of political power during the Middle Horizon and Late Horizon. Besides the area's significance as the source of rare high-quality volcanic glass for tools and other purposes, the drainage may have been a major supplier of gold, which occurs in unusually rich veins rather than alluvial accumulations and was apparently mined in pre-Hispanic times (Petersen 1970:68; Trawick 1994). Another local resource is rock salt, which was mined across the valley from Cotahuasi near the village of Huarhua (Llano 1904[1761]). This material continues to be highly valued and widely traded by pastoralists in the region today (Inamura 1981:72). The movement of obsidian, gold, and salt in prehistoric times was facilitated by the Cotahuasi area's location along a natural route of communication between Cuzco and the Peruvian south coast. This route can be broadly described as follows. After crossing the lengthy puna to the sOllthof Cuzco, the route descends the Cotahuasi or Huarcaya drainage, passing the source of Alca obsidian; below the town of Cotahuasi the road detours out of this drainage to avoid the deeply cut canyon mentioned previously, crosses a section of puna, and continues its descent along the Chala drainage immediately to ~e north. Unfortunately, the pre-Hispanic road system in this region remains poorly known (Hyslop 1984) and additional field work is necessary to confirm its use by Inca and pre-Inca peoples. Nevertheless, there are some suggestive data available. In his study of the Inca road system, Victor von Hagen and his associates identified this route up the Chala River as the transverse road through Kunti Suyu along which fresh fish was brought to the Inca capital (von Hagen 1955:234-252). Hyslop likewise shows this route as one of the principal transverse coast-highland roads, although he did not carry out field work there (Hyslop 1984:262,308). Yon Hagen illustrates several views of the coastal section of the road. In these photographs the road can be seen to be wide, bordered with stones and with sections

ANDEAN PAST 5 (1998) of paved stairs. Although von Hagen did not document most of the highland section of the lateral road, he did trace a portion of it from where it splits off from the main coastal road and ascends the Chala valley through the Parcoy Canyon towards the sierra. From a lowflying airplane, he was able to observe a highland portion of this same road and viewed where it disappeared behind Lake Parinacochas, and then made its way toward Cuzco (von Hagen 1955:245). We know from both Garcilaso and later accounts that a road from Lake Parinacochas continues across the puna into the Cotahuasi drainage, from where it continues its ascent into the Cuzco region. Hiram Bingham's mule train followed this route during his 1911 explorations (Bingham 1922:50-75). Long before, Garcilaso de la Vega (1966: 152-153) described how the Inca emperor Maita Capac advanced fro~ Cuzco across the Apurimac River through Chumbivilcas province [Chumpivillca] encountering resistance at Villilli [Velille], then went across 16 leagues of high desert into the -ruggedAlca area. After defeating the local population, he continued down the Cotahuasi valley through the settlements of Taurisma and Cotahuasi before crossing into the neighboring drainage to visit Lake Parinacochas. It is possible that this same route was followed by pre-Inca travelers and was subsequently incorporated into the transverse Inca road. Current use of this ancient route offers some insight into the journey required to transport Alca obsidian to the Cuzco basin. At the present time, the trip on horseback from Alca to Cuzco takes about six days (Trawick 1994:33). If the trip were made by llama caravans carrying loads of obsidian, the duration of the journey would more than double. Thus, it would appear that the peoples of the Cotahuasi area played a significant role during much of Andean prehistory, not only as the herders of alpacas and llamas on the high pasture lands and the farmers of an oasis-like setting within the arid highlands of Arequipa, but also as the suppliers of obsidian and other rare raw materials. It remains for fu~e in-

-194 vestigators to reveal more clearly the changing context within which the Alca Source of obsidian was mined and exchanged with the inhabitants of the Cuzco Valley and more distant regions. Acknowledgments We would like to thank the National Science Foundation and the Fulbright-Hays Commission for support of the anthropological investigations in Cotahuasi, and The Yale Council for Archaeological Studies and The Fisher Foundation for additional support in Peru. We appreciate the meticulous care of David Gan in preparing and weighing the samples reported in this work. We are indebted to Melissa Gibbs, Irradiation Coordinator, and the staff of the University of MissouriColUmbia Research Reactor Facility for the neutron irradiations necessary for this project. This work was support~d by the Department of Energy under contract DE-ACO 3-76SF 00098.
References Cited
Alvarez, L.W., W.A. Alvarez, F. Asaro, and H.V. Mi-

chel 1980 Extraterrestrial Cause for the CretaceousTertiary Extinction: Experiment and Theory. Sci-

ence 208:1095-1108.

Alvarez, W., F. Asaro, H.V. Michel, and L.W. Alvarez 1982 Iridium Anomaly Approximately Synchronous with Terminal Eocene Extinctions. Science 216:886-888. Alvarez, L.W., F. Asaro, F.S. Goulding, D.A. Landis, N.W. Madden, and D.F. Malone 1988 Instrumental Measurement of Iridium Abundances in the Part-per-Trillion Range Following Neutron Activation. Book of Abstracts of the 196th ACS National Meeting at Los Angeles, Sept. 25-30, 1988. Washington, DC: American Chemical Society. NUCL-30. Asaro, Frank, Emesto Salazar, Helen V. Michel, Richard L. Burger, and Fred H. Stross 1994 Ecuadorian Obsidian Sources Used for Artifact Production and Methods for Provenience Assignments. Latin American Antiquity 5:257-277. Bingham, Hiram 1922 Inca Land: Explorations in the Highlands of Peru. New York: Houghton-Mifflin. Burger, Richard L. 1980 Trace-element Analysis of Obsidian Artifacts nom Pachamachay, Junin. In Prehistoric Hunters of the High Andes, by John Rick, pp. 257-261.
New York: Academic Press.

1951981 La procedencia de artefactos de obsidiana en los sitios fonnativos en Ayacucho: Chupas y Wichqana. Boletin del Museo Nacional de Antropologia y Arqueologia 7:9-10. Lima. 1984 Archaeological Areas and Prehistoric Frontiers: the Case of Fonnative Peru and Ecuador. In Social and Economic Organization in the Prehispanic Andes, edited by David Browman, Richard L. Burger, and Mario Rivera, pp. 37-71. BAR International Series 194. Burger, Richard L. and Frank Asaro 1977 Trace Element Analysis of Obsidian Artifacts from the Andes: New Perspectives on Pre-Hispanic Interaction in Peru and Bolivia. Berkeley: Lawrence Berkeley Laboratory Report LBL-6343. 80 pp. 1978 Obsidian Distribution and Provenience in the Central Highlands and Coast of Peru during the Preceramic Period. Contributions of the University of California Archaeological Research Facility 36:61-83. Berkeley. 1979 Amilisis de rasgos significativos en la obsidiana de los Andes Centrales. Revista del Museo NacionaI43:281-326. . 1993 La distribuci6n y procedencia de artefactos de obsidiana durante el Periodo Inicial y Horizonte Temprano. In Emergencia de la civilizaci6n en los Andes: ensayos de interpretaci6n, by Richard L. Burger, pp. 180-231. Lima: Universidad Nacional Mayor de San Marcos. Burger, Richard, Fran\cAsaro, and Helen Michel 1984 The Source of Obsidian Artifacts at Chavin de Huantar. In The Prehistoric Occupation ofChavln de Hutintar, Peru, by Richard L. Burger, pp. 263270. Berkeley: University of California Press. Burger, Richard L., Frank Asaro, Helen V. Michel, Fred H. Stross, and Ernesto Salazar 1994 An Initial Consideration of Obsidian Procurement and Exchange in Prehispanic Ecuador. Latin American Antiquity 5:228-255: Burger, Richard L., Sergio Chavez, and Karen MohrChavez n.d. Through the Glass Darkly: Analysis of Obsidian trom the Preceramic through Early Intennediate Period in Southern Peru and Northern Bolivia. Manuscript in the possession of the authors. Chavez Chavez, Jose Antonio 1982 Evidencias arqueo16gicasen la cuenca del Rlo Cotahuasi-Ocona. Unpublished Tesis de Bachiller, Universidad Nacional de. San Agustin, Arequipa. Garcilaso de la Vega (EI Inca) 1966[1609] Royal Commentaries of the Incas and General History of Peru: Part One. Translated by Harold Livennore. Austin: University of Texas Press. Hyslop, John 1984 The Inka Road System. New York: Academic Press.
Inamura, Tetsuya

Burger et al.: Alca Obsidian Source


cos del Peru meridional, edited by Shozo Masuda, pp.65-83. Tokyo: University of Tokyo Press. Isbell, William H. and Gordon F. McEwan 1991 A History of Huari Studies and Introduction to Current Interpretations. In Huari Administrative Structure: Prehistoric Monumental Architecture and State Government, edited by William H. Isbell and Gordon F. McEwan, pp. 1-17. Washington, DC: Dumbarton Oaks. Linares MAlaga,Eloy 1990 Prehistoria de Arequipa. Arequipa: CONCYTEC, Universidad Nacional de San Agustin. Llano Zapata, Jose 1904[1761] Memorias hist6rico-jisico-apologeticas en el America Meridional. Lima: Imprenta y Libreria de San Pedro. Michel, H.V., F. Asaro, and W. Alvarez 1991 Geochemical Study of the Cretaceous-Tertiary Boundary Region in Ocean Drilling Hole 752B. Proceedings of the Ocean Drilling Program, Scientific Results 121:415-422. Michel, H.V., F. Asaro, W. Alvarez, and L.W. Alvarez 1990 Geochemical Study of the Cretaceous-Tertiary
- Boundary in

ODP Holes 689B and 689C. Pro-

1981 Adaptaci6n ambiental de los pastores altoandinosen el sur del Peru. In Estudiosetnograji-

ceedings of the Ocean Drilling Program, Scientific Results 113:159-168. Moseley, Michael E., Robert Feldman, Paul Goldstein, and Luis Watanabe 1991 Colonies and Conquest: Tiahuanaco and Huari in Moquegua. In Huari Administrative Structure: Prehistoric Monumental Architecture and State Government, edited by William H. Isbell and Gordon F. McEwan, pp. 93-119. Washington, DC: Dumbarton Oaks. Neira, Maximo 1994 Arequipa prehispanic~. In Historia general de Arequipa, by Maximo Neira, pp. 5-184. Arequipa: Fundaci6n M.l. Bustamente de la Fuente. OIchauski, E. and David Davila 1994 Geologia de los cuadrangulos de Chuquibamba y Cotahuasi. Boletin 50, Instituto Geo16gicoMinero y Metalmgico. Lima. Perlman, I. and F. Asaro 1969 Pottery Analysis by Neutron Activation. Archaeometry 11:21-52. 1971 Pottery Analysis by Neutron Activation. In Science and Archaeology, edited by R.H. Brill, pp. 182-195. Cambridge, MA: MIT Press. Petersen, Georg 1970 Minerfa y metalurgia en el antiguo Peru. Arqueo16gicas 12. Lima: Museo Nacional de Antropologfa y Arqueologia. Schreiber, Katharina J. 1992 Wari Imperialism in Middle Horizon Peru. Anthropological Papers 87, Museum of Anthropology, University of Michigan, Ann Arbor. Stross, Fred H., Payson Sheets, Frank Asaro, and Helen V. Michel 1983 Precise Characterization of Guatemalan Obsidian Sources and Source Detennination of Artifacts trom Quirigua. American Antiquity 48:323346.

ANDEAN PAST 5 (1998)


Trawick, Paul B. 1994 The Struggle for Water in the Andes: A Study of Technological Change and Social Decline in the Cotahuasi Valley of Peru. Unpublished Ph.D. dissertation, Yale University. von Hagen, Victor W. 1955 Highway of the Sun. New York: Duen~Sloan, and Pearce.

-196

197-

Burger et al.: Alca Obsidian Source

Table 1. Comparison of diagnostic element abundances- measured in Alca, Cotahuasi Province, Peru source samples with the previously measured Cuzco Typeb chemical Group.
FJement

! .

Rosa-1
(X)

I I

Rosa-2

(Y)

(X-Y)/ i Y (%)

Cuzco Type'

(C)

(C-X)/! y(%)

(C-Y)/ Y(%)

Standanf

abundances

Best-measured elements Sc Th Cs Fe % Eo Ta Rb 1.90 i 14.9 I 2.87 i ! 0.574 I 0.506 ! 0.904 i 136.6 ! Z 1.7 i

1.95 15.1 2.90 0.571 0.505 0.914 143.2

-2.6
-1.3 -1.0 +0.5 +0.2 -1.1 -4.6 1.6

1.95
14.75 2.90 0.565 0.495 0.895 133

Z
Z Z % Z Z Z

0.03
0.12 0.13 0.015 0.010 0.012 9

+2.6 -1.0 +1.0 -1.6 -2.2 -1.0 -2.6 1.7

0.0 -2.3 0.0 -1.1 -2.0 -1.0 -7.1 U

20.55 13.96 8.31 1.017 1.291 1.550 64.5

Z Z Z Z Z Z Z

0.33 0.39 0.550 0.012 0.034 0.044 3.2

AveragedeviatioD(%)

Elementspoorly measuredin eitherthe Cuzco7'ypechemicalgroupor Alca sourcesamples Ce4 Co Zn i 61.0 0.246 z.OO9 37.4 61.3 0.233 z.OO6 38.8 -0.5 +5.6 -3.6 59.0 0.22 Z ::I: 1.4 0.07 80.3 14.06 457 Z Z ::I: 3.9 0.15 22

-10.6

-5.6.

Abundancesare givenin ppm exceptfor those of Fe, which are given in percent (%). One sigma R. Burger and F. Asaro (1977). The publishederrors for the CuzcoType obsidiangroup are the

counting errors in this work are 1% or less unless a larger value is indicated.
b

root-mean-square deviations for 24 artifacts. As some of the Rb measurements were based on flux monitors, those Rb values have been recalculated relative to the standard (Standard Pottery) used in the measurements. As suggested by Asaro et ale (1994), a correction of 0.048:f: .059 ppm was removed from the old LBL Co abundances.
c S~dard Pottery (perlman and Asaro 1969)was used to calibratethe abundancesof all the listed elements,exceptZn. The abundancesand one-sigmauncertaintieswere taken from Perlman and Asaro (1971). The Rb abundances,however,have since been revisedto the listed values. Zinc was calibrated versus the standardDINO-l (Alvarezet ale 1982)and the abundance(in the DanishCretaceous-Tertiary boundarysample)was reported in Alvarez et ale (1980). The accuraciesof the measurements,whichare useful for comparingwith the work of other laboratories calibratedversus different standards, must incorporateboth the precisionsof the measurementsand the uncertaintiesin the standard.

d Ce has not been corrected for 235U fission in either the Cuzco Type chemical group or the A1ca source samples.

ANDEANPAST 5 (1998)

-198

Table 2. Backgroundabundancesfrom contaminationin the LWAICSand the 99.9999%-pure


AI encapsulating foil. a
Element

Energy of 'Yrays (keV)

LWAICS bkg

Averages for 10 empty capsules

Rosa-l AI capsule

Rosa-2 AI capsule

Elements without obvious heterogeneous contamination . in (or on) the Al encapsulating . foils Sc Th Ceb Cs Fe %. Eu Zn Ta Rb Co Hf Hf 1121 i 312 145 i 796 1099! 344-779 1116 1189 1077 1173-1332 482 133-482

I I

0.006 0.003 0.005 0.007 0.001 0.004 0.09 0.001 0.0 0.055 0.000 0.000

:I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :1::I:

0.000 0.002 0.008 0.000 0.001 0.000 0.01 0.000 0.3 0.003 0.001 0.000

0.372 0.003 0.008:1: 0.000:1: 0.002:1: 0.000:1: 6.73 0.000:1: 0.0 0.017:1: 0.004:1: 0.002:1:

:I: :I:

:I: :I:

0.004 0.014 0.036 : 0.000 0.002! 0.000 0.55 i 0.000 0.3 0.011 0.010 . j 0.003 I

0.391

0.375

0.003:1: .001 i 6.29 :I: .12!

0.002:1:.000 5.90 :I: .07

0.018:1:.002 0.000:1:.004 0.000:1:.004

I
I . i
I

0.018:1:.002 0.000 :I: .002 0.000 :I: .002

Elements with obvious heterogeneous contamination in (or on) the Al encapsulating foils which may be due in part to laboratory contamination Sb

1691

i
I

0.002

:I:

0.001

0.083:1:

0.006

0.168 :I: .014

0.121 :I: .011

Se
(Ppb)

136-265!.

7.3

:I:
54.5

0.9

17

:I:
54.5

22

133 :I: 10
53.6

64 :I:4
55.4 259.2

Weight for abundance calc. (mg) True weight (mg)

254.9

:I:

2.0.

256.0

Abundances are givenin ppmexceptfor thoseof Fe, whichare givenin percent(%), andthose

of Se, which are given in parts-per-billion(Ppb). Listed errors for individual samples are the best estimatesof the one sigma values of the precision of the measurement. Where no error is given, the precisionis estimatedas 1%. The errors for the group of 10 emptyAl capsulesare root-mean-square deviations. Wheretwo gammaray energiesseparatedby a hyphen are givenfor an element,radiations of those energieswere used in coincidencemeasurements.
b Ce abundances have not been corrected for 23SU fission.
C

Mean and root-mean-squaredeviation.

199-

Burger et aI.: Alca Obsidian Source

Table 3. Backgroundabundancesfrom contaminationin the LWAICSand 99.9999%-pureAl


encapsulating foil. a
Flemeat EDergy01 'Yrays (keY) Splits 01irradiated Rosa-l sample Rosa-l A 20.75 Rosa-l B 30.38
Splits 01 irradiated Rosa-2 sample

Rosa-2A 25.60

Rosa-2B 28.45

Weight(mg)

Elementswithoutobviousheterogeneous contamination in (or on) the Al encapsuladngfoils


Sc 111 Ce' Cs Fe % Ell Zn Ta Rb Co

1121 312 145 796 1099 344-779 1116 1189 1077 1173-1332

1.964 15.25 62.9 2.92 0.591 0.519 38.5 0.923 141.0 0.263

:f: :f:

2.1 0.012

1.977 15.33 62.9 2.95 0.598 0.516 38.4 0.926 141.9 0.262

:f: :f:

2.0 0.009

2.000 15.500 '63.400 2.970 0.588 0.522 39.500 0.932 147.7 0.241

:f: :f:

2.5 0.010

1.933 15.010 61.300 2.890 0.572 0.508 38.300 0.914 144.3 0.236

:f: :f:

2.5 0.010

Elements displaying heterogeneity within the irradiated samples oj Rosa-lor Rosa-2 Al encapsulatingjoils Hf Hf

Rosa-2 but not appearing as a contaminant on the Rosa-lor

482 I 133-482!.

3.54 3.57
within the irradiated

4.04 4.03
samples oj Rosa-lor

3.90 3.88
Rosa-2 and also appearing

3.60 3.58
as a contaminant on the Rosa-lor

Elements displaying heterogeneity Rosa-2 Al encapsulatingjoils

Sb So (Ppb)

1691! 0.540 136-265 i 154

:f: :f:

0.015 0.707 6 ! 403

:f: :f:

0.016! 0.416 7 i 103

:f: :f:

0.016! 5

i 117

0.047

:f: :f:

0.016 5

a Abundancesare given in ppm exceptfor those of Fe, which are given in percent (%), and those of Se, which are given in parts-per-billion(Ppb). Listed errors for individualsamples are the best estimatesof the one sigma values of the precision of the measurement. Where no error is given, the precisionis estimatedas 1%. Where two gammaray energiesseparatedby a hyphen are given for an element, radiationsof those energieswere used in coincidencemeasurements.
b Ce abundances have not been corrected for 235U fission.

a
b

Abundancesare given in ppm except for those of Fe, whichare given in percent(%).
A.C.E

= Average Counting Error

of obsidian sample measurements. I t-.) 8

E.r.

= Error residue = 100 * [Square Root (RMSOZ- A.C.E?)]I Mean

d Ce abundances have not been corrected for 23SUfission.

201-

Burger et al.: Alca Obsidian Source

o I

20 40 60 60 100 I I I I I Kllomete,s

KEY

. obsidian
source

. referred places
to in text

Figure 1. Locations of various obsidian"sources in southern Peru.

ANDEAN PAST 5 (1998)

-202

o I

1 I

2 I
Kilometers

3 I

4 )

@
.

obsidian source location of analyzed source specimens modern settlement

Figure 2. Detailed location of the Alca obsidian source in southern Peru

THE CHIV AY OBSIDIAN SOURCE AND THE GEOLOGICAL ORIGIN OF TITICACA BASIN TYPE OBSIDIAN ARTIFACTS Richard L. Burger Yale University Frank Asaro Lawrence Berkeley National Laboratory Guido Sala~ Universidad Nacional de San Agustin

Fred Stross LawrenceBerkeleyNationalLaboratory


Background Of the major chemical types of obsidian utilized for tools by pre-Hispanic cultures in the Titicaca Basin of southern Peru and northern Bolivia, one chemical group overshadowed all others in frequency regardless of time period. In our initial study of the provenience of obsidian artifacts carried out at what is now called the Ernest Orlando Lawrence Berkeley National Laboratory (LBNL). between 1974-1975, obsidian with this chemical signature was referred to as Titicaca Basin Type obsidian for ease of discussion (Burger and Asaro 1977, 1979). Obsidian from this unlocat~d obsidian source was used at major Peruvian sites in the Department of Puno' such as Qaluyu and Sillustani and also at Bolivian sites on the other side of Lake Titicaca, such as Qallamarka and Tiahuanaco. Our subsequent research (1977-1978) confirmed this general pattern of pre-Hispanic obsidian distribution and, in addition, demonstrated that this type of obsidian was the predominant type present at six archaeological sites in southern Arequipa (Burger et al. n.d.). The abundance of this chemical type of obsidian at Preceramic sites such as.Sumbay and the volcanic character of southern Arequipa led us to suggest that the source of this obsidian might be found in this region rather than in the Titicaca Basin (Burger and Asaro 1993:222-223, 230), but the absence of detailed geologic data on obsidian deposits hindered further progress. obsidian from a mine in the Chivay area in the Colca Valley of Arequipa, and x-ray fluorescence (XRF) analysis of the sample at LBNL suggested a match with the Titicaca Basin Type obsidian artifacts. Unfortunately, information was not available concerning the exact location where the sample had been collected.
_

Subsequently, Sarah Brooks, a University of Wisconsin geographer working in the Colca Valley, encountered small obsidian pebbles at 3700-3800 meters above sea level (masl) on the flanks of Pampa Finaya, across the river and approximately 1.5 km west from the town of Chivay. Collaborative research with Brooks and Michael Glascock at the Missouri Research Reactor (MURR) demonstrated through instrumental neutron activation analysis (INAA) that a chemical match existed between these samples and the composition of Titicaca Basin Type obsidian artifacts (Brooks et al. 1993). However, the consistently small size of these pebbles precluded the possibility that Pampa Finaya was the source for the TiticacaBasinType obsidianartifacts. The Chivay Obsidian Source: Location And Geology (Figures 2, 3) In an effort to locate more precisely the primary geologic source of this material, Burger suggested to one of his students, Eli Gould, that a reconnaissance visit to the Chivay area might be worthwhile, and in January of 1994, Gould successfully located quantities of obsidian above the Quebrada de los Molinos, roughly 4-6 km east of the modem town

In 1989, Rosalia Avalos de Matos provided Burger with a sampleof non-artifactual


ANDEAN PAST 5 (1998):203-223.

ANDEAN PAST 5 (1998) of Chivay. Nodules of unworked obsidian were found at approximately 4900 masl on a level surface at the foot of the western slopes of Cerro Ancachita (5131 masl); the obsidian was mixed with deposits of what appeared to be volcanic tuff. Gould observed obsidian specimens up to 30 em on a side. These were substantially bigger than those found at Pampa Finaya and sufficiently large to have served as source material for artifact production. According to a local farmer, a German resident of Bolivia had visited this obsidian deposit in order to collect obsidian for export to La,Paz, where it was to be transformed into craft products. Three of the samples studied in this work, Chivay-1, -2, and -3, came from this

-204 nodules also occur to the east of Cerro Ancachita in the area known as Pampa Ichocollo (4500-4900 masl), roughly 7 km east of Chivay. Obsidian blocks were on the southern slopes of Cerro Ancachita, and they are remarkably free of flaws and impurities. Some of the obsidian recovered has the same distinctive purplish hue observed in obsidian arti-, facts of the Titicaca Basin Type. The obsidian pieces are variable in shape and, as would be expected, are covered with cortex or weathering rind; most show no evidence of having been worked. However, an obsidian core and obsidian flakes were observed in the Quebrada de los Molinos. . The full extent of the primary deposit of obsidian has not been determined but the information available suggests that it extends over several kilometers. No effort was made to locate ancient quarry areas or workshops, and such essential information can only be obtained by additional research in the field. Four of the samples studied in this work, Chivay-4, -5, -6, and -7, came from this area. Sarah Brooks, who has been working independently in the Colca Valley, has recently reported locating a huge obsidian quarry and her work promises to shed light on obsidian procurement (Sarah Brooks, 1996, personal communication). 1

location.

In his brief reconnaissance of the Chivay area, Gould noted that some obsidian is found in Chivay itself (including the town's bullring), but these specimens appear to have been redeposited by erosional or ~uman forces. In contrast, the material from the slopes of Cerro Ancachita appeared to be in its original geological context. A June, 1998 survey of the glacial geology around Chivay by Harold W. Borns, Daniel H. Sandweiss, and Bernardino Ojeda determined that obsidian from Cerro Ancachita and .vicinity was brought down Quebrada de los Molinos and a small adjacent quebrada by glacial activity. Small nodules of this obsidian are present in morrainal deposits under the town of Chivay as well as in the Quebrada itself (H. Borns, personal communication). In July of 1995, Burger and Arequipa geologist Guido Salas traveled to the Chivay area to collect additional geological information and obsidian samples from the source area visited by Gould. During the walk up the Quebrada de los Molinos, Burger and Salas
.

observed obsidian and other volcanic rocks

eroding into the ravine from the volcanic deposits above. They also encountered a local farmer who had been collecting quantities of obsidian from these deposits for future sale. Field observations indicate that the obsidian deposit extends beyond the area on the western slopes of Cerro Ancachita documented by Gould. Large and small obsidian spherical

1 Following the submission and acceptance of this article, Brooks et al. (1997) published a short article in the Scientific Correspondence section of Nature in which an obsidian quarry whose chemical signature matches the Titicaca Basin Type is discussed. This quarry, which is referred to as Cotallalli, is located in the Colca Valley but its relationship to the obsidian source described here is difficult to determine without further information. No information is provided in the Brooks et al. article which would allow the quarry t~ be located with precision, and the general map illustrating the article shows Cotallalli to be 50 km north of Sumbay, which would place it some 20 km northeast of the area described here. We were unable to locate a volcano or mountain named Cotallalli in the current maps of Peru's Instituto Geografico Nacional or in the geological literature. This ambiguity will no doubt be resolved with the fuller publication of the work by Ms. Brooks and her colleagues.

205In 1997, Salas returned to the Chivay area for two days in order to document better the context of the obsidian within the local geological formations. To the south of Cerro Ancachita, Salas identified a large rhyolitic dome, known as Cerro Hornillo (Figure 4). The dome extends for approximately 2.6 Ian (N-S) by 2.2 km (E-W). Along its edges are vitrified deposits (i.e., obsidian), including nodules and large blocks of volcanic glass. The vitrification along the northern edge of the dome accounts for the obsidian encountered in the earlier visits to Cerro Ancachita, and a still larger concentration of obsidian blocks was encountered on the dome's western edge (Figure 5), some 5 km west of Chivay. A reading by Salas using the Global Positioning System, GPS, measured the location of the dome's obsidian deposits, and it yielded the following location (in Universal Transverse Mercator, UTM, coordinates): E.227,234 and N 8,268,421. The obsidian source has a generallocation of 1531'13"-1532'46" S latitude, 7138'6" W longitud;? We propose to refer to this obsidian deposit as the Chivay Source, because of its proximity to the wellknown town of Chivay. Judging from a regional geologic study by Peru's INGEMMET
Geol6gico Minero y Metalmgico)

Burger et al.: Chivay Obsidian Source appears to occur where lavas have cooled rapidly when they came into contact with the contiguous . deposits of the older Tacaza Group. Significantly, Pampa Finaya, like Cerro Ancachita and Pampa Ichocollo, corresponds to the Barroso Group. The intervening area between Pampa Finaya and Cerro Ancachita, including the land beneath the modern town of Chivay, consists of later fluvial con~ glomerates, lacustrine materials, and glacial morraines deposited during the Pleistocene. Instrumental Neutron Activation Analysis In the INAA process, there are measurement uncertainties concerned with counting gamma rays (called counting errors) which can be made smaller by counting for longer periods of time, irradiating larger amounts of sample, or irradiating for longer periods of time. As a practical matter there are other significant errors (which can overshadow small counting errors) which come from a multitude of sources and sometim~s can be attacked only one at a time. In earlier studies a 1% precision in the abundance values was about the best that could be accomplished, although between 1% and 2% was more routine. The present measurements of element abundances by neutron activation analysis are considerably more precise than previous ones. Sample preparation As the details will be described in another publication, only a brief sketch will be given here. The obsidian nodules collected in the field were broken, and the pieces sent to LBNL. There, some of the pieces were crushed and coarsely powdered with an agate mortar and a pestle. Powder samples of about 100 mg were weighed with a precision better than 0.1 mg and encapsulated in a weighed amount of 99.9999% pure Al metal. Each capsule was completely wiped twice with a cotton swab wetted with ethyl alcohol in order to remove any loose Al flakes or sample powder. The cleaned capsules containing obsidian powder along with many other samples were placed in stacks of 38 in quartz tubes that had been sealed on the bottom. Each stack had an empty Al capsule at the top. One tube (out of

. (Instituto

that included Chivay (Palacios et al. 1993), the volcanic deposits containing the obsidian belong to the Barroso Group, which, according to the INGEMMET study, includes andesitic lavas, trachytes, and tuffs (such as pumice). In the Chivay area, the Barroso Group consists primarily of andesitic lavas with concentrations of plagioclase crystals. The Barroso Group includes transversal flows that are responsible for ridges and crests, such as Cerro Saylluta and Cerro Ancachita, and the volcanic dome known as Cerro Hornillo. These Barroso Group deposits (Ts-Ba) directly overlie strata of the Tacaza Formation (Tm-Ta) which are deep (200 m) deposits of lavas and andesitic breccias that date to the Early/Middle Miocene. This superposition can be observed at the lower end of the Quebrada de los Molinos. In very rough chronological terms, the Barroso Group dates to the Late MiocenelPliocene between 6 million years ago (Ma) and 1 Ma (Palacios et al. 1993:191). Salas observed that the obsidian

ANDEAN PAST 5 (1998) a total of 7) contained 16 standards distributed at the bottom, middle, and top of the tube. For calibrating most elements, the standards were 25, 50, and 100 mg of Standard Pottery (Perlman and Asaro 1969, 1971) at each of the three positions. For calibrating Ir and Zn, the standards were 25 and 50 mg samples of DINO-1 (Alvarez et a1. 1982) at each of the three positions. One sample of CaC03 was also in the standard capsule to measure any impurities introduced in the sample preparation process. Four of the obsidian samples, Chivay-4, -5, -6, and -7, were in the same quartz tube as the standards. The other three obsidian samples, Chivay-1, -2, and -3, were in adjacent positions in another tube. After correcting for half-lifes, time of decay, sample weight, counting time, counting rate, and interferences, calibration coefficients were calculated for each element of interest for each standard. The calibration coefficients (abundance per 100 mg weight per count-perminute corrected to the end of irradiation) for the various elements in Standard Pottery were respectively about 1.2% and 6.0% higher in the middle and top of the tube with the standards than at the bottom, and the exact values for each element were interpolated for each position in that tube. The same calibration coefficients had to be used for the obsidian in the other quartz tube containing obsidian as it contained no standards. As will be shown later, the abundances of samples Chivay-4, -5, and -7 from the tube with the standards and Chivay-1 and -2 from the other tube with obsidian agree very well. This indicates that fortuitously, the two tubes had not shifted significantly in the axial direction with respect to each other. In the future for very precise work, multiple standards of Standard Pottery will be included in all tubes. The open quartz tubes were heated to 1900 C in an oven for several hours to drive off water and prevent the subsequently sealed tubes from cracking because of internal pressure during the irradiation. The quartz tubes were each individually tightly wrapped in kitchen foil to promote heat transfer during the irradiation and keep the temperature in the stacks below the melting point of AI, and the cluster of 7 tubes was loosely wrapped in kitchen foil for the same reason. The cluster of tubes was sent to the

-206 University of Missouri reactor in Columbia, Missouri and irradiated for 48 hours at a flux of 2.5 x 1013 neutrons per second per cm2. One month after the end of irradiation, the samples were sent back to LBNL, and subsequently the Al was unwrapped from the quartz tubes, and the tops of the latter were sawed off. As needed, samples were removed from' the tubes and completely cleaned with cotton swabs wetted with ethyl alcohol about 4 times. Because small leakages of powder had occurred since the previous weighings, it was necessary to reweigh some of the obsidian samples after the irradiation and use those weights in the calculations. Measurements Measurements on the standards and obsidian samples were made with the Luis W. Alvarez Iridium Coincidence Spectrometer. This instrument had been designed and constructed specifically to measure instrumentally and with high sensitivity the abundance of iridium in deep sea sediments. It has been modified to measure many other elements simultaneously with Ir, and recent changes in the instrument permit significantly higher precision. The instrument can measure gamma rays with either of two Ge detectors or coincidences with both detectors. The measurements can be made with or without an antiCompton shield, which reduces background from scattered gamma radiation. For obsidian studies, only the gamma ray detector with the best resolution was used for singles measurements of 9 elements and coincidences were used for Co, Hf, and Eu measurements. Selenium abundances were also very sensitively measured, but Se abundances are very sensitive to low-level laboratory contamination and are not included in this report. Count rates of samples in each Ge detector can go as high as 150,000 per second and still give useful results, but 40,000 per second is optimal for coincidence measurements. The obsidian samples in the present work counted about 14,500 per second; the empty Al foils counted about 700 per second. The anticoincidence shield normally counts about

207500,000 counts per second, but was not used for the obsidian measurements. The standards of Standard Pottery and DINO-I were measured first, then the instrument background was determined and then the average impurity levels in or on the empty Al capsules. These levels for a 100 mg sample are shown in Table I. It is seen that there is a very large Sc impurity in the Al foil, but it can be accurately subtracted from each sample because it is very homogeneous in the foil. Zinc, on the other hand, is also present as a large component in the Al foil, but it is not very homogeneous and therefore adds about 2% uncertainty to each Zn measurement. (Unwrapping the samples and counting the AI wrapping foils by themselves would remove this uncertainty.) The levels of background and Al impurities appropriate for each sample's weight are automatically removed from the ab~dances. One Al empty capsule had a large impurity of Co, over 0.1 parts-per-

Burger et al.: Chivay Obsidian Source the ratio. The geometry for the singles measurements is calculated from the sample position. No corrections are needed for geometry variations in the radial direction. No geometry corrections at all are needed for coincidences measurements as movement toward one detector is compensated by movement away from the other. Errors Known random errors in the present work come from three sources. One is the error as.sociated with counting radioactivity and is random for all elements in each sample. Another is concerned with the error in the efficiency measurements. This is random for each sample, but every element measured by singles measurements for a given gamma ray count will have the same error. The errors in the coincidence measurements will be slightly larger, will also vary coherently among themselves, and in large part will vary coherently with the variations in, the singles measurements. What one might then see is that all element abundances are larger or smaller in one sample compared to another by a nearly fixed amount. There is also a random error for the singles measurements due to uncertainties in the measurement of the 46SCgamma rays necessary to determine the exact position of the sample between the Ge detectors as it is being counted. This error, called the geometry error, will be the same for all elements determined by singles measurements and is somewhat smaller than the error due to the pulser efficiency. There will be systematic errors in measurements of some gamma rays because they contain interferences from other radiations. We have studied these interferences in the past and corrected the data for them, but our studies have been at the 1% level of precision, not the 0.1% level which is desirable for 0.3% overall precision for the data taken over an extended period of time or by different laboratories. Where we had a choice of radiations or techniques we used that which had the least interferences or other problems.

million (ppm). . This value.was considereda


low-probability contamination fluke and .was deleted from the subsequent calculations. It is necessary to -determine the Ge detector efficiency (which varies as a function of count rate) for each sample for both singles and coincidences measurements. A pulser alternately feeds pulses corresponding in amplitude to that of the 46SC 889 keV gamma ray to one detector and then pulses corresponding in amplitude to that of the 46SC1121 keV gamma ray to the other detector. These pulses are treated by the electronic systems in the same fashion as gamma rays. The efficiency for singles measurements is the ratio of the pulser pulses passing through the detector to the input rate (raised to the power 1.02 to adjust for variations with count rate). The coincidence efficiency is the product of the efficiencies from both detectors (with one raised to a power of 0.427 or higher depending on the count rate in the Ge detector). The position of the sample along the central axis between the two detectors can vary by a few millimeters. The exact position is determined automatically by counting the 46SC1121 keV gamma ray abundances in each detector and determining the position for each sample from

ANDEAN PAST 5 (1998) Results Five Chivay samples had a very similar chemical abundance pattern and their data are shown in Table 2. Also shown are the mean abundances for each element for the group of 5~their root-mean-square deviations and coefficients of variation (C. of V. = root-meansquare deviations divided by the mean values). For the 6 elements with counting errors better than 0.3% the average coefficient of variation was 0.26%. This value~ which is an upper limit to the inhomogeneity in the measured samples~ is slightly larger than expected (0.19%) from the average counting error (0.13%) and the errors in measurement of the gamma-ray efficiency (0.11%) and the gamma-ray geometry (0.08%). Table 3 shows the data for samples Chivay-6 and Chivay-3. Chivay-6 differs from the main group by very close to +0.4% for all of the 6 best measured elements and the abundances of the other elements are all consistent with that value within thei~ counting errors, except for that of Sb~which is low by about 6%. Except for Sb~ an error of -0.4 mg in weight~-0.4% in the sample geometry or efficiency calculations~or an error of +3.5 mm in the sample position in the reactor irradiation could all give the same effect. Because the weight was checked after the irradiation and agreed to 0.06%~the difference is not due to a loss in weight. (This agreement is somewhat fortuitous because small amounts of water in the obsidian could be lost in the irradiation.) The sample was in the same capsule as the standards of Standard Pottery~and also in its proper position in the irradiation quartz tube with samples Chivay-4, -5 and -7~hence there was not a shift of 7 mm from its assigned position. Table 2 shows that the average coefficient of variation for the three runs for the 6 best measured elements is only 0.19%~ consistent with the expected counting, geometry, and efficiency errors. So there was not a +0.4% error in the geometry or efficiency determination. We do not know the source of the +0.4% difference, but it may be due to the loss of water or one of the elements that we do not measure~ e.g. ~Si~AI~K, or 0, in the geo-

-208 chemical history of the sample~combined with a 6% loss in Sb. Chivay-3 is somewhat similar to the abundance profile of the main Chivay group~but is easily distinguishable because its Ta abundance is over 1% lower and Hf~ Fe~ and Co abundances are higher by 4~ 4~ and 16%~respectively. Table 4 compares the abundances of the main Chivay group with those of the Titicaca Basin Chemical Group of 21 artifacts measured by INAA at LBNL and published in 1977 and 1978. The old data readily fall into two precision groups: eight elements had a rootmean-square deviation (RMSD) of less than 4% and two elements had a RMSD of greater than 4%. The average deviation between the suites of data from the Main Chivay Source Group and the Titicaca Basin Chemical Group in the first group of elements is only 1.1% and the deviations in the second less precise group of elements are consistent with the uncertainties. This excellent agreement between the source samples and the previously analyzed artifacts confirms the assignment of the Titicaca Basin Chemical Group to the Chivay Source in Arequipa~and indicates that the old INAA data~while not as precise as the present work~ gave group values precise at the 1% level. As the source loci studied in the present work have obsidian suitable for artifact production~one or more of them may have been the origin of at least some of the artifacts assigned to the Titicaca Basin Chemical group in 1977 and 1978~ and the general volcanic deposit in Chivay from which the source samples were collected can be considered with a high degree of certainty as the origin of the obsidian used to produce artifacts of the socalled Titicaca Basin Chemical Group. Archaeological Ramifications (Figure 1) The location of the source of the Titicaca Basin Type obsidian above the town of Chivay in the Colca Valley has important implications for understanding the regional prehistory of southern Arequipa. It also provides crucial evidence for re-evaluating obsidian distribution patterns outside the Arequipa area.

209The latter task has been undertaken as part of a long-term collaboration with Sergio Chavez and Karen Mohr Chavez (Burger et al. n.d.) and the current discussion will focus more narrowly on the Colca Valley and the immediately surrounding area. With a few notable exceptions, the Colca Valley has been neglected by archaeologists until recently. Perhaps the most influential early study of Colca was the aerial photographic survey of the Shippee-Johnson Expedition in 1929 and 1931. Based on these efforts, dramatic oblique views of the extensive pre-Hispanic terracing in the Colca Valley were published (Johnson 1930; Shippee 1932, 1934). Many of these impressive agricultural systems were no longer in use and the Shippee-Johnson photographs became increasingly relevant as scholars became interested in the question of terrace and canal abandonment. Pioneering archaeological studies of the Colca Valley were carried out by Maximo Neira (1961, 1990), and these were complemented by the investigations of Eloy Linares MaIaga (1981, 1990) in the lower sections of the drainage (knoWn as Majes and Camana). Although these early studies were important contributions, the results stimulated little additional research. Finally, between 1984 and 1986, The Rio Colca Abandoned Terrace Project was carried out under the direction of geographer William Denevan. This project addressed the problem brought to the attention of scholars by the Shippee-Johnson Expedition, and involved the participation of a range of specialists to achieve this objective. From the perspective of this paper, it was particularly significant that several archaeologists (Pablo de la Vera Cruz, Michael Malpass, Daniel Shea) were included in the project. Their preliminary results were presented in 1985 at the 45th International Congress of Americanists in Bogota (Denevan 1987). Nevertheless, even after this renewed archaeological activity, the Colca's prehistory remains poorly understood and it has been largely neglected in recent syntheses of Central Andean prehistory (e.g., Bonavia
1991; Moseley 1992; Richardson 1994).

Burger et al.: Chivay Obsidian Source At the present time, little is known of Colca's prehistory before the Middle Horizon. However, it is likely that this is a function of the limited archaeological investigation carried out thus far, and a comprehensive program of survey and excavation, like that done in Moquegua, will probably produce a lengthier and more complex sequence. The procurement of raw obsidian from the Chivay source and its widespread distribution through what is now southern Peru and northern Bolivia indicates unambiguously that something significant was happening in Colca long before the Middle Horizon occupation detected . in the valley. Neira's excavations of preceramic occupations in the caves and rockshelters of Sumbay, located across the puna some 40 km to the south of Chivay, produced excellent evidence of hunting and gathering groups adapted to the puna environment (4127 masI). The third and fourth strata in the principal cave at Sumbay (Su-3) produced C-14 measurements of 3400:f:90 BC (BONN-1559) and 421O:f:120 BC (BONN-1558), respectively. These dates fit well with the typological study of the lithics (Neira 1990:50). Although retinite was the most popular lithic material used, obsidian artifacts were found in all strata at Su-3, including stratum 4 (ibid :28-34). As reported in detail elsewhere (Burger et al. n.d.), twentyfive samples were analyzed at LBNL and all of them, including one sample from stratum 4 of Su-3, proved to be from the Chivay Source. Thus, the prehistoric inhabitants of southern Peru knew of the Chivay obsidian deposit by the Middle Preceramic Period and exploited it for the production of projectile points, scrapers, and unmodified flakes. During the Preceramic Period, as in later periods, the manner in which the obsidian source was exploited cannot yet be determined. Was it procured from the source by outsiders or by local residents of the Colca Valley? Whichever the case, such activities were likely to have produced contact between the pre-Hispanic peoples of Colca and the puna-dwelling consumers of the raw material.

There is no informationon local cultural developmentin the Colca Valley for the next .

ANDEAN PAST 5 (1998) three millennia, but artifacts made from Chivay Source obsidian appear at early agricultural sites such as Qaluyu in the northern Lake Titicaca Basin and Chiripa in the southern Lake Titicaca Basin. Obsidian from the Chivay deposit also appears in small quantities at Pikikallepata near Sicuani in the Upper Vilcanota and Marcavalle in the Cuzco Basin. This pattern of obsidian distribution appears during the Initial Period and continues during the Early Horizon and Early Intermediate Period (Burger and Asaro 1979, 1993; Burger et al. n.d.). After some significant changes during the Middle Horizon, it re-emerges during the Late Intermediate Period and Late Horizon (Burger and Asaro 1979). Large amounts of obsidian from the Chivay source are known in the northern Titicaca Basin during the Late Intermediate Period at sites such as Incatunuhuiri,Llalli, and Sillustani. From the standpoint of Colca's prehistory, it is particularly significant that the main consumers of the Chivay Source obsidian were highland settlements in the Late Titicaca Basin. This is understandable given the location of the Chivay source and the surrounding topography. Natural routes lead up the valley from Chivay into the Colca's headwaters and across the puna and north into Llalli and Ayaviri -and east towards Juliaca. Although still another route leads west towards Chumbivilcas and Cuzco, the Alca source of obsidian is significantly closer (approximately 130 km) to the Cuzco Valley. Llama caravans from Cuzco could have saved weeks of travel by acquiring obsidian from the Alca Source rather than from the Chivay Source. By the Initial Period, domesticated camelids became widespread in the Cuzco Valley and the adjacent altiplano (Miller 1979; Miller and Burger 1995), and it is likely that the mining, transport, and exchange of obsidian from the Chivay Source would have reflected the role of llama caravans in the interzonal movements of bulky and heavy commodities such as obsidian (Browman 1974, 1975; Flores 1968). The appearance of artifacts made from Chivay Source obsidian at distant sites such as Chiripa and Tiahuanaco, suggests the degree to which obsidian became

-210 integrated into these larger long-distance exchange systems. Given the radical changes in sociopolitical structure experienced on the altiplano between the Initial Period and the Late Horizon, it is likely that the exploitation and distribution of Chivay Source obsidian underwent significant transformations, but its continuation in whatever form would have maintained contact between the populations o the Colca and those of the altiplano. If the ethnohistoric and ethnographic cases can serve as a guide, these exchange relationships usually involve social linkages that reinforce the economic relationships. Both archaeologists and ethnographers studying the Colca Valley have emphasized that its location makes it a natural point of contact between quechua zone farmers and puna herders (e.g., Shea 1987:81-84). Even today, pastoralists descend from the puna to Coporaque in Colca to trade dried meat, wool, textiles, and pottery for maize (Vera Cruz 1987:96). In pre-Hispanic times, besides maize, fruit, and obsidian, the Colca was possibly a source of precious metals, copper, and textile dyes derived from the insects that breed on local cacti (ibid.:115-116). Given the small amount of archaeological research conducted thus far, it should not be surprising that, except fOf the obsidian data considered here, evidence for these interzonal connections remains slight. Daniel Shea (1987) reports that his excavations of a late prehistoric site near Achoma in the Colca region uncovered storage facilities that contained camelid charqui, and he concludes that this find demonstrates a complex economy that included products acquired by trade with pastoralists. The structural links of the ancient inhabitants of the Colca region with the puna herders may have also been reinforced by the dependence of the valley agriculturalists on canal systems that use water from natural acquifers (springs, streams, lakes) situated in the puna. This is essential because the Colca Valley is so deeply entrenched that it is not feasible to raise water from the river level to the adjacent agricultural lands (Guillet 1987).

211One of the most problematic and interesting interpretive problems presented by the location of the Chivay Source is the nature of obsidian procurement and distribution during the Middle Horizon. Traditionally, the Colca Valley has been viewed as being near the frontier between the Huari and Tiahuanaco states or spheres of influence. In Rowe's early formulation, this hypothetical frontier was placed between the Majes and Sihuas Valleys (1956; cf. Lumbreras 1974: figure 162). Subsequent work by Linares Malaga, Neira, and others confirmed the presence of Huari ceramics mainly dating to MH2 in the valleys of Caraveli, Ocofia, Majes (or Camana) and Sihuas (Linares MaIaga 1990; Neira 1990). Recent surveys have provided additional evidence of the Huari presence in Ocofia (Chavez Chavez and Salas Hinojoza 1990) and Majes (Manrique Valdivia and Cornejo Zegarra 1990; Garcia Marquez and Bustamante Montoro 1990). A detailed survey of $e more southern Moquegua (or Osmore) drainage by the Programa Contisuyu yielded strong evidence for a Tiahuanaco-related occupation (Goldstein 1990). Significantly, scarce Huari-style materials in Moquegua were concentrated in the fortified Huari civic-ceremonial center. of Cerro BaUl, and investigations at the site led arch~eologists to conclude that the settlement was a short-lived intrusive political colony or outpost (Moseley et al. 1991; Goldstein 1990). The work in Moquegua reinforced the impression that the interaction along the HuariTiahuanco frontier was competitive and at times hostile (Goldstein 1990:101). Recent research by Vera Cruz and Malpass has demonstrated a strong Huari presence in the middle section of the Colca Valley. Vera Cruz argues that a Huari center was established at Achachiwa (3131 masl), located 40 km downstream from Chivay, and Huarirelated ceramics were excavated at Chijra near Coporaque, only 4 km below the modern town of Chivay. In fact, Vera Cruz and Malpass conclude that some of the terraces at Chijra and elsewhere in the Colca Valley may have been built while the Valley was under Huari control (Malpass 1987:62-64; Vera Cruz

Burger et al.: Chivay Obsidian Source 1987:89). As noted earlier, Huari cultural materials also exist in the lower coastal sections of the drainage. The foregoing review of the literature suggests that the obsidian deposit near Chivay was situated near the southern limit of Huari influence during the Middle Horizon. The procurement and distribution of raw obsidian during this period might be expected to reflect the special political realities of those troubled times. It could be suggested, for example, that obsidian from Chivay may have been distributed widely to Huari centers throughout the Central Andes, as has been observed for other types of obsidian from major deposits within the Huari sphere of influence (Burger and Asaro 1979). This, however, does not appear to be the case. With the exception of Cerro BaUl, obsidian artifacts made from Chivay source material have not been encountered thus far at Huari centers such as Huari, Jincamocco, or Pikillaqta. Although the Chivay source was the closest obsidian deposit to the Huari center at Cerro Baul, over 70 percent of the obsidian sampled (n=42) came from the more distant Alca source. Small quantities of Andahuaylas A Type obsidian were also documented at Cerro Baul; its source is probably located in the Department of Apurimac. Analysis of Cerro Baullithics also showed the presence of Quispisisa obsidian. All three of these sources lie within the Huari heartland, and artifacts from them also occur at Huari itself (Burger and Asaro 1979). In contrast, only a single sample at Cerro Baul (2% of the obsidian analyzed) came from the much closer Chivay Source. The situation is complicated further by evidence from the site of Tiahuanaco. All of the sixteen obsidian samples from Tiahuanaco's surface tested at LBNL in our original study were of Titicaca Basin Type obsidian (Burger and Asaro 1979); this means that the raw material for all of the artifacts had been procured at the Chivay Source. This finding does not appear to be consistent with the frequent assumption that the frontier between the Huari and Tiahuanaco spheres of influence

ANDEAN PAST 5 (1998) impeded the movement of materials between the two zones. At the same time, it should not be forgotten that the samples tested from Tiahuanaco were all surface materials and consequently, they could all postdate the decline of Huari. If so, they would not be relevant to discussions of the impact of the HuariTiahuanaco frontier. However, regardless of the dating of the Tiahuanaco obsidian, the patterning of obsidian procurement in the Cerro BaUlassemblage needs to be explained. Could the Chivay source have been under Tiahuanaco control despite the Huari presence in the middle and lower sections of the valley? Alternatively, could local residents of the Colca region have maintained a special relationship with the peoples of the altiplano during the Middle Horizon and consequently continued to supply these groups with raw material? A third alternative is the possibility of a decline or hiatus in obsidian procurement at the Chivay source during the crucial period of the Middle Horizon. Unfortunately, before trying to evaluate the relative merits of these (and other) alternatives, it would be necessary to have additional obsidian artifacts analyzed from contexts in southern Peru and northern Bolivia which could be situated within a unified Middle Horizon chronological framework. Whatever the patterning of obsidian procurement and distribution during the Middle Horizon, it is clear that the Chivay Source continued to provide raw obsidian to the Peruvian altiplano in later pre-Hispanic times. In this light, it is interesting to recall discussions of the ethnic configuration of the Colca Valley described in the early Colonial documents and discussed at length by Neira (1961, 1990) and Pease (1977). Basically, the historical documents refer to two distinctive and contrasting ethnic groups: the Collaguas and the Cavana. The Collaguas are described as an Aymaraspeaking group that dominated the upper portion of the Colca Valley, including Chivay and Coporaque. The Cavana are said to have been Quechua-speaking people who occupied the lower elevations in the Colca Valley including the area around Cabanaconde. Given the long-standing exchange links between the area

-212 in which the Chivay source obsidian is located and the altiplano, it is intriguing to fmd that at the time of the Spanish conquest the Chivay area was dominated by an Aymara-speaking group linked to the puna habitat. Neira and others have argued that the Collaguas represent the original pre-Inca population of the Colca Valley (Neira 1990:178), despite the general scarcity of altiplano influence visiblein the current archaeological sample from Colca (Vera Cruz 1987:121). Whatever the ultimate outcome of this and other debates, the presence of a major obsidian source near Chiyay underlines the need to understand the Colca Valley's prehistory within a regional or even pan-regional framework, as well as in terms of local carrying-capacity and other narrow concerns. References Cited
chel 1980 ExtratelTestrial Cause for the CretaceousTertiary Extinction: Exp.eriment and Theory. Science 208:1095-1108. Alvarez, W., F. Asaro, H.V. Michel, and L.W. Alvarez 1982 Iridium Anomaly Approximately Synchronous with Terminal Eocene Extinctions. Science 216:886-888. Bonavia, Duccio 1991 Peru, hombre e historia de los origenes al Siglo XV. Lima: Ediciones EDUBANCO. Brooks, Sarah, Richard Burger, Michael Glascock, and Hector Neff . 1993 Preliminary Report on Obsidian Samples from the Colca Valley, Peru. Manuscript in the possession of the authors. Brooks, Sarah Osgood, Michael Glascock, and Martin Giesso 1997 Source of Volcanic Glass for Ancient Andean Tools. Nature 386:449-450. Browman, David 1974 Pastoral Nomadism in the Andes. Current Anthropology 15(2):188-196. 1975 Trade Patterns in the Central Highlands of Peru in the First Millennium B.C. World Archaeology 6(3):322-329. Burger, Richard and Frank Asaro 1977 Trace Element Analysis of Obsidian Artifacts from the Andes: New Perspectives on Pre-Hispanic Interaction in Peru and Bolivia. Berkeley: Lawrence Berkeley Laboratory Report LBL-6343.; 80 pp. 1979 Amilisis de rasgos significativos en la obsidiana de los Andes Centrales. Revista del Museo NacionaI43:281-326. 1993 La distribuci6n y procedencia de artefactos de obsidiana durante el Periodo Inicial y Horizonte
.

Alvarez, L.W., W.A. Alvarez, F. Asaro, and H.V. Mi-

213Temprano. In Emergencia de la civilizacion en los Andes: ensayos de interpretacion, by Richard L. Burger, pp. 180-231. Lima: Universidad Nacional Mayor de San Marcos. Burger, Richard L., Sergio Chavez, and Karen MohrChavez n.d. Through the Glass Darkly: Analysis of Obsidian from the Preceramic through Early Intermediate Period in Southern Peru and Northern Bolivia. Manuscript in the possession of the authors. Chavez Chavez, Jose A. and Ruth Salas Hinojoza 1990 Catastro arqueol6gico de la cuenca del Rio Ocofia. Gaceta Arqueo16gica Andina 18/19:15-20. Lima: INDEA. Denevan, William 1987 Terrace Abandonment in the Colca Valley, Peru. In Pre-Hispanic Agricultural Fields in the Andean Region, edited by W. Denevan, K. Mathewson, and G. Knapp, pp. 1-43. Proceedings 45 International Congress of Americanists, Bogota (1985). BAR International Series 359(i). Flores Ochoa, Jorge 1968 Los pastores de Paratia: una introduccion a su estudio. Serie Antropologia Social 10. Mexico: Instituto Indigenista Interamericano. Garcia Marquez, Manuel and Rosa Bustamante Montoro 1990. Arqueologia del Valle de Majes. Gaceta Arqueo16gicaAndina 18/19:25-40. Lima: INDEA. Goldstein, Paul . 1990 La ocupaci6n TiwanakU en Moquegua. . Gaceta Arqueol6gica Andina 18/19:75-104. Lima: INDEA. Guillet, David 1987 Contemporary Agricultural Terracing in Lari, Colca Valley, Peru: Implications for Theories of Terrace Abandonment and Resource Programs of Terrace Restoration. In Pre-Hispanic Agricultural Fields in the Andean Region, edited by W. Denevan, K. Mathewson, and G. Knapp, pp. 193-206. Proceedings 45 International Congress of Americanists, Bogota (1985). BAR International Series 359(i). Johnson, George 1930 Peru from the Air. American Geographic Society, Special Publication No. 12. Linares Malaga, Eloy 1981 Proyecto Especial Majes: evaluaci6n de recursos hist6ricos-arqueol6gicos de los monumentos del Proyecto y zonas de influencia. Arequipa: Ordearequipa. 1990 Prehistoria de Arequipa. Arequipa: CONCYTEC, Universidad Nacional de San Agustin. Lumbreras, Luis G. 1974 Peoples and Cultures of Ancient Peru. Translated by Betty J. Meggers. Washington, DC: Smithsonian Institution Press. Malpass, Michael 1987 Prehistoric Agricultural Terracing at Chijra in the Colca Valley, Peru: Preliminary Report II. In Pre-Hispanic Agricultural Fields in the Andean Region, edited by W. Denevan, K. Mathewson, and

Burger et af.: Chivay Obsidian Source


G. Knapp, pp. 45-66. Proceedings 45 International Congress of Americanists, Bogota (1985). BAR International Series 359(i). Malpass, Michael and Pablo de la Vera Cruz Chavez 1990 Cronologia y secuencia de la ceramica de Chijra, Valle de Colca. Gaceta Arqueo16gica Andina 18/19:41-57. Lima: INDEA. Manrique Valdivia, Julio and Manuel Cornejo Zegarra 1990 Visi6n sobre la arqueologia del Valle de Camana. Gaceta Arqueologica Andina 18/19:2124. Lima: INDEA. Miller, George R. 1979 An Introduction to the Ethnoarchaeology of Andean Camelids. Ph.D. .dissertation, University of California, Berkeley. Ann Arbor: University Microfilms. Miller, George R. and Richard L. Burger . 1995 Our Father the Cayman, Our Dinner the Llama: Animal Utilization at Chavin de Huantar, Peru. American Antiquity 60(3):421-458. Moseley, Michael E. 1992 The Incas and Their Ancestors. London: Thames and Hudson. Moseley, Michael E., Robert Feldman, Paul Goldstein, and Luis Watanabe 1991 Colonies and Conquest: Tiahuanaco and Huari .in Moquegua. In Huari Administrative Structure: Prehistoric Monumental Architecture and State Government, edited by William H. Isbell and Gordon F. McEwan, pp. 93-119. Washington, DC:

DumbartonOaks.

Neira, Maximo 1961 Los Collaguas. Unpublished doctoral thesis. Arequipa: Universidad Nacional de San Agustin. 1968 Un nuevo complejo litico y pinturas rupestres en la Gruta Su-3 de Sumbay. Revista de la Facultad de Letras 5:43-75. Arequipa: Cooperativa Editorial Universitaria. 1990 Arequipa prehispanica. In Historia general de Arequipa, edited by M.' Neira, G. Galdos, A. Malaga, E. Quiroz, and J. Carpio, pp. 5-184. Arequipa: Fundaci6n M.J. Bustamante De La Fuente. Palacios, 0., J. de la Cruz, N. de la Cruz, B.A. Klinck, R.A. Allison, and M.P. Hawkins 1993 Geologia de la Cordillera Occidental y Altiplano al oeste del Lago Titicaca-sur del Peril. Boletin No. 42. Lima: INGEMMET, Republica del Peru, Sector Energia y Metalurgico. Pease, Franklin 1977 Collaguas: una etnia del siglo XVI. Problemas iniciales. In Collaguas 1, edited by Franklin Pease, pp. 131-167. Lima: Pontificia Universidad Cat6lica Fondo Editorial. Perlman, I. and Frank Asaro 1969 Pottery Analysis by Neutron Activation. Archaeometry 11:21-52. 1971 Pottery Analysis by Neutron Activation. In Science and Archaeology, edited by R.H. Brill, pp. 182-195. Cambridge, MA: MIT Press. Richardson, James B. III 1994 People of the Andes. Montreal and Washington, D.C.: Smithsonian Institution Press.

ANDEAN PAST 5 (1998)


Rowe, John H. 1956 Archaeological Explorations in Southern Peru: 1954-1955. American Antiquity 22:120-137. Shea, Daniel 1987 Preliminary Discussion of Prehistoric Settlement and Terracing at Achoma in the Colca Valley, Peru. In Pre-Hispanic Agricultural Fields in the Andean Region, edited by W. Denevan, K. Mathewson, and G. Knapp, pp. 67-88. Proceedings 45 International Congress of Americanists, Bogota (1985). BAR International Series 359(i). Shippee, Robert 1932 Lost Valleys of Peru: Results of the ShippeeJohnson Peruvian Expedition. Geographical Review 22:562-581. 1934 A Forgotten Valley of Peru. National Geographic Magazine 65(1):110-132. Vera Cruz Chavez, Pablo de la 1987 Cambio en los patrones de asentamiento y e1 uso. y abandono de los andenes en Cabanaconde, Valle del Colca, Peru. In Pre-Hispanic Agricultural Fields in the Andean Region, edited by W. Denevan, K. Mathewson, and G. Knapp, pp. 89128. Proceedings 45 International Congress of Americanists, Bogota (1985). BAR International Series 359(i).

-214

215-

Burger et af.: Chivay Obsidian Source

Table 1. Effective background abundances from contamination in the LWAICS and the 99.9999%-pure Al encapsulating foil for 100 mg sample weight.a
Means and RMSD for 5 empty Al capsules (assume 100 mg)C 0.2079:1: .0020 0.0061:1: .0028 0.0017:1: .0022 0.0012:1: .0003 0.0001 :I:.0003 0.011 :I:.010 Means and RMSD for 5 empty Al capsules (true weight)C 0.0812 :I:.0003 0.0024 :I:.0011 0.0007 :I:.0009 0.00045 :I:.00010 0.0000 :1:.0001 0.004 :1:.004

Isotope and/or element 46Sc 233Pa(Th) 134Cs 59Fe(%) 182Ta 141Ced 181 Hf 152Eu 86Rb 60Co 65Zn 124Sb 75Se(Ppb)

Energy of y rays (keV)b 889.25 312.01 795.87 1099.25 67.75 145.44 132.9-482.0 344.29-778.92 1076.69 1173.2-1332.5 1115.52 1690.98 136.00-264.66

LWAICS bkg 0.0017:1:.0000 0.0011:1:.0001 0.0044:1:.0001

0.0005:I: .0000
0.0002 :I:.0000 0.0040:1:.0006 0.0003:I: .0000 0.004 :I:.000

0.0026 :I: .0006


0.0004:1: .0004 0.10 :I: .14 0.014 :I:.006 3.5 :1:.6

0.0010

:I:

.00025

0.007

:I: .009

0.0362 :I:.0005

0.067 :I: .002 0.0003:I: .0002


1.75 :1:.08 100

0.00015:1: .00014 0.038 :1:.055 0.0056 :I:.0023 1.38 :1:.23

0.043
6.1

:I: .005

0.0167
2.4

:I:

.0018

:1:2.2 100

:I:1.6

Weight for abund. calc. (mg) True weight (mg)

255.7:1: 2.2

aAbundances are given in parts-per-million (ppm) except for those of Fe, which are given in percent (%), and those ofSe, which are given in parts-per-billion (Ppb). LWAICS indicates measurements were made with the Luis W. Alvarez Iridium Coincidence Spectrometer. bWhere two gamma ray energies separated by a hyphen are given for an isotope, radiations of those energies were used in coincidence measurements. cRMSD is the root-mean-square deviation. dCe abundances have not been corrected for 235U fission.

Table 2. Comparison of diagnostic element abundances measured in samples from the Chivay obsidian source in Perna.
Main Chivay source group Elementb Chivay-l Saylutta S16-104 Chivay-2 Saylutta S16-105 Chivay-4 S16-266 Chivay-5 S16-267 Chivay-7 S16-269 3.365 25.10 10.21 0.509 1.682 43.34 :1:.007 :I:.06 :I:.03 :1:.002 :1:.005 :I:.12

> 2: = tr.I E.
(%)e 0.06 0.10 0.11 0.23 0.12 0.14

C.of V. Mean & RMSDc for Chivay Main Group (%)d 3.372 25.14 10.26 0.508 1.681 43.61 :I:.0059 :1:.045 :1:.029 :I:.0013 :1:.0043 :I:.18 0.18 0.18 0.28 0.26 0.26 0.41

M.S.

Th Cs
.

Sc

Elements with sample counting errors of less than 0.3% (a// are singles measurements) 25.12 10.26 0.509 1.677 43.76

3.378

:I: .003 :I: .03

Ta Cef

Fe(%)

:1:.02 :1:.001 :1:.003 :I:.07

25.11 10.27 0.507 1.681 43.57

3.373:1: .003 :I: .03


:1:.02 :I:.002 :1:.003 :1:.05

25.21 10.26 0.507 1.688 43.59

3.378:1: .007 :I: .06


:1:.03 :I:.002 :I:.005 :I:.12

. 3.368 :I:.004
25.16 10.29 0.506 1.678 43.78 :1:.05 :1:.03 .:1: .001 :I:.004 :1:.09

~ ~ > 00 ~ tit -.. ..... 1..0 1..0 00 . '-"

Average (%) 0.26 Average efficiency uncertainty (singles) (%) Average geometry uncertainty (singles) (%) Overall uncertainty (%) Elements with sample counting errors greater than 0.3% 3.882 H 3.934 :I: .017 3.873 :1:.014 0.294 :1:.002 0.290 :1:.002 0.297 Eug Rb 249.3 :1:3.7 248.3 :1:3.5 250.6 0.333 :1:.007 0.320 :1:.006 0.329 Cog Zn 32.4 :1:.6 31.7 :1:.6 31.9 Sb 0.896 :1:.018 0.953 :1:.016 0.899 :1:.027 :1:.005 :I:3.8 :1:.015 :1:.6 :1:.029 3.878 0.295 247.7 0.324 33.1 0.900 :1:.020 :1:.003 :I:3.9 :I:.010 :1:.6 :1:.022 3.860 0.286 247.4 0.336 32.2 0.920 :1:.027 :1:.005 :I:4.1 :1:.015 :1:.6 :1:.030 3.885 0.292 248.7 0.328 32.3 0.914 :1:.028 :1:.004 :I:1.3 :1:.007 :I:.5 :1:.024 0.7 1.5 0.5 2.0 1.7 2.6 1.5

0.13 0.11 0.08 0.19 0.5 1.1 0.7 2.5 1.8 2.1 1.5

Average (%)

aAbundances are given in parts-per-million (ppm) except for those of Fe, which are given in percent (%). Errors for individual samples are usually the counting errors due to both the samples and the standards. The exceptions are the Zn values which include the uncertainties in the variable amount ofZn impurity in the AI capsules. bStandard Pottery (Perlman and Asaro, 1969) was used to calibrate the abundances of all of the listed elements, except Zn. The abundances in the standard were taken from Perlman and Asaro, 1971. The Rb abundance in Standard Pottery, however, has since been revised to 64.5:1: 1.2 ppm. Zinc was calibrated versus the standard DINO-l (Alvarez et al. 198.) and the abundance (in the Danish Cretaceous-Tertiary boundary sample) was reported in Alvarez et al. 1980. The accuracies of the measurements, which are useful for comparing with the work of other laboratories calibrated versus different standards, must incorporate both the precisions of the measurements and the uncertainties in the standards. CRMSD is the root-mean-square deviation. d C. of V. is the coefficient of variation, i.e. the root-mean-square deviation divided by the mean value. eM. S. E. is the mean sample counting error (excluding the counting errors due to the standards). f Ce abundances have not been corrected for 235U fission. 9 Best measurements are by coincidence techniques.

I tV ..... 0\

N
I )

Table 3. Deviation of element abundances of Chivay-6 and Chivay-3 from those of the main Chivay source

groupa
Main Chivay Group Mean & (RMSDCor M.S.E.e)

Elementb Elements Sc Th Cs Fe(%) Ta Cef

Chivay-6 3 counts ofS16-268 C.ofV. Mean & RMSDe (%)d

M.S.E. (%)e 0.06 0.10 0.11 0.24 0.11 0:16 0.13% 0.15%

(Chivay-6 -Main Group) / Main Group (%) 0.39 0.24 0.44 0.49 0.59 0.37 :i:.14 :i:.15 :i:.20 :t:.18 :t:.15 :i:.21

Chivay-3 Saylutta S16-106 3.349:i: .003 25.14 :i:.04 10.07 :i:.03 0.528 :i:.002 1.657 :i:.004 43.54 :i:.l0

(Chivay-3 -Main Group) / Main Group (%)


-0.7 0.0 -1.9 3.9 -1.4 -0.2 :i:.1 :i:.2 :i:.3 :i:.4 :i:.3 :i:.3

with sample counting errors of less than 0.3% 3.372:1:.006 3.385 :i:.007 0.19 25.14 :i:.05 25.20:i: .05 0.20 10.26 :i:.03 10.305:i: .031 0.30 0.508:i: .001 0.5105 :i:.0003 0.06 1.681 :i:.004 1.691:i: .003 0.19 43.61 :i:.18 43.77 :i:.07 0.17 Mean Effic. and geo. uncertainties Best value 0.19%

0.42 :i:.07 % 0.6 1.1 0.9 2.5 0.5 2.3 1.3% -0.10 1:0 -0.1 -0.6 1.9 -5.9 :i:0.5 :i:1.0 :i:0.6 :i: 1.7 :i: 1.1 :i: 1.8 4.025 :i:.031 0.300:i: .005 244.2 :i:3.9 0.382:i: .010 32.5 :i:.6 0.920:i: .033 3.6:t: .9 2.7:i: 1.8 -1.8 :i: 1.6 16.5 :i:3.2 0.6 :i:2.0 0.7:i: 3.8
b::I \::
(\) "'t

Elements with sample counting errors greater than 0.3%' H~ 3.885:i: .028 3.881:i: .025 0.6 Eug 0.292 :i:.004 0.295:i: .004 1.4 Rb 248.7:i: 1.3 248.4 :i:2.1 0.8 Cog 0.328:i: .007 0.326:t: .005 1.5 Zn 32.3:i:.5 32.9:i: .5. 1.5 Sb 0.914:i: .024 0.860:i: .023 2.7 Mean 1.4%

~
~ ~

.....

:-

9
~.

a,b,c,d,e,f,gSeefootnotes in Table 2. Exceptions are: 1) errors for the Chivay-6 and Chivay-3 deviations from the main group presume without proof that the precision of the group averages can be expressed by the RMSD divided by the square root of the number of measurements; 2) "RMSD or M.S.E." is the larger of the root-mean-square deviation for the 5 samples or their mean counting error (excluding the counting errors due to the standards); 3) Effic. and geo. uncertainties = uncertainties in the singles measurements of gamma ray efficiency and geometry, respectively.

~ ;

~ (\)

ANDEAN PAST 5 (1998) Table 4. Comparison of element abundances measured in samples from the Chivay source in Peru with the Titicaca Basin Chemical Group of Burger and Asaro, 1977a
Main Chivay Source Group (Mean & RMSDc for 5 samples) Titicaca Basin Chemical Group (Mean & RMSDc for 21 samples)

-218

EImentb

Deviation (%) -0.5 -0.2 -0.5 -2.0 -2.1 -1.9 0.0 -1.7

Elements with RMSD better than 4% in the old LBNL INAA 3.355 :i: .047 Sc 3.372 :i: .006 Th 25.14:i: .05 25.10:i: .34 Cs 1O.26:i: .03 10.21:1: .37 Fe(%) 0.508:i: .001 0.498 :i:.017 Eu 0.292 :1:.004 0.286 :1:.007 Cef 43.61:1: .18 42.8:1: 1.1 Ta 1.681:1: .004 1.681 :1:.031 Hf 3.885:1: .028 3.82:i: .12

Average deviation (8) = 1.1% Elements with RMSD poorer than 4% in the old LBNL INAA Co 0.328:i: .007 0.31.:i: .08 Rb 248.7:1: 1.3 240:i: 13h

a,b,c,fSee footnotes in Table 2. Deviation (%) = 100 x (Chemical groupSource group) / Source group hTh~ abundance is the best value for the recalibrated Rb abundance that could be deduced for 21 samples measured in 4 irradiation groups. The most realistic RMSD comes from 3 of the groups with comparable error. As the value is coincidentally small, 4 ppm, the average uncertainty of the 3 groups, 13 ppm, is used.

219-

Burger at al.: Chivay Obsidian Source

o 20 40 60 80 100
I I I I I Kilometers I

KEY obsidian source

places referred to intext

Figure 1. Location of archaeological sites and other places mentioned in this article.

ANDEAN PAST 5 (1998)

-220

o
(

1 I

2 3 I I Kifometers

I @ location of analyzed
.

source specimens

modern settlement

Figure 2. Location of obsidian samplesanalyzed in this study.

221-

Burger at aZ.: Chivay Obsidian Source

Ts-Ba

Ancachlta Pampa Ichocollo

TmTa

Ts-Ba

QA-n

2 I
Kilometers

4 111 rhyolitic dome o concentration of'obsidian . modem settlement

Figure 3. Geology of the Chivay area, including the rhyolitic dome at Cerro Homillo and the ChivaySourceobsidiandepositsassociatedwith its perimeter(basedon Palacios et af. 1993and observations by Guido Salas).

ANDEAN PAST 5 (1998)

-222

Figure 4. Rhyolitic dome at Cerro Homillo. Photograph by Guido Salas.

223-

Burger at aZ.: Chivay Obsidian Source

"

Figure 5. Obsidian layers on the western edge ofCerro Hornillo's rhyolitic dome (see Figure 4). Photograph by Guido Salas.

THE JAMPATILLA OBSIDIANSOURCE: IDENTIFYING THE GEOLOGICAL SOURCE OF PAMPAS TYPE OBSIDIANARTIFACTS FROM SOUTHERN PERU Richard L. Burger
Yale University

Katharina J. Schreiber
University of California - Santa Barbara Michael D. Glascoc~
University of Missouri

- Columbia

Jose Ccencho Huaca Pucllana, Lima

Background In 1974 and 1975, a study of Peruvian archaeological obsidian at the Lawrence Berkeley National Laboratory (LBNL) identified a distinctive chemical composition of obsidian utilized in artifacts. For ease of reference it was called Pampas Type obsidian. Pampas Type obsidian was one of eleven common kinds of obsidian utilized in pre-Hispanic times in what now is Peru.l Pampas Type obsidian was documented as common only in the samples from sites iJ].the Carhuarazo Valley of southern Ayacucho (Figure 1).2 These obsidian artifacts had been collected by William Isbell, .Patricia Knobloch, and Katharina
lIn the original LBNL study, eight types of obsidian were identified as having been exploited with some frequency, and another eight distinctive chemical compositions were documented as existing among a small group of rare artifacts. Subsequent research on the obsidian of Cuzco by Burger in collaboration with Sergio Chavez and Karen Mohr Chavez (Burger et al. n.d.) indicates that at least three of these "rare types" were frequently used in the southern highlands. Additional obsidian sources are known to exist further south in Bolivia and Chile. 2Although we refer to the valley here as the Carhuarazo Valley, this is not a name that is used locally. Today, the river that flows north through this valley has a series of names, including Mayobamba, Sondondo, Jatun Mayo, Yanamachay, and Pampamarca, derived often from the name of whatever local village it passes at a particular point. Older maps of the valley indicate the name Carhuarazo for the river; hence this name is used to refer to the valley (Schreiber 1992:132).

Schreiber during a project of archaeological explorations (Burger and Asaro 1977:21-31, 1979:310-311). In the LBNL study, obsidian of the Pampas Type was found to be present, but less frequent, in some regions adjacent to the Carhuarazo Valley. The diagnostic trace element composition of Pampas Type obsidian was determined at LBNL by the analysis. of archaeoldgical obsidian by instrumental neutron activation analysis (n=7) and X-ray fluorescence (n=36). These results suggested a single distinctive geological source of obsidian that was relatively homogenous in composition and was easily distinguishable on chemical grounds from the other major and minor obsidian types (Burger and Asaro 1979:' tables 1, 2B). The scarcity or absence of Pampas Type obsidian outside the Carhuarazo region suggested that the geological source was probably located somewhere in this area. Subsequently, Schreiber encountered small unworked obsidian nodules near Huaycahuacho; these materials came from a secondary deposition, apparently eroded from the primary geological source of the obsidian (Burger and Asaro 1977:310). Samples from Huaycahuacho were analyzed at LBNL and were found to match the composition of the Pampas Type obsidian (Asaro and Burger unpublished data). In 1981, Schreiber located the primary deposit of geological obsidian at Jampatilla during an archaeological survey near Huaycahuacho, and she obtained geological obsidian samples from three loci within this source

ANDEAN PAST 5 (1998):225-239.

ANDEAN PAST 5 (1998) (Figure 2). Obsidian workshop areas and associated sites were also encountered. In 1991, ,archaeologist Jose Ccencho, a resident of Huaycahuacho, returned to the obsidian source at Jampatilla and made more detailed observations of the deposits and workshop areas. Two years later, Michael Glascock and Hector Neff of the Missouri University Research Reactor (MURR) began to collaborate with Burger on additional studies of Central Andean obsidian, and six samples from the Jampatilla source were analyzed by instrumental neutron activation at MURR. The results of these analyses definitively confirmed that Jampatilla was the source of the Pampas Type obsidian. This article presents a description of the Jampatilla geological source, summarizes the analytical methods and results of the trace element analyses at MURR, compares them with the earlier LBNL data, and then briefly explores some of the ,archaeological implications of the identification of the Jampatilla source in light of the existing provenience studies of obsidian !lf1:ifacts. The Jampatilla Obsidian Source: Location and Geology The Jampatilla source is located 3-4 km to the east-northeast of the town of Huaycahuacho, Lucanas Province, Department of Ayacucho (Figure 2).3 It is found about 55 km north in a straight line from the town of Puquio and about 125 km south of the city of Ayacucho (Figure 1). The local geology is characterized by Tertiary-Quaternary basalt flows at elevations exceeding 4000 meters above sea level (masl), producing broad tablelands that surround the valley to the' east, west, and south (INGEMMET 1975). Where exposed, the basalt has a columnar aspect, indicating slow
3Jampatilla Source Location: Latitude-Longitude, generallocation: 1413'S,735'W UTM coordinates of recorded deposits: north side ridge deposit: E 615500 to E 616100, N8428900; south side ridge deposit: E 616200, N 8428750 hillside deposit: E 616800, N 8428700 Toqsa deposit: E 615700, N 8426200.

-226

cooling. Steep-walled valleys were eroded into this tableland by fluvial activity, exposing the underlying deposits of Tertiary tuffs between elevations of 2950 and 4000 masl. These tuff deposits vary in density and produce local topography characterized by gentle slopes, nearly level shelves, and very steep cliff faces. Below 3600 masl, soils are moderately well developed in gently sloping and level areas, and support local agriculture.. Below elevations of 1950 masl, fluvial action has exposed Cretaceous-Tertiary sandstone formations. Soils are poorly developed in these areas. The east side of the valley, as viewed from Huaycahuacho, is a roughly semicircular basin flanked to the north and south by the Jampatilla and Toqsa ridges, respectively. The upper edge of the basin is formed by the exposed basalt cliffs that define the upper perimeter of the valley. Several outcrops of obsidian were recorded in July and August of 1981 during the course of an archaeological survey (Schreiber 1982, 1992:115-163). The largest outcrop was located along the north side of the Jampatilla ridge, where obsidian nodules up to 8 cm in diameter were observed in a deposit of loosely consolidated tuff at 3500 masl. A small area at the same elevation on the south side of the ridge was also recorded. The deposit may actually be continuous around the west end of the ridge, but the density of chipped obsidian debitage made it impossible to determine whether there was a natural deposit in this area as well. A second outcrop was observed to the east and very near to the first, but separated from it by a steep-sided ravine, on the flanks of the basin. Here the deposit also comprised nodules within loosely consolidated tuff, but the nodules were somewhat larger than those of the ridge deposit, reaching 10-12 cm in diameter. The elevation of this deposit is some 25-50 m higher than the first. A third deposit was observed on the north face of the Toqsa ridge, in very loosely consolidated tuff. This deposit, at 3450 masl, is somewhat lower than the larger Jampatilla deposits. Here the nodules were much smaller,

227reaching only 4-5 cm, and in lower concentrations than elsewhere. This stratum may extend farther to the northeast toward the second outcrop, but the extreme steepness of the topography prevented further observation. Very small nodules (1-2 cm in size) are seen throughout the Huaycahuacho area, eroding down from the source deposits. On the basin slopes where soils are moderately well developed, obsidian is observed in low frequencies in agricultural fields, but in higher frequencies on small hilltops and in exposed erosion cuts and ravines. Below 2950 masl, where soils are poorly developed, the density of obsidian nodules on the surface is high. Although no systematic study has been conducted of the physical characteristics of Jampatilla Source obsidian, most of the material examined was of good purity and flaking quality. It varied in clarity from transparent colorless to black, and the coloration was uneven with black patches or ~treaks contrasting with colorless clear areas. Archaeological Sites Associated with the Source Archaeological sites associated with the Jampatilla Source were recorded both during the 1981 survey, and by a second survey undertaken by Jose Ccencho (Ccencho Huamani 1991). Numerous workshop areas are found in the vicinity of the source, and some of these are quite large (Figure 2). The entire length of the summit of the Jampatilla ridge is littered with obsidian debitage in greater and lesser densities. In one area of high density it was estimated that there were in excess of 200 pieces of obsidian debitage per square meter. Another workshop area extends from the northwest end of the ridge and down the slope; it covers roughly 25 ha and is densely covered with lithic debris. Other workshop areas were recorded below the Jampatilla ridge to the west. Although finished tools were relatively rare at any of the workshops, the few projectile points observed range in date from the Preceramic Period through the Middle Horizon. The only major archaeological site, other than workshops, located adjacent to the Jam-

Burger et al.: Jampati/la Obsidian Source

patilla ridge is Willkaya (Ay5-46), which was a small Huari imperial facility (Schreiber 1992: 155-157). Neutron Analyses at MURR Six source specimens were submitted to MURR from the Jampatilla Obsidian Source. This sample included obsidian from the three outcrops identified by Schreiber. Sample Preparation and Analysis Analytical samples were prepared at MURR from the Jampatilla source specimens by using a ceramic mortar and pestle to break off small fragments weighing 50-100 mg. Samples were prepared for short irradiation by weighing 100 mg of material into small polyvials. Standards of the SRM-278 Obsidian Rock and SRM-1633a Fly Ash reference materials were similarly prepared. Samples for long irradifltion were prepared by weighing 250-300 mg of fragments into highpurity quartz vials which were sealed in an oxygen torch flame. For the smaller samples, the short irradiation sample was transferred from the polyvial into a quartz vial. Further details concerning the procedures for sample preparation and standardization are available in Glascock et al. (1988). . MURRis pneumatic-tube irradiation system was used for short irradiation of all samples and standards. Samples were sequentially activated for five seconds in a thennal neutron flux of 8 x 1013n cm-2S-I. Following activation, the samples were allowed to decay for 25 minutes before measurement on a highresolution gennanium detector for 12 minutes. By comparing the unknowns with the reference standards, the concentrations of five short-lived elements were detennined: CI, Dy, K, Mn, and Na. The quartz vials containing the samples and standards for long irradiation were wrapped in aluminum foil and irradiated for 70 hours in a thennal neutron flux of 5 x 1013 n cm-2S-I. After irradiation, the samples were allowed to decay for 9-10 days before con-

ANDEAN PAST 5 (1998) ducting a measurement for 2,000 seconds each to determine seven medium-lived elements: Ba, La, Lu, Nd, Sm, U, and Yb. After an additional decay of 4-5 weeks, the samples were measured for 10,000 seconds each to determine the following fifteen long-lived elements: Ce, Co, Cs, Eu, Fe, Hf, Rb, Sb, Sc, Sr, Ta, Th, Th, Zn, and Zr. Results and Interpretation Results for the six Jampatilla source samples are presented in Table 1. Earlier data from the LBNL for seven Pampas Type artifacts reported by Burger and Asaro (1977) are presented for comparison in Table 2. The LBNL data were normalized to the MURR standard through unpublished analyses of Perlman's Pottery Standard at MURR. Agreement between the best elements is quite good and strongly suggests that the artifacts of Pampas Type obsidian artifacts came from the Jampatilla source. This is further supported by two accompanying bivari~te plots for Cs vs Hf and Ba vs Eu (Figures 3 and 4). The plots illustrate the compositional similarity of the Jampatilla source specimens and Pampas Type artifacts to one another and their differentiation from other significant obsidian types reported in the Berkeley study. Ninety-five percent confidence ellipses are shown surrounding each of the obsidian types. Archaeological Ramifications Confirmation that the Jampatilla Obsidian Source provided the raw material for artifacts of Pampas Type obsidian permits a more precise discussion of pre-Hispanic obsidian procurement and exchange in the south central highlands of Peru. The Jampatilla Obsidian Source appears to have been exploited throughout the prehistory of the Carhuarazo Valley (Table 3). As noted above, Schreiber identified lithics dating to the Preceramic Period at workshops near the obsidian outcrops. Although numerous Formative (i.e., Initial Period/Early Horizon) sites have been located in the Carhuarazo Valley, and these are characterized by an abundance of obsidian artifacts, no samples have yet been analyzed. Obsidian from the Jampatilla Obsidian Source

-228 has been recovered at the Initial Period/Early Horizon sites of Waywaka in Andahuaylas and Hacha in Acari. Subsequent use of Jampatilla obsidian during the Early Intermediate Period and Middle Horizon is suggested by the artifacts analyzed from Corralpata (Ay5-5) and Jincamocco (Ay5-6). The exploitation of the Jampatilla Source during the subsequent Late Intermediate Period and Late Horizon is' implied by the results of our analyses from Willka Cahuana (Ay5-7) and Canichi (Ay5-1). In the Central Andes, this pattern of prolonged exploitation of an obsidian source throughout the prehistoric sequence is by no means limited to Jampatilla. On the contrary, it is apparently common and has been well documented for Quispisisa obsidian (Burger and Asaro 1979). Metal tools never entirely replaced stone artifacts in pre-Hispanic times and obsidian, the sharpest and most easily flaked of lithic raw materials, continued to be popular until the Spanish Conquest. This was especially true in areas ~ith nearby geological sources of obsidian. The documentation of a small Huari outpost immediately adjacent to obsidian workshop areas and less than 1 km from the obsidian outcrops at Jampatilla suggests that the local representatives of the Huari state may have been interested in controlling exploitation of the obsidian source. Additional evidence that the Jampatilla Obsidian Source was a resource for the Huari state is reflected in the numerous obsidian artifacts of Jampatilla Source obsidian that have been recovered 10 km away at the local Huari administrative center of Jincamocco (Burger and Asaro 1979; Schreiber 1992:246). Our analysis at LBNL of 51 obsidian artifacts from Jincamocco indicated that 45% of them were made of obsidian from Jampatilla (Table 3). While it might be tempting to see the Jampatilla Obsidian Source as one of the reasons for establishing the Huari administrative center in the Carhuarazo Valley, the existing evidence from Huari centers in other regions seems to indicate that Jampatilla obsidian was rarely exported during the Middle Horizon. Provenience studies of obsidian from the ad-

229ministrative centers at Huari (n=55), Pikillacta (n=8) and Cerro Baul (n=29) have all failed to encounter examples of Jampatilla obsidian (Burger and Asaro 1979, unpublished data). The case of Cerro Balli in the Osmore drainage is particularly interesting because obsidian artifacts made from Quispisisa obsidian were documented at the site, while obsidian from the closer source at Jampatilla was entirely absent from our sample of artifacts. One of the unexpected features of preHispanic obsidian procurement in the Carhuarazo Valley is the failure of the Jampatilla Source obsidian to dominate the assembhlges of. archaeological sites near the geological source. In the LBNL study (Burger and Asaro 1979), samples from four sites in the Carhuarazo Valley were analyzed. In three of these four sites, (Ay5-5, Ay5-6, and Ay5-7) obsidian from the Quispisisa source in Huancavelica outnumbers artifacts from the local obsidian source in our sample. While the Jampatilla source was 'ocated 10 km or less from these sites, acquisition of Quispisisa obsidian required a lengthy journey. Our sample and selection procedures from most sites is too small and unsystematic to permit anything approaching statistical significance, but the rough consistency between the sites is suggestive. Moreover, analyses of 51 artifacts from the Middle Horizon center of Jincamocco (Ay5-6) determined that the raw material for 47% of the artifacts were of Quispisisa obsidian, compared to only 45% from the more local source at Jampatilla. Ay5-1 and Ay5-7 are habitation sites that postdate the Middle Horizon, thus, the popularity of imported Quispisisa obsidian continued into late prehistoric times and was not merely a product of Huari domination. In fact, of the obsidian samples analyzed from these two late prehistoric villages, three samples (37.5%) were from the Quispisisa source. While the sample is small, it does not indicate a sharp drop in the procurement of obsidian from Huancavelica, despite the continued exploitation of the local source. How can this patterning be explained? The types of volcanic glass from Jampatilla and Quispisisa are visually similar. Although some

Burger et al.: Jampatilla Obsidian Source samples from Quispisisa do show some distinctive qualities, such as red spots from iron impurities, most obsidian from the two sources is indistinguishable without laboratory analysis. In general, the visual properties of obsidian do not seem to have been an important consideration in the selection of raw material for what were usually crude utilitarian lithic flakes and tools. Thus, appearance does not seem to have been the reason that exotic obsidian from Quispisisa and elsewhere was sought with such great frequency. An alternative possibility is that the geological character of Jampatilla Obsidian Source made its raw material inferior to that of Quispisisa from a technological perspective for some purposes, and as a consequence of this, obsidian from distant sources at Quispisisa was imported despite easy access to raw material from the local source. Exploration of this hypothesis would require more detailed studies of the geological deposits and workshop-s at Jampatilla and other sources. Nevertheless, it is worth noting that the investigations at the Jampatilla outcrops have only encountered obsidian as relatively small nodules. Of the three primary outcrops located, the largest nodules of obsidian encountered measured only 10-12 cm and in some portions of the primary deposit obsidian nodules measured only 4-5 cm. This pattern contrasts with that of some other obsidian sources, where massive blocks of high quality obsidian can be observed. A 10-12 cm obsidian nodule, while large enough to produce a small projectile point or utilizable flakes, presents inherent technological limitations, particularly when compared to obsidian found in larger form. While information available is insufficient for any conclusive statements, further research into this question is warranted. Schreiber has argued that in Middle Horizon times, a major road that passed alQngside of the Jampatilla obsidian deposit connected the site of Huari in the highlands of Ayacucho with Nazca on the southern Peruvian coast via the Carhuarazo Valley (Schreiber 1984:274). Given this long-standing pattern of movement, it might be presumed that obsidian from the

ANDEAN PAST 5 (1998) Jampatilla source would be found with some frequency at sites in these somewhat distant areas. Despite this expectation, no obsidian from the Jampatilla source has been documented from the eleven sites in Ayacucho (including Huari), although a substantial number (n=137) of obsidian artifacts has been analyzed from sites of different ages in this region (Burger 1981; Burger and Asaro 1978, 1979). Similarly, Jampatilla source obsidian was not encountered in the sample tested (n=52) at LBNL from two sites (San Nicolas and Poroma) in the Nazca drainage (Burger and Asaro 1979:table. 3). The highland trunk of a second pre-Hispanic road passing through the Carhuarazo Valley led northeast to the Cuzco Valley (Schreiber 1987:276, figure 5); this early road was used later by the Incas. Over three hundred obsidian artifacts from sites of different ages have been analyzed from Cuzco, but only a single artifact proved to be made of material from the Jampatilla soUrce (Burger and Asaro 1979:311). Unfortunately, it comes from the site of Wimpilla, a poorly known multi-component site, and the explanation of this anomalous surface find remains elusive. Thus, the existing_evidence suggests that the Jampatilla Source was exploited largely by and for the local populations of the Carhuar~o Valley. Building upon our discussion of the popularity of imported Quispisisa obsidian in the Carhuarazo Valley, it can be suggested that the reasons for the limited distribution of Jampatilla obsidian in distant areas may have more to do with the geology of the Jampatilla obsidian deposit than it does with the marginalization or isolation of the Carhuarazo Valley from the rest of southern Peru in pre-Hispanic times. Nevertheless, obsidian from the Jampatilla Obsidian Source was occasionally utilized at sites outside the Carhuarazo Valley. One of the areas for which this has been documented is the lower Acari Valley, some 150 air kilometers from and 3300 meters in altitude below the obsidian source at Jampatilla. Another area that utilized obsidian from the Jampatilla Source is Andahuaylas, some 86 Ian away in the highlands to the northeast.

-230 The evidence for the Acari Valley comes from the site of Hacha. Hacha appears to have been a small agricultural village occupied during the end of the Initial Period and the beginning of the Early Horizon. Judging from the most recent studies by Riddell and Valdez (1988) and Robinson (1994), the obsidian points that were recovered on the surface by Rowe, Menzel, and others (Rowe 1967) most. likely date to the final episode in the site's occupational history (known as Haldas 2), probably during the Early Horizon. A radiocarbon measurement of 2590j:200BP (UCR2207) from a human burial associated with an obsidian point gives a rough idea of the age of the Hacha 2 component (Robinson 1994:1417, figure 18). A sample of 64 points from Hacha was analyzed at LBNL, of which 7 (11%) came from the Jampatilla Source; 40 (62.5%) came from more distant Quispisisa. In the Andahuaylas area, excavations and surveys by Joel Grossman provided an excellent sample (n=84) from the Waywaka site and a small sample (n=2) from the nearby Kunka Taka site (Burger and Asaro 1979:table 5; Grossman 1985). At Waywaka, about 20% of the obsidian artifacts analyzed were made of obsidian brought from the Jampatilla Source, and examples of the Jampatilla obsidian were found in all of the site's phases, including those preceding and coeval with the Hacha projectile points. As noted previously (Burger and Asaro 1979:31), the presence of Jampatilla obsidian in Waywaka and Hacha suggests a pattern of cultural and economic interaction running up the Acari and Sondondo Rivers. Such a pattern of interaction would have traversed the axis of the Andes and linked ecologically disparate but complementary zones in the manner discussed by Jo1m Murra (1972) and others (e.g., Flores Ochoa 1968). The presence in small quantities of Andahuaylas A Type and Andahuaylas B Type obsidian at Hacha, collectively constituting 6% of the sample, (Burger and Asaro 1993:211, table 11) reinforces the argument that such an exchange network spanning these diverse ecological habitats existed during the beginning of the Early Horizon (if not before). The geological sources for Andahuaylas A and Andahuaylas B obsidian have yet to be 10-

231cated, but given the restricted range of artifacts made from this material, it is likely that the obsidian deposits will be found somewhere in the Department of Apurimac, perhaps in the Andahuaylas region, where these types constitute the majority of the obsidian utilized (Burger and Asaro 1979:311, tables. 4, 5). This sphere of interaction linking Andahuaylas with Acari would have functioned independently of the Chavin sphere of interaction which never reached this far south. Moreover, the continued appearance of Jampatilla obsidian in the later occupation layers at Waywaka and the late site of Kunka Taka (as well as the continued presence of Andahuaylas A Type obsidian in the Carhuarazo Valley) suggests that this pattern persisted long after the Early Horizon. The relatively restricted distribution of Jampatilla obsidian beyond the Carhuarazo Valley has parallels with many other kinds of Peruvian obsidian. These obsidian types, because of their localized distribution, might be considered as deriving from "minor" so~ces. Obsidian types of localized importance that were only r~ely carried beyond the hinterland of their respeCtive source area include the Acari Type, the Ayacucho Type, the Andahuaylas A Type, and the Andahuaylas B Type. In contrast, raw material from the Quispisisa Obsidian Source, the A1ca Obsidian Source (Cuzco Type obsidian), and the Chivay Obsidian Source (Titicaca Basin Type obsidian) all had pan-regional distributions.4 These sources can be considered of "major" importance, particularly from the perspective of long-distance exchange networks. Why some sources of obsidian, like the one at Jampatilla, remained minor, while others achieved panregional significance remains an intriguing problem. Its resolution will require consideration of the geological quality and location of the different sources, as well as the mechanisms of exchange, and the changing socio4The source of the Titicaca Basin Type obsidian has been identified in the Chivay area and the source of Cuzco Type obsidian has been discovered at Alca in the Cotahuasi area. Both of these "major" sources are located in the Department of Arequipa (Burger and Asaro 1993; Burger et al. 1998a, 1998b).

Burger et al.: Jampatilla Obsidian Source economic and political contexts in which this exchange occurred. Acknowledgments We gratefully acknowledge funding from the National Science Foundation through an Archaeometry Program grant (DBS 9102016) to MURR which provided partial support fOT this research. Additional support for this work was generously provided by Yale University. Permits for export of the obsidian artifacts analyzed were graciously provided by the In.stituto Nacional de Cultura, Lima, Peru. We also acknowledge the assistance of Ms. Sindy Hays who supervised the students who prepared the specimens for instrumental neutron activation analysis and Sandra Ruggiero who typed the manuscript. Of course, the authors are responsible for any errors of fact or interpretation found in this report. We also acknowledge NSF grant BNS80-06121 from the Anthropology Program that supported the archaeological survey in the Carhuarazo Valley. References Cited
Burger, Richard L. 1981 La procedencia de artefactos en los sitios formativos en Ayacucho: Chupas y Wichqana. Boletin del Museo Nacional de Antropologia y Arqueologia 7:9-10. Lima. Burger, Richard L. and Frank Asaro 1977 Trace Element Analysis of Obsidian Artifacts from the Andes: New Perspectives on Pre-Hispanic Interaction in Peru and Bolivia. Berkeley: Lawrence Berkeley Laboratory Report LBL-6343. 80 pp. 1978 The Distribution and Provenience of Preceramic Obsidian Artifacts from the Central Highlands and Coast of Peru. Contributions of the University of California Archaeological Research Facility 36:61-83. Berkeley. 1979 Analisis de rasgos significativos en la obsidiana de los Andes Centrales. Revista del Museo Nacional 43 :281-326. 1993 La distribuci6n y procedencia de artefactos de obsidiana durante el Periodo Inicial y Horizonte Temprano. In Emergencia de la civilizacion en /os Andes: ensayos de interpretacion, by Richard L. Burger, pp. 180-231. Lima: Universidad Nacional Mayor de San Marcos. Burger, Richard L., Frank Asaro, Paul Trawick, and Fred Stross 1998a The Alca Obsidian Source: The Origin of Raw Material for Cuzco Type Obsidian Artifacts. Andean Past 5:183-200.

ANDEAN PAST 5 (1998)


Burger, Richard L., Frank Asaro, Guido Salas, and Fred Stross 1998b The Chivay Obsidian Source and the Geological Origin of Titicaca Basin Type Obsidian Artifacts. Andean Past 5:201-222. Burger, Richard L., Sergio Chavez, and Karen MohrChavez n.d. Through the Glass Darkly: Analysis of Obsidian from the Preceramic through Early Intermediate Period in Southern Peru and Northern Bolivia. Manuscript in the possession of the authors.
.

-232
Andes, edited by Ann Kendall, pp. 75-94. B.A.R. International Series 210. 1987 Conquest and Consolidation: a Comparison of the Wari and Inca Occupations of a Peruvian Highland Valley. American Antiquity 52(2):266284. 1992 Wari Imperialism in Middle Horizon Peru. Anthropological Papers 87, Museum of Anthropology, University of Michigan, Ann Arbor.

gen Derecha del Rio Sondondo, Lucanas - Ayacucho." Final report submitted to the Instituto Nacional de Cultura, Lima. Flores Ochoa, Jorge 1968 Los pastores de Paratia: una introduccion a su estudio. Mexico: Instituto Indigenista Interamericano. Glascock, M.D., J.M. Elam, and R.H. Cobean 1988 Differentiation of obsidian sources in Mesoamerica. In Archaeometry '88, edited by R.M. Farquhar, R.G.V. Hancock, and L.A. Pavlish, pp. 245-251. Toronto: Archaeometry Laboratory, University of Toronto. Grossman, Joel 1985 Demographic Change aild Economic Tr.ansformation in the South-central Highlands of preWari Peru. Nawpa Pacha 21:45-126. INGEMMET (Instituto Geologico Minero y MetalUrgico) .. 1975 Mapa geologico del Peru. Lima: Instituto Geol6gico Minero y Metahlrgico. Murra, Jo4n 1972 El "control vertical" de un mwdmo de pisos ecol6gicos en la economia de las sociedades andinas. In Visita de la Provincia de Leon de Hucmuco en 1562. Documentos por la Historia y Etnologfa de Huaouco y Selva Central 2:427-476. Huaouco: Universidad Nacional Hermilio Valdizan. Riddell, Francis and Lidio Valdez 1988 Hacha y la ocupaci6n temprana del Valle de Acari. Gaceta Arqueol6gica Andina 16:6-10. Lima: INDEA. Robinson, Roger 1994 Recent Excavations at Hacha in the Acari Valley, Peru. Andean Past 4:9-37. Rowe, John H. 1967 An Interpretation of Radiocarbon Measurements on Archaeological Samples from Peru. In Peruvian Archaeology: Selected Readings, edited by J. Rowe and D. Menzel, pp. 16-30. Palo Alto, California: Peek Publications. Schreiber, Katharina J. 1982 Informe final: exploraci6n arqueol6gica del valle Carhuarazo, Lucanas, Ayacucho, Peru. Final report submitted to the Instituto Nacional de Cultura, Lima. 1984 Prehistoric Roads in the Carhuarazo Valley. In Current Anthropological Projects in the Central

CcenchoHuamani,JoseE. 1991 Informe de los estudios arqueol6gicos del proyecto"EIPeriodoIntermedioTardioen la Mar-

233-

Burger et al.: Jampatilla Obsidian Source

Table 1. Element concentrations


Element Ba (ppm) La (ppm) Lu (Ppm)
Nd (ppm) Sm (ppm) U (ppm) Yb (ppm) Ce (ppm) Co (ppm) Cs (ppm) Eu (ppm) Fe (%) Hf (ppm) Rb (ppm) Sb (ppm) SC (ppm) Sr (ppm) Ta (ppm) Tb (ppm)

in obsidian source specimens from the Jampatilla source.


anid RLB041 754 40.6
0.458

anid RLB039 785 40.4


0.431

anid RLB040 753 40.7


0.437

anid RLB042 778 40.7


0.435

anid RLB043 744 40.0


0.433

anid RLB044 771 40.5


0.441

I
j

I i I

27.4
5.35 8.18 2.55 75.3 0.553 12.3 0.975 0.868 4.84 152 1.742 2.33 354 1.61

27.1
5.45 8.29 2.57 76.6 0.572 12.4 1.008 0.889 4.92 153 1.708 2.38 385 1.61

29.2
5.48 8.36 2.55 76.0 5.61 12.3 0.994 0.887 4.91 153 1.712 2.35 411 1.61

27.1
5.39 8.11 3.00 75.s 0.566 12.4 1.003 0.883 5.28 155 1.749 2.36 388 1.65

27.4
5.30 8.21 2.77 75.0 0.553 12.4 0.986 0.877 4.83 153 1.706 2.34 335 1.61

27.3
5.36 7.78 2.61 76.3 0.573 12.4 1.003 0.886 4.94 155 1.698 2.36 356 1.63

i i I i

Th (ppm) Zn (ppm) Zr (ppm) Cl (ppm) Dy (ppm) K (%) Mn (ppm) Na (%).

i
! .
i

i i

i !

0.688 12.5 61.4 205 602 .4.94 4.01 641


3.58

0.688 12.6 80.7 202 659 4.55 3.63 622


3.51

0.713 12.5 80.4 194 621 4.26 3.69 639


3.55

0.682 12.5 81.4 215 614 4.92 3.94 648


3.62

0.617 .12.5 81.0 195 583 4.54 4.12 639


3.57

0.638 12.5 82.0 214 628 4.50 3.60 635


3.56

i
I

ppm

- parts

per million

ANDEANPAST 5 (1998)

-234

Table 2. Comparison between mean concentrationsand standard deviations for Jampatilla source specimensand Pampas-typeartifacts.
I

Element Ba Ce Cl Co Cs Dy Eu Fe Hf K La Lu Mn Na Nd Rb Sb Sc Sm Sr Ta Tb Th U Yb (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (%) (ppm) (%) (ppm) (ppm) (ppm) (%) (ppm)
(ppm)

i
i

Jampatilla source specimens analyzed at MURR (n = 6) !

i
.

Pampas-type
analyzed at Berkeley

artifacts
(n

= 7)'

764

75.8
618 0.56 12.4 4.62 0.995 0.882 4.95 3.83 40.5 0.44 637 3.57 27.6 153 1.72 2.35 5.39 371 1.62 0.67 12.5 8.16

! i I I

!
i

:t :t :t :t :t :I: :I: :I:


:I:

16

794

:t

22

0.6
25 0.01 0.1 0.26 0.013 0.008
0.16

77.7
0.61 12.8 4.23 1.03 0.898 5.09 3.45 42.8 584 3.33 155 1.84 2.45 5.44 280 1.76 13.3 7.80

:t
:t :t :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :t

2.5
0.08 0.3 0.12 0.02 0.018 0.18 0.18 1.6 10 0.04 8 0.14 0..03 0.04 65 0.03 0.1 0.07

i
I

I
i

(ppm)
(ppm) (ppm) (ppm)

(ppm)

(ppm) (ppm) (ppm) (ppm)

.! !

i !
!

! i
i

:I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I: :I:

0.22 0.3 0.01 9 0.04 0.8 1 0.02 0.02 '0.06 28 0.02 0.04 0.1 0.20

i .

Zn Zr

(ppm) (ppm)
I

2.68 78 204

:I: :t: :t:

0.18 8 9

2.63

:t:

0.05

ppm -- parts per million

Normalizedto MURRthroughunpublishedanalysesof Perlman's Pottery standardrelative to the

SRM-278 Obsidian Rock reference standard previously conducted at MURR.

235-

Burger et al.: Jampatilla ObsidianSource

Table 3. Archaeologicalsites with artifacts of Jampatillaand other types of obsidian.


I Jampatilla Quispisisa Andahuay-! ! source Ji source ! /as A type
Sites in the Cararu.'

!
I ,

1 !

'

I
! i .
I ! :
1

huarazo Valley

I
i I
:

I
I
I I

Aka source

! Andahuay-!

i
:
'

Acari

1
I

Other

/as B Type!

type I types n.._...................................

I
i i I
i

I
!
; I
1 I
,i
I

. PAy 5-1 . PAy 5-5


. Jincamocco

3 4 22
1 :
I

1 6 24
2

I
:

1 1

Iii 1 ,,:

i I

. i i Iii
. "" :

i 'I ill 1
::

Ii! I,

II! I
I

I 5

'

.: Sites in the Pampas


.

PAy5-7

I
! 1
!
,

!
i
,

!
I

i :
i :
i

i i
i 1
33

i :....................................................................-........

i !

. i i I .

"" i:

. 1 .

Valley

. Waywaka
. .Kunka Taka . . .L Acar( S 'te,nt"e
Valley
" ~ '

16 i

15
" 'i"
I

1.

i
" : 1

j i

I Iii

"

I : i :

i i
i ! . i
Z

i :

2 i " i Iii j ,

11

3
~

.......

I
! :
I 'j
Z

Iii I

. .
I I
I

Iii
1 :!:
:

1
2

. Hacha
Site in the Cuzco Valley

.
I
! '

7 .." J.""... I ...""" 40 "

! . :
:
!

I " .""..."... :
I
I

I
I I

I
I
,

! :." : ' I I
I

! ::
I
:

12

:.."

.......

I .

. W1D1pilla

i I

1:

i !
:

i :
:

! I i I
:

-1

i I

i i

!
.

ANDEAN PAST 5 (1998)

-236

Wlmpllla

.o

Piklllaqla

river
..I. department

a [

50 kilometers

AREQUIPA

.,. boundary o moderI'!town . obsidian source . archaeological site with Jampatilla obsidian present o other sites . mentioned in text

'Figure 1. Map of southwestern Peru showing the location of obsidian sources and archaeological sites mentioned in the article.

237-

Burger et al.: Jampatilla Obsidian Source

612000

1
KM

Figure 2. Map of the Jampatilla source. Obsidian outcrops observed in 1981 are marked with X's. Outcrops occur along the north side of the Jampatilla ridge and on the valley side to the southeast. Another smaller outcrop was observed on the north side of the Toqsa ridge. Shaded areas indicate prehistoric obsidian workshops. Ay5-46 is an archaeological site pertaining to the Huari Empire (AD 750-900). Map based on Carta Nacional, 1:100,000 series, sheet 29-0 "Querobamba," Instituto Geografico Milit, Peru.

ANDEAN PAST 5 (1998)

-238

C'i

to co ..0

Andahuaylas B type

Jampatilla
Pampas type

Lri 0
Lri

~
Ayacucho type

... .

artifacts
JampatiUa source samples

Q. J:

E to Q.

..,:
C'i

llticaca Basin type

co ,..)

~
,..)

~
4

And8huaytas A type

0 ....,

:~~~
2

~
6

AcariW.
8 10

12

14

16

Cs (ppm)

Figure 3. Bivariate plot of Cs vs. Hf for obsidian artifacts and source specimens from Peru analyzed at LBNL and MURR with 95% confidence ellipses surrounding source groups with five or more members. The confidence ellipse for the Jampatilla source samples and Pampas type artifacts was calculated after combining them into a single group.

239-

Burger et al.: Jampatilla Obsidian Source

Pampas type artifacts

o
Jampatilla source

samples

Andahuaylas B type

_0 E

CXJ

Q. Q.

::I

<0

Andahuaylas A type

~
Trticaca Basin type

~
Oulsplsisa
AcarI typo A A

Cuzco type

ci~

~
800

t"I o O.

~yacucho
200

4'pe
400

600

1000

1200

Sa (ppm)

Figure 4. Bivariate plot of Ba vs. Eu for obsidian artifacts and source specimens from Peru analyzed at LBNL and MURR with 95% confidence ellipses surrounding source groups with five or more members. The confidence ellipse for the Jampatilla source samples and Pampas type artifacts was calculated after combining them into a single group.

UNIFACES IN EARLY ANDEANCULTURE HISTORY: THE NANCHOC LITHIC TRADITION OF NORTHERN PERU
Jack Rossen Ithaca College & University of Kentucky

Introduction
This paper describes and discusses a lithic industry of northern Peru that has been studied for several years. The Nanchoc Lithic Tradition (NLT) is associated with sites on the forested western slopes of the upper Zafia Valley dating to the Middle Preceramic IV Period (8000-6200 B.P.). The chipped lithics described here were recovered from a series of residential sites in a lateral quebrada or side canyon of the valley with alluyial fans (the Quebrada de Las Pircas). The sites are characterized by intact midden and features, including quincha (wattle-and-daub) hut floors, burials, and a variety of activity areas (Djllehay et al. 1989; Rossen 1989, 1991). Other artifact assemblag~s such as ground stone, faunal remains, marine shell, daub, and botanical remains were also recovered and have been described in these sources and elsewhere (e.g., Rossen et al. 1996). Chipped lithics are the focus of this paper for several reasons. Foremost, lithics are usually the dominant assemblage in South American Preceramic sites, and thus lithic traditions have come to be strongly identified with those cultures. Lithics also provide the broadest basis for comparing Preceramic Andean cultures, many of which (such as the Paijan complex of northern Peru) have been defined primarily on the basis of surface-collected sites with minimal context or few non-lithic assemblages (see Chauchat 1988). In addition, there have been conceptual advances in the study of lithic technologies that allow examination of organizational implications of lithic assemblages (Binford 1987). In this regard, archaeological correlates have been developed to match different types of lithic assemblages with cultural traits such as (1) varying degree of group mobility or sedentism, (2) generalized versus specialized economies, and (3) long- versus ANDEAN PAST 5 (1998):241-299.

short-term tool use (Henry and Odell 1989; Johnson and Morrow 1987; Torrence 1989). A basic premise of this paper is that these conceptual advances in the study of lithics offer means to produce new insights into early Andean culture history. The widespread presence of unifacial preceramic lithic traditions in northern Peru and southern Ecuador has been interpreted in a number of ways. As examples, unifacial industries have been viewed as indicative of the presence of dispersed, related social groups (Malpass 1983; n.d. [c. 1984]), as repetitive, independent adaptations to similar environments and resources (Richardson 1978:280), and as representative of more plant-oriented and woodworking functions (Ranere 1975; Richardson 1981). This analysis of the Nanchoc Lithic Tradition (NLT) adds detailed contextual data to the analysis. It will be argued that, in this case, unifacial lithics represent a shift toward a more diversified economy favoring plant exploitation, a lowmobility localized permanence, and the subtle beginnings of intensification. Here, intensification means a social reorganization or shift toward the potential to produce more, coupled with a commitment to change; specifically, an irreversible move toward sedentism, foodproduction, and group-level integration (Bender 1978). Lithic technology is but one of several data sets that led to this conclusion (Rossen 1991). However, because lithics constitute the largest data set of the Nanchoc sites, and because lithic contexts were documented in detail, it is useful to examine these unifaces in their own right. This article thus aims to characterize the lithic industry and touch on some related issues of how lithic technology may relate to economy and culture change in the forested slopes of Middle Preceramic northern Peru. Considerations of space preclude any detailed incorporation and discus-

ANDEAN PAST 5 (1998) sion of the smaller supporting associated assemblages such as faunal and floral remains (see Rossen et al. 1996). The organization of this paper is as follows. First, the research area of the upper Zafia Valley will be described, along with a summary of the twenty-one-year history of the Zafia-Niepos Archaeological Project. A background section provides a broader context of the problems and hypotheses within which the Zafia Valley research has been conducted. The NLT is then described in terms of its stages of reduction, attribute-variables, formal tool typology, and site contexts. The final section discusses some conceptual and cultural historical issues related to the NLT, including the possibility of the existence of an AndeanPanamanian Unifacial Complex (Malpass n.d. [c. 1984]) and the possibility that Middle Preceramic unifacial industries signal changes in social organization, economy, and a new direction in cultural evolution. The Upper Zaiia Valley The Zafia Valley, in terms of water carried and arable land, is the sixth largest of Peru's north coastal valleys (Moseley 1983:785). It is situated parallel to, and immediately to the north of; the Jequetepeque Valley (Figure 1). The valley has two branches, with the Zafia Valley proper being the northern branch and the Nanchoc River the southern branch. These two branches have their confluence near the town of Oyorun, at what is considered to be the boundary between the middle and upper valleys. Today, the upper valley contains a string of small towns of maize farmers and herders. Other valley inhabitants live in dispersed homesteads on hill spurs and flats above the valley floor. The valley is ecologically unique and important. It is situated at one of the lowest points of the Andes chain, and contains Peru's closest juxtaposition of coast, sierra, and tropical forest. The valley ecology includes a relict tropical montane forest on its western, uppermost slopes (Craig 1985; Dillehay and Netherly 1983; Dillehay et al. 1989:734-736; Koepcke 1954; Koepcke and Koepcke 1958;

-242 Weberbauer 1945:27, 47, 444-446, 465-471). This forest is a remnant of a once-continuous band of tropical forest that crossed several northern valleys of Peru (Simpson 1975; Vuilleumier 1971). The conquistador Francisco Pizarro travelled up the Zafia Valley on his way to Cajamarca in 1532, stopping to rest at an Inca ad-' ministrative site at Nanchoc for several days (Trujillo 1948 [1571], discussed in Hyslop 1984:66). The valley and its vegetation was described by the. Italian naturalist Antonio Raimondi, who visited the upper valley in 1868 (Raimondi 1940 [1874]:262-263). Raimondi noted that the forest stretched from the top of the valley at 3500 m down to Nanchoc, then an hacienda, located at only 360 m in elevation and 80 km from the coast. Even then, hacienda Indians were cutting trees for export to the coast. In 1905, and again in 1934, the region was visited by the German botanist Augusto Weberbauer, who described the forest as reaching pown to about 5 km above Nanchoc (Weberbauer 1945:27, 47, 444-6). By the 1950s, when the region was again described by Hans and Maria Koepcke, the remaining forest ended 15 km above Nanchoc (Koepcke 1954; Koepcke and Koepcke 1958). The densest remaining pocket of this forest is located on the Zafia side of the valley near Taulis. At present, this surviving pocket of western slope tropical forest, located approximately from 1300-2400 masl grades into a humid, ceiba (Ceiba sp.) forest at elevations of about 800-1300 masl, and then into a semiarid thorn forest that covers the lower portion of the upper valley (500-800 masl) including what are now the towns of Nanchoc, Bolivar and Carahuasi (Dillehay and Netherly 1983; Dillehay et al. 1989:734-736). The present-day vegetation of the Nanchoc area is a mixture of semi-arid and temperate species, including various types of columnar cactus (Cereus sp. and Cephalocereus sp.) together with small trees and shrubs (Bombax discolor, Puya sp., and Deuterocohnia sp.). Along watercourses are temperate trees (Celtis sp. and Myrica pubescens). This portion of the valley (ca. 80 km from the coast) is characterized by a series of branching, lateral que-

243-

Rossen: Nanchoc Lithic Tradition valley floor as agriculture intensified during the Formative Period, requiring more cultivable land and water. Also suggested was that important advances of the Preceramic Period, such as adoption of cultigens, sedentism, and clear separation of domestic and non-domestic space, originally occurred above the main valley floors, in narrow side quebradas that had received little archaeological attention (Dillehay and Netherly 1985; Dillehay et al: 1989).

brada and alluvial fan networks that contain access routes to the tropical forest above and the arid middle valley and coastal plain below (Figure 2). Previous Research

Throughout the history of the Zafia-Niepos Project (now encompassing nine field seasons) the ,ecology and archaeology of the Zafia Valley has offered new insights into the cultural developments of the Middle and Late Preceramic Period in northern Peru. Prior to the Among many results, Dillehay and Netherly's research (I) provided evidence from a 1987 research, four field seasons directed by small Middle Preceramic public site (the CeTom D. Dillehay and Patricia J. Netherly (1976, 1978, 1981, 1984-85) provided an ini- menterio de Nanchoc Site [CA09-04]) and (2) defined many habitation middens, including at tial understanding of the complex ecology, least two with intact subsurface deposits, and geology, and archaeological potential of the suggestive evidence of complex internal valley. Their research laid the foundation for structure and activity areas. Although not further, specific investigations by locating and segmented in the classic crafts specialization test. excavating numerous sites and defining a sense (Flannery 1969), there existed speciallocalized chronology (Dillehay 1985; Dillehay and Netherly1983,1985;Dillehayet al. 1989; ized, separated public activity and segmentation in terms of the dispersed settlement patNetherly and Dillehay 1985, 1986). tern. The Dillehay-Netherly hypotheses preDillehay and Netherly's early surveys in- sented elements of an alternative model of the dicated a substantial Preceramic habitation in circumstances surrounding Preceramic subsistence intensification, including the possible the upper Zafia Vailey. In particular, surveys adoption of plant cultivation, and the subseconducted in 1984-5' succeeded in locating quent deyelopment of early complex society in sixty-two such sites, of which forty-eight are northern Peru. The author's 1987 research concentrated in a ceiba montane and arid thorn forest ecotone of the Nanchoc area. Within was designed to follow-up with an intensive that ecotone, the sites are further clustered in investigation of a related- series of one quebrada's habitationmiddens in order to (1) de10 square kilometers of elevated alluvial fans fine Middle Preceramic domestic activities near stream headwaters in the upper portions and lifeway, (2) evaluate and further develop of the lateral quebradas (Figure 2). Dillehay and Netherly's localized model of The location of dense clusters of Pre- cultural development, and ultimately (3) relate ceramic sites on selected alluvial fans in side the findings to broad conceptual and theoreticanyons well above the valley floor, and the cal debates concerning intensification and cultural evolution. A few of these issues will total absence of overlying ceramic occupations were considered conspicuous and sig- be explored as they specifically relate to the nificant features of the Zafia Valley's archae- chipped stone industry. ology. Although clustered, these sites appeared to represent an intact relict of a dis- The Quebrada de Las Pircas persed settlement pattern of many small sites, Today the lateral quebradas of the Nanwhich tended to segment activities. From choc area are semi-dry to dry, with only a few these features, Dillehay and Netherly (1985) hypothesized that earlier Preceramic populawidely-scattered springs, and are uninhabited tions were concentrated high in the lateral on a permanent basis. The quebradas are only quebradas on alluvial fans, and that, through sporadically utilized, for hunting, except durtime, they gradually migrated toward the main ing years with heavier-than-normal rainy sea-

ANDEAN PAST 5 (I 998) sons (December to April) like 1986-87, when the area is used for foraging herds. Of the many quebradas in the Nanchoc region, the Quebrada de Las Pircas was a natural choice for intensive research because (1) it was the largest, most geologically developed quebrada and alluvial fan system in total area surveyed during the 1984-85 field season, (2) its Preceramic sites were more numerous and intact than those in other quebradas, and (3) test excavations at two sites in 1985 revealed intact deposits and Middle Preceramic radiocarbon dates (Dillehay et al. 1989). A 1969 air photo of the southwest to northeast sloping Quebrada de Las Pircas depicts a sinuous dry streambed flowing through a canyon with little vegetative cover or zonation (Figure 3). This rocky dry bed that is up to 10-15 m deep runs along the southern foot of the hill, CelTOTingues. During the 198587 investigations, the quebrada had much heavier vegetation, and a major curve in the dry streambed depicted in the air photo was straightened by the 1982-83 El Nifio rains. This change in stream course is depicted by a dotted line in Figure 4. A lower pampa with dense undergrowth, passable by only two footpaths, opens into a higher, less vegetated, more dissected pampa containing alluvial fans, dissected alluvial fan remnants, and hill spurs. At its top, the quebrada is enclosed by three long hills, CelTo Tingues, CelTo Taza, and CelTo Conquis, which range in height from 1000 to 1200 m. South American Unifacial Lithic Industries As a unifaciallithic industry, the Nanchoc Tradition must be conceptualized in a different manner than bifacial industries. There are several examples of primarily or exclusively unifacial lithic industries in lower Central America and the Andes, some of which are briefly listed below from north to south with their locations, environmental reconstructions, and approximate dates: Chiriqui (Ranere 1972, 1975; Linares and Ranere 1980): location: western Panama, Pacific side

-244 environment: semi-evergreen seasonal forest date: 6500-5000 B.P. economy: advanced hunting-gathering, incipient gardening? Aguazuque (ColTealI989): location: Cundinamarca Dept., central, Colombia environment: dry low montane forest date: ca. 4000 B.P. economy: advanced hunting-gathering and

gardening

Las Vegas (Stothert 1974, 1985, 1988): location: Santa Elena Peninsula, southwestern Ecuador environment: mangrove swamp, thornscrub with woodland and thicket patches date: 10,000-6600 B.P. economy: hunting-gathering, incipient gardening Sangay (polTas 1988): location: eastern Andean slopes, southeastern Ecuador environment: tropical montane forest date: ca. 5000-4000 B.P. economy: horticulture-agriculture Cubiltin (Temme 1982): location: eastern cordillera, southern Ecuador environment: tropical montane forest date: unknown economy: hunting-gathering? Siches (Richardson 1969, 1978, 1981): location: Talara, far northern coastal Peru environment: mangrove swamp, savanna woodland, savanna with extensive riparian forests date: 8000-5000 B.P. economy: hunting-gathering, gardening at end of sequence? Mongoncillo (Malpass 1983): location: lower Casma Valley, central coastal Peru environment: open park forest, coastal valley

245date: ca. 9000-8000 B.P. economy: hill1ting-gathering

Rossen: Nanchoc Lithic Tradition grouped together into a "Unifacial Complex" (ibid.). In discussing these concepts and perspectives on unifacial lithic industries, it is first important to emphasize that the NLT is not an anomalous phenomenon, but is instead one of several Preceramic unifacial industries located throughout forested or semi-forested zones of tropical America. Second, it will be argued that despite broad similarities with other ill1ifacial industries, specific traits of the NLT differentiate it from other known unifacial industries. Three particular examples are (1) the presence of recurring formal tool types, (2) the amOill1tof edge trimming and modification, and (3) the substantial amOill1tof associated groill1d stone. Because of the above distinctions, a slightly different perspective is taken here: that the various unifacial industries of Ecuador and Peru represent industries associated with less mobile groups and primarily plant-oriented economic strategies. In part, this perspective recognizes that (1) great variation may exist between unifacial industries, and (2) ill1ifaciallithics represent part of a plant-oriented economy rather than a cultural anomaly requiring the genetic linking of widely-distributed human groups. That is, this perspective suggests that ill1ifacial lithics are part of an adaptive option related to emphasis on plant subsistence, woodworking, and in the particular case of the NLT, subsistence intensification in the form of a diversified economy emphasizing plants and perhaps gardening. NLT Site Descriptions Controlled surface collection and multiple block excavation were performed at three Nanchoc-area sites (CA09-27, CA09-28, CA09-52) and supplementary limited test excavation was performed at three other sites (CA09-85, CA09-86, CA09-87) (Figure 5). Lateral quebrada Preceramic sites in the Nanchoc area tend to be quite small (with a few exceptions), often ill1der1000 m2 in total area. This can be a research advantage, in that block excavations can examine a substantial percentage of a site to sample its activities and contents reliably (Mueller 1976). The three sites chosen for intensive investigation were

In considering this list, it should be noted that ill1ifacial lithic industries related to forested environments have also been documented in other areas of South America, particularly Brazil (see Dillehay et al. 1992; Schmitz 1987 for general discussions). Also, South American unifacial lithic industries are not limited to the Preceramic Periods. For example, a coastal split-pebble tradition noted on the Peruvian north coastline in the Zafia and Jequetepeque Valleys continued ill1til Inca times (Dillehay et al. 1990). The above list is given merely to highlight examples of some industries that are closer to the Nanchoc study area and/or have been well-documented. Of these, the works of Ranere, Richardson, Malpass, and Stothert constitute the most extensive analyses. In comparing these industries, two important similarities may be noted: 1. Many unifacial industries are associated with forested environments. These are regions where plant and wood exploitation would be expected to be important. Conversely, hill1ting.(with associated bifacial lithics) would be a more important subsistence activity in less forested, more open environments where animals might be more accessible. 2. Many (but not all) tools in these industries are thought to have been used to make primary tools of wood or other perishable material, rather than having been primary tools themselves (Richardson 1969, 1978). When the north-central Andes (particularly Peru and Ecuador) are specifically considered, there are currently two perspectives on the series of ill1ifaciallithic industries that have been recorded. The first perspective is that these industries represent independent adaptations to similar environments and resources (Richardson 1978:280). The second perspective is that these industries are genetically related and represent actual intergroup contacts (Malpass 1983; n.d. [c. 1984]:3). Under this second perspective, the various industries are

ANDEAN PAST 5 (1998) documented during survey in 1985 (Dillehay and Netherly 1985), while the three sites chosen for limited test excavation, located in the upper reaches of the quebrada, were documented during the 1987 field season (Rossen 1988). All sitys but CA09-86 and CA09-87 consisted of moderate to dense lithic scatters with occasional ground stone present. Those sites contained from 30 to 60 cm of intact midden and included subsurface features such as a quincha (cane and mud) hut floor with surrounding post holes, and both intact and fragmented human burials. Site CA09-86 was a light lithic scatter without underlying midden and site CA09-87 was a modified outcrop and heavy scatter of deep violet-colored quartzite. These sites were designated as follows: CA09-27 - Quebrada de Las Pircas 1 CA09-28 - Quebrada de Las Pircas 2 CA09-52 - Quebrada de Las Pircas 3 CA09-85 - Quebrada de Las Pircas 4 CA09-86 - Quebrada de Las 'Pircas 5 CA09-87 - Quebrada de Las Pircas 6 Site Stratigraphy and Integrity Multiple block excavation at three sites (CA09-2.7, 28, and 52) and test excavation at two other sites (CA09-85 and 86) conducted in 1987 produced consistent stratigraphic profiles of eolian (wind-blown) midden deposits sandwiched above the hard alluvial fan subsoil and below a thin (5-8 cm) hard cap of colluvium (Figure 6). The hard cap is attributed to seasonal rains and the resulting surface deflation, discussed briefly below. Regardless of depth of deposits, which, similar to the 1985 testing phase, ranged from 30 to 60 cm below surface, this general "stratigraphic sandwich" was present throughout the 120 square meters of excavation. Furthermore, all subsurface features, ranging from a quincha (wattle and daub) hut floor and post holes to a small stone structure and partial and complete human burials, began in a narrow range of approximately 22 to 28 cm below the modem surface. That is, all features appear to be interrelated within the same subsurface midden zone without crowding or overlap, giving the impression at each site of a single major occupational

-246 episode (Figure 7). All excavations thus produced evidence of relatively shallow middens with simple stratigraphy and apparently interrelated, well-spaced features. A recognition that all sites are disturbed to one degree or another, and a discussion of that disturbance is crucial to evaluating all ar-, chaeological evidence (Hassan 1987; Nash and Petraglia 1987; Schiffer 1987). In the Quebrada de Las Pircas, site disturbance was considered in terms of (1) the extent of surface versus subsurface deposits at these small sites, (2) surface artifact frequencies and lengths located within versus outside the boundaries of subsurface deposits, and (3) artifact frequencies and lengths in various 10 cm subsurface levels. The detailed analyses of the sites' integrity' and. its experimental correlates (sensu Petraglia and Nash 1987; Schick 1987) have been presented elsewhere (Rossen 1991:103-125). The results of those analyses appear to indicate that the Nanchoc sites experienced only a low level of disturbance, consisting of surface deflation and downslope horizontal "smearing" of surface artifacts. Consistent means, modes and standard deviations of artifact size, both horizontally and vertically, suggest that more damaging post-depositional processes such as "size winnowing'" of artifacts that remove selective portions of assemblages or vertical displacement are not major factors at these sites. Stylistic consistency of artifacts at all levels of individual sites and between different sites in the QLP suggest that all site deposits are related to the same occupation, or at least the same cultural stock (ibid:153-155). Radiocarbon Dates A series of fourteen radiocarbon dates has thus far been received. Table 1 presents the ten accepted dates and their contexts. The four problematic dates are discussed in Rossen et al. (1996). The samples of aggregate wood charcoal were submitted from the best contexts of the sites, including the quincha hut floor (depth 25 cm from ground surface), a narrow midden level just beneath that floor (depth 32 cm) and an articulated burial (depth

24736 cm). Nine of these ten samples produced Middle Preceramic Period dates. The acceptable dates (that is, dates that match the evidence of artifacts, stratigraphy, and previous project data) range from 7630 B.P.:I: 80 (Beta30778) on the hut floor of site CA09-27 to

Rossen: Nanchoc Lithic Tradition five sites. Each site appears stratigraphically to consist of a single component, representing occupation by a single group with a consistent material culture. Furthermore, the QLP sites appear to be culturally interrelated by their similar artifacts, stratigraphy, and spatial proximity. As subsequent field seasons refined the broad Preceramic sequence of the Nanchoc area, the interrelatedness of the QLP sites was further reinforced (Rossen and Dillehay 1994). The lithics recovered from these sites may thus be discussed as representative of a single stone-working tradition. It is further believed that these data may represent a broad, valley-wide cultural pattern of the Middle Preceramic Period (Dillehay et at. 1990). ANALYSIS FRAMEWORK The chipped stone collection (n=35,537) was analyzed from both more traditional perspectives such as reduction sequencing (Bradley 1972; Collins. 1975; Schiffer 1976), attribute-variable analysis, and formal morphological typology (Brew 1946; Ford 1954; Krieger 1944; Rouse 1939, 1960), as well as from the perspectives of microscopic use-wear (performed by Dillehay) and spatial analyses to define the basic lithic tradition elements and to assign functions. Due to space limitations, only the reduction sequence and certain aspects of the attribute-variable and typological analyses are presented. An effort was made to present portions of the analysis that broadly define the characteristics of the NLT. Analyses that are not discussed here either generally corroborated the data presented (such as micro-wear analysis) or were oriented toward defining activity areas at individual sites. In addition to the above-listed traditional forms of lithic analysis, newer concepts and archaeological correlates have been developed to match different types of lithic assemblages with such cultural traits as varying degree of group mobility or sedentism, generalized versus specialized economies, and long versus short-term tool use (Henry and Odell 1989; Johnson and Morrow 1987; Torrence 1989). Organizationally-oriented studies are based on conceptual dichotomies that apply to lithic

8410 B.P. :I: 140(Beta-33526)from a test unit


of site CA09-85, high in the quebrada. As may be noted in Table 1, three dates with acceptable results were processed on the extended count method that assures greater accuracy for small samples. Wood charcoal samples were also examined for possible contamination by carbonates and none were found (Murray Tamers, personal communication, 1989). An unacceptable date of 2270 B.P. :I: 140, from wood charcoal of level 4, site CA09-52, would relate to the Late Formative Period. The absence of ceramics or other artifacts of this period at these sites suggests this date represents either charcoal . from a root burned after the site occupation, sample contamination, or another unspecified problem. Other problematic radiocarbon dates, run on botanical remains using the Accelerator Mass Spectrometer (AMS) method, are not relevant to a discussion of the lithics and have been discussed in detail elsewhere (Rossen 1991:149-153,517-528; Rossenetal. 1996). Interrelated, Single Component Sites In summary, six sites were investigated in the Quebrada de Las Pircas, one of severallateral canyons and alluvial fans located throughout the Nanchoc branch of the upper Zafia Valley. This series of sites was investigated both (1) as a continuation of the longterm research founded and nurtured by Dillehay and Netherly's Zafia-Niepos Archaeological Project (Dillehay and Netherly 1983, 1985; Dillehay et al. 1989) and (2) as an independent research project with somewhat different conceptual and theoretical goals, including an emphasis on past subsistence intensification activities (Rossen 1991). A total of 120 m2 was excavated. One of the more important results of the 1987 research was the recovery of a large collection of lithics from intact subsurface deposits, house floors, and other features of the

ANDEAN PAST 5 (1998) assemblages in their entirety. Of these, Binford's (1977, 1979, 1987) expedient versus curated industries dichotomy is most relevant to the NLT. Expedient industries contain tools that are manufactured, used, and discarded according to momentary needs, while curated industries contain tools that are manufactured in anticipation of future use, are maintained, transported, and often recycled or rejuvenated (ibid.; Bamforth 1986). In the classic sense of this conceptual dichotomy, expedient tools are technologically simpler and morphologically less-patterned than more technologically sophisticated and formally distinct curated tools (Bamforth 1986). The essential difference is one between an industry geared toward immediacy, a wider variety of tasks, utilization of resources on an encounter basis, and a generalized economy (expedient), versus an industry oriented toward logistical planning and more specialized tasks and economies (curated) (Binford 1980, 1987). Related to this dichotomy is another conceptual dichotomy, that of embeddedness versus trade (Binford 1979). The concept of embeddedness is essentially that exotic raw material may be procured in the natural course of hunting or other mobile economic strategies, even for basically sedentary groups (e.g., Morrow and. Jefferies 1989). Embedded procurement of non-local material may then be low in cost. The archaeological correlate of this would be whether non-local (exotic) raw material occurs in a more reduced state than local material at a site (reflecting exchange), or whether both local and non-local raw materials are treated the same in reduction terms (reflecting embedded procurement) (Torrence 1989:30). Thus, at a series of interrelated sites such as the QLP sites, where local versus nonlocal materials are easily distinguished, analysis related to raw material types can be utilized to infer degree of embeddedness, mobility, and economic generalization or specialization of the group. Of course, in the broader research milieu, other site assemblages were utilized to support the observations based on these lithic conceptual dichotomies, and the realities of analysis produced a result more complicated than simple dichotomies allow. However, these new conceptual trends in lithic

-248 analysis produce interesting supplemental information to the traditional analyses of reductive stage, attribute-variables, and formal tool typology. Before discussing preceramic Andean lithic industries, it is important to note the variety of analytical terminology and methodology that is present in the literature. Termi-' nologies utilized in connection with unifacial industries range from several variations on simple descriptive terms (e.g., Malpass 1983; Richardson 1969) to an emphasis on flake production technique and flake shapes (Stothert 1974, 1988). These differences reflect differences in the excavation contexts and morphological characteristics among unifacial industries. The present terminology employed differs somewhat from those previously utilized for the same reasons the previous analyses differ from each other. The combination of attribute-variable analysis with formal typology and reductive stage compilations in part represents an attempt to minimize comparison problems between industries. The present analysis differs even more from the terminology and style used to examine bifacial preceramic lithics of northern Peru. The Paijan industry, the latter portion of which slightly predates the NLT, is defined by its elongated bifacial prepared cores and stemmed projectile points (Chauchat 1975, 1978; Uceda 1986, 1987). There is no doubt that the Paijan industry represents a hunting economy, mobile settlement system, and bifacial lithic technology that are fundamentally different than those related to the NLT. The Paijan lithic studies placed an emphasis on morphological description and functional typology and heavily deemphasized attribute and contextual analyses, partly because the assemblages were collected from surface sites with poor contexts (for a more complete discussion, see Rossen 1991:172-174,656-657). Supporting microscopic use-wear analysis was not conducted. The present analysis differs greatly in analytical terminology and emphasis from the Paijan studies because of the fundamental differences between unifacial and bifacial industries, and because of differing analysis philosophies (Adams and Adams 1991).

249Stages of Reduction The stage of reduction concept was originally developed and utilized to better understand manufacturing reduction of bifacial lithic industries (Bradley 1972; Collins 1975; Schiffer 1976). Other specialists consider lithic reduction to be a behavioral continuum, of which a series of stages is an inadequate representation (e.g., Flenniken 1985). The approach taken here is to consider the NLT as a behavioral reduction continuum within which stages of reduction can be isolated and examined. Bifacial industries to which the reduction stage concept has been applied are commonly char~cterized in five to six distinctive stages of reduction (e.g., Boisvert et al. 1979:60-62). In the present case of a unifacial industry, only three stages of reduction were isolated (Figure 8). Stage 1 is represented by blocky, angular cores chipped from nodules and boulders, primarily collected in the rocky quebrada floors adjacent to the research sites. Stage 2 consists of secondary flake tool types that have not been further shaped, trimmec;l,or thinned. Stage 3 consists of secondary flake tool types that ha~e been shaped, trimmed, or thinned into what are considered here to be recurring formal tool types. Figure 8 also illustrates both products and by-products of reduction within the NLT. That is, stages of reduction are defined in terms of production goals (cores, secondary flakes, and worked secondary flakes), but the lithic assemblage also contains by-products of the reduction process (primary flakes, exhausted cores, and tertiary flakes). Primary flakes are defined as flakes containing cobble cortex. Exhausted cores are cores too small for further removal of flakes. Tertiary flakes cussion that are smaller than 1.5 cm in their largest dimension. In terms of artifact use or consumption, tools exhibit -little retouch and no rejuvenation (that is, reworking and reshaping of worn-out tools for different func~ions). Thus, tool consumption figures as only a minor component of this reduction sequence. Although formulated for complex bifacial industries, the stage of reduction concept is

Rossen: Nanchoc Lithic Tradition useful in this case to demonstrate the relative simplicity of this unifacial industry in terms of manufacturing reduction. The complexity of the NLT lies not in its principles of reduction, but in the variety of tool shapes and recurring formal types that exist within the Stage 3 (modified secondary flake) level. The NLT contains an extensive, proliferated Stage 3 instead of additional lower stages of reduction. , In his writings on lithics, the master flintknapper Don E. Crabtree emphasized the importance of the quality of available raw material above all other factors in characterizing and understanding a lithic industry reduction sequence (Crabtree 1973). In the NLT, the relatively brittle nature (compared to silexes) of the most common materials in the collection (basalt, tuff, diorite, andesite, and rhyolite) may help account for the attenuated series of reduction stages and the relatively great amount of tertiary flakes. But even the Nanchoc lithics with their generally coarse raw materials differ substantially from the Las Vegas, Ecuador lithics,.where low quality raw material was considered an important reason for the preponderance of irregular small flakes, unprepared cores, and lack of formal tools (Stothert 1974, 1988). In another case, comparison of Nanchoc lithics with contemporary Junfn puna Preceramic lithics of central Peru, made of much harder, finer raw materials such as silex, indicate that the Junin lithics have more reduction stages (including substantial retouch and rejuvenation) and fewer tertiary flakes (LaVallee et al. 1985; Matos 1975; Rick 1980; Matos, personal communication 1987). Despite the undeniable effect of raw material. quality on a lithic industry, the position taken here is that the type of economy represented by the industry and thus, the intended functions of the produced tools, is a more important factor than raw material quality in explaining the reduction sequence of an industry. It will be argued that this unifacial industry with its three reductive stages is primarily related to procurement, production, and processing of vegetal products in the forest setting.

are flakes-without platforms or bulbs of per-

ANDEAN PAST 5 (1998) Attribute-Variables A long list of attribute-variables were recorded and processed. For the purposes of this abbreviated description, the most important attribute variables are raw material type, tool angle (the angle of the general tool body, indicating acuteness of tool shape away from the use edge), use edge angle, length, width, thickness, core type, core shape, and degree of core exhaustion. Formal Tool Typology Morphological descriptions were utilized in the analysis, while functional descriptions were avoided as misleading and oversimplifying. A more polythetic, visually-based classification approach was taken (Adams and Adams 1991; Gardin 1984), with a formal tool typology based on easily recognizable characteristics utilized alongside a detailed attribute analysis. Formal type definition was in part based on observations aI).daggregations of the attribute lists of Tom Dillehay, developed for the Zafia-Niepos Project. In other words, a classification scheme was adopted that stresses human definition and recognition of types independent of and parallel to attribute analysis, leaving computers relegated to compilation and statistical tasks. Type definitions were thus constructed on the basis of simple visual traits such as form and size. For example, distinctive, recurring rectangular, pentagonal, semi-lunar, and triangular flake tools were recognized during analysis. These tools were produced by a nonmarginal percussion technique that was appar.

-250 tools belong to Stage 1 (cores), and the common unmodifed utilized flakes belong to Stage 2 (uIimodified secondary flakes). The formal tool typology was constructed on the basis of four principles. First, basic shape or outline when viewing the dorsal surface. is the most important visual characteristic for defining a formal type. Second, size differences in forms that appear to be morphologically similar of-' ten indicate tools of different functions (Stockton 1977:341). Third, relative steepness of tool and edge angles was viewed as an important characteristic in describing types and ultimately, in differentiating functions (e.g., Gould et al. 1971). The strength of norms, or degree of standardization within a type varies considerably between types. Tool types with curved useedges generally exhibit less standardization than types with straight use-edge types. In some cases, the lower standard deviations on metrics of the straight use-edge types (versus curved use-edge type~) also indicate the greater strength of their norms. Other types are more generally-defined, and encompass a greater variety of specimens. The strength of type boundaries also varies greatly. Boundaries are relatively poorlydefined between types with similar morphologies that were separated by size, such as the quadrilateral types (1-3), semilunar types (47), and triangular types (17-19). However, as morphological subgroupings, their boundaries are well-defined and easily recognizable. Certain types may have considerable variation within the type, but also have relatively welldefined boundaries because of their fairly complex morphologies. ANALYSIS RESULTS Computerized analyses of the NLT illustrate the variability within this unifacial industry and demonstrate how complexity may be hidden within apparent simplicity. In this instance, the term complexity refers to the variety of recurring, trimmed flake tool types. Complexity was also expressed in other ways not discussed here, as in the spatial relationships of particular tool types with other as-

ently designed to produce repetitive flake


shapes with specific use-edge characteristics in terms of configuration and edge angle (Abler 1989; Speth 1972). Within the NLT, the presence of recurring, simple, unifacial flake tools in particular forms suggested a simple formal tool typology (Table 2). Rather than representing various stages of reduction, formal tool types all belong to the proliferated Stage 3 (trimmed secondary flakes) of the reductive sequence presented above, with two exceptions. The rare core

251semblages within discrete activity areas (e.g., Price 1981; White et al. 1977). Analyses were performed on an IBM 3090-600J mainframe at the University of Kentucky Center for Computational Sciences. The principle statistics program utilized was SPSSX. In the course of this descriptive analysis, attribute tendencies and modes, particularly related to edge angle, edge angle mode, and tool edge type, are defined and utilized as descriptive attributes of individual formal tool types. Some formal types are described as (1) generalized or specialized, (2) for fine or heavy use, and (3) as apparently related to specific actions such as slicing, cutting, scraping, and planing. Formal types are also referred to as expedient or curated (Binford 1977, 1979), based on whether the type is haphazardly-made and lightly-used (expedient) or carefully-made and heavilyused (curated). These are merely general and inexact distinctions that aid in categorizing and understanding diversity within the industry. Some of these assumptions and generalizations were further tested and corroborated by microscopic' use-wear. analysis (Rossen 1991:446-483). . Descriptive Analysis. Some basic compilations show the general nature of the NLT. Black basalt is by far the most conunon raw material type (46.5%) (Table 3). Other important material types are diorite (15.7%), tuff (12.9%), limestone (7.8%), andesite (7.5%), and rhyolite (6.5%). The tuff is a gray, veined material that was examined in thin-section by Charlotte Allen of the University of Kentucky Department of Geology. She wrote: "The material is an altered lappilli tuff. The word lappilli refers to particle size, with the thin-section having "spots" of equal size. This material was the air fall part of a silicic volcanic eruption that was laid down, and altered by hot water flowing through it and depositing silica in it. The result is a hard, fine-grained, flinty rock" (Charlotte Allen, personal communication, letter, 1990).

Rossen: Nanchoc Lithic Tradition

Together, the six primary raw material types comprise 96.9% of alliithics. They are all locally available as alluvial gravels in the rocky stream beds adjacent to or near the sites. A series of non-local silexes, jaspers and quartzes comprise only 1.1% of the assem;' blage. Exotic silexes were recovered in a variety of colors and color combinations: cream, coffee, dark green, black, black with coffee, gray, gray with red streaks, maroon, black with yellow, and banded. Jaspers were black with red streaks and cream, black, and red. Highly translucent crystaline quartz was also recovered. These materials, with the exception of crystaline quartz, are all more finegrained than locally available materials. If the collections in the Museo Nacional de Antropologia, Arqueologia e Historia in Lima are representative of the locations of lithic raw materials in northern Peru, it is most likely that many of these materials came from the Cajamarca highland area, approximately 100 km southeast of (up-valley from) Nanchoc. In comparison, silexes in coastal collections often exhibit a dull red color, such as a single Paijan point base specimen that was recovered from site CA09-27. Tools of the industry (Table 4) may be discussed in three categories: (1) unmodified utilized flakes, (2) common tool types, and (3) uncommon tool types. Unmodified utilized flakes comprise nearly half the tools (49.0%). Unmodified utilized flakes are flakes that have not been trimmed through non-marginal percussion, and thus modification refers to the tool manufacturing process following detachment of the secondary flake. These tools are thus distinguished from the so-called "modified" types where trimming (such as removal of the dorsal ridge and non-marginal edge shaping) occurred (Speth 1972). Despite the presence of recurring tool forms with nonmarginal edge trimming, one-half the tools used were merely unshaped flakes. The high percentage of these unmodified utilized flakes is characteristic of an expedient industry, because they reflect quickly-made, lightly-used tools made for immediate use and rapid discard.

ANDEAN PAST 5 (1998) Figures 9-16 illustrate representative formal tool types of the NLT. Common tool types consist of recurring, non-marginally edge-trimmed forms that include more than fifty specimens (Table 4). Types 2 (mediumsized quadrilaterals), 5 (medium-sized semilunars), 7 (thick semi-lunars), 12 (pentagonals), 13 (incurved edges), 15 (blocky tools with pentagonal cross sections), and 26 (long-handled, short use-edge forms) are the common types of the assemblage. Uncommon tool types recur in form but are not present in great frequency (Table 4). These include the large and small variants of quadrilateral (Types 1 and 3) and semi-lunar forms (Types 4 and 6), the triangular forms (Types 18, 19, 20), and heavier tool types such as thick rectangulars (Type 16) and thick pentagonals (Type 17). Pointed forms (Types 8 and 9) are present but rare (n=12). Uncommon tool types are important to understanding site activities despite their low frequencies. For ex-

-252 Third, there are some low frequency tool types, including pointed tools (Types 8 and 9) and large core tools (Type 10), that may be underrepresented in the assemblage. These tools are relatively well-made and heavily used, and represent a curated element in the industry. The implications of this mixture of expedient and curated tools will be discussed in the conclusions of.this article. Recently; the concepts of expediency and curation have come to focus on reduction sequence and particularly on amount of exhaustion (Henry and Odell 1989; Johnson and Morrow 1987; Torrence 1989). In this sense, expediency is viewed as wasteful, while curation is more conservative of raw material. Expedient industries should thus (1) be associated with regions that contain an abundance of suitable raw material, and (2) be relatively less reduced or exhausted. It has already been stated that the Nanchoc region has an apparent abundance of locally-available raw material. Despite this, the' NLT displays high rates of raw material exhaustion. Core type is the best single measure of relative amount of exhaustion in a lithic industry. Table 5 describes 772 recovered cores in terms of both exhaustion and percentage of cortex remaining. Exhausted cores were defined as cores that do not contain platforms and/or dimensions suitable to further removal of 3 cm long flakes. (In final analysis, the average size of small Type 14 utilized flakes was slightly smaller, 2.6 cm). Because most raw material is in the form of streambed cobbles, lower percentages of remaining cortex also indicate more complete use (that is, greater exhaustion) of material. A large majority of cores, 79.3%, are exhausted, with the largest single category being exhausted cores with less than 10% cortex (48.7%). When core type is crosstabulated with material type, it is apparent that all major material types are highly exhausted (Table 6). Basalt, the most common material in the assemblage as well as the most locally abundant, has only 19.8% non-exhausted cores. Other local materials such as rhyolite and diorite are also heavily exhausted. The least exhausted major material type, tuff, has 27.6% non-

ample, .only eight large core tools (Type 10)


are present, but these disti,nctive specimens appear to be related to heavy woodworking and plant processing, two important activities. Once worn out, these tools were further reduced to produce flake tools instead of having been discarded. Thus, these relatively large tools may be underrepresented in the assemblage. The compilation of tool types thus depicts a varied industry (Table 4). Approximately half the specimens are hurriedly made, unmodified (that is, untrimmed) utilized flakes (Type 14). These flakes represent expediency in the industry. Within the formal tools, there are frequently made, non-marginally worked recurring types that differ from flakes in having been trimmed to particular shapes, especially quadrilateral, semi-lunar, and pentagonal forms. These common tools probably represent intermediacy between. expediency and curation, in that they have been shaped, and some specimens show heavy, repeated usewear patterns on their edges. Some of these tools are made by striking only four to six blows along the edges of a flake. Others, including some pentagonal specimens, have many more flakes removed along their edges.

253exhausted cores. All 15 cores of exotic silexes are exhausted. In this case, local material abundance, general industry expediency, and lack of material conservation appear to not be associated. Prior to the general discussion below, it may be stated that our concepts of either abundance or expediency may be refined in order to understand this unexpected combination of industry characteristics. A further examination of cores and their forms (Table 7) adds detail to our understanding of industry reduction and exhaustion. Most core forms, particularly trapezoidal, subtriangular, and rectangular cores, are heavily exhausted. One relatively uncommon core form, spherical cores, and one extremely rare form, conical cores, display high percentages (48.4% and 100%, respectively) of nonexhaustion. These two core forms tend to contain few acute angles, making detachment of suitable flakes for tools more difficult. This suggests that spherical and conical cores were undesirable core forms. There may thus be a technological/morphological reason for. the non-exhaustion of some cores that further emphasizes the overall exhaustion and material conservation of this industry. Also noteworthy in Table 7 is the ability to place most cores within a fundamental form category. The core form categories ~tilized here denote core form as it is viewed towards its dorsal surface. Only 11.4% of all cores were classified as "amorphous", while trapezoidal, pentagonal, sub-triangular, and rectangular core forms together comprise 73.0% of all cores. While these are not prepared cores in the classic sense (e.g., Johnson and Morrow 1987), there are still preferred, blocky angular core types that facilitate removal of flakes for tools and may be better reduced to exhaustion. Recent publications have increasingly linked expedient, including unifacial, industries with "amorphous core technology" (see several articles in Henry and Odell 1989; Johnson and Morrow 1987; Torrence 1989). The basic premise in doing so is that expedient tools are quickly made with little thought to core preparation, and thus the end result of lithic reduction is an amorphous core. How-

Rossen: Nanchoc Lithic Tradition

ever, this strict dichotomy between prepared and unprepared cores omits the possibility that multiple, angular, blocky core forms may be present in expedient industries. In this sense, perhaps the term "multiple form" is sometimes preferable to "amorphous" in describing the cores of generally expedient industries. Exotic Lithic Materials All non-local silexes and quartzes are lumped together under the term "exotics" to facilitate a comparative discussion between local and non-local lithic resources. Though still a small sample (n=446), the lumped exotic material type specimens are important because of their relatively distant sources (Table 8). Most lithic studies treat non-local materials as inherently valuable, and expect them to have been treated, distributed, and reduced differently than locally available materials. The concept of embeddedness challenges that fundamental notion in some cases by including the collection of at least a portion of a group's exotic materials within a suite of procurement activities conducted by the home group (Binford 1977, 1979). This removes some non-local goods from the necessity of having been procured through exchange and other external relationships and thus reduces their value. A summary of the reduction of non-local lithic materials shows them to have a pattern of reduction similar to the dominant local materials (Table 8). The percentage of tools is higher among non-local materials, but this appears to be a function of the unusually high percentage of non-local material tools (12.6%) at one site (CA09-28) associated with specialized mortuary activities. At other sites, the percentage of tools made of non-local materials (5.0%) is much closer to the percentage of tools made of local materials (3.7%). It is noteworthy that the specialized activities of site CA09-28 apparently produced skewing in many categories of lithic data, and thus, despite containing all the basic aspects of the NLT, this site is less useful in statistically analyzing the basic characteristics of the industry. For example, the relatively high proportion of exotic tools at site CA09-28 occurs

ANDEAN PAST 5 (1998) where exotic material is aIso the lowest percentage (1.8%) of utilized unmodified flakes. It thus appears that there was more manufacture of exotic material tools, and more exotic material conservation in conjunction with mortuary activities. These data show that despite their possible association with special activities, exotic materials were generally treated much like local materials in terms of reduction and conservation. That is, both local and exotic materials have similar reduction sequence proportions and even when lithic reductive proportions vary .between sites, these differences are equally expressed in both local and exotic materials. This pattern suggests that exotic materials may have been at least partially "embedded" within the general lithic industry. This observation does not deny the probable role of sporadic trade and exchange in procuring non-local materials, but instead suggests one possible explanation of why exotic lithic materials, despite their scarcity, do not display archaeological patterns indicative' of greater value or utility than local materials. That is, exotic materials may have been sometimes procured by the "Iocalgroup themselves during forays for other resources (quartz crystals, shells, etc.) as well as having been received through trade or exchange. Edge Angles, Tool Angles, and Tool Edge Types Edge angles, tool angles, and tool edge types also provide important insights into the NLT. These attributes aid in differentiating at the morphological level slicing and cutting functions that require low edge angles from scraping and planing functions that utilize higher angles (Vaughan 1985:59-63; White et al. 1977). Tool edge type is determined from the relationship between edge and tool angles. When edge angle is greater than tool angle, a strong edge is formed that allows heavier tasks to be performed with the tool. When edge angle is lower than tool angle, a finer edge is formed that allows use for more delicate and precise tasks (Figure 17). As a combination of two attributes, tool edge type aids in differentiating heavier edges, required for use on

-254 harder materials, from finer edges used on softer materials. In general, analysis of these attributes contributes to a depiction of a versatile industry associated with a diverse economy requiring manipulation of a wide variety of materials. Specifically, slicing and cutting action on softer materials, probably plants, is common (Rossen 1991: Chapter 5). Certain tool types have narrow ranges for these three' attributes, while others show wide ranges or multiple peaks in attribute values suggestive of multiple-function implements. Figure 18 displays the frequencies of edge angles of the NLT. Two peaks exist, one at 30 and one at 40 degrees. A significant decline or valley is present at 35 degrees, between the two frequency peaks. This may represent two different types of fine cutting or slicing, or the working of different materials that require slightly different edge angles. Despite the presence of substantial frequencies of tools at high edge angles, the industry emphasis is clearly on relatively low.edge angles probably related to slicing and cutting actions. The same conclusion is drawn from the mean edge angle of 38.755 and the mode edge angle of 30. Diversity and versatility in the industry is reflected by a large standard deviation, 16.25, nearly one-half the mean. A different way of representing this is by plotting edge angles against tool types. Figure 19 has a cutpoint of five, meaning that one to five cases equals one, six to ten cases equals two, and so forth. Unmodified utilized flakes have been omitted. This figure shows that some tool types have a wide, equally distributed range of edge angles, while other types have one or two clearly defined edge angle modes. Note that Types 6 (small semi-lunar forms), 7 (thick semi-lunar forms), 16 (thick rectangular forms), 21 (the amorphous edgetrimmed category), and 26 (long-handled, short use-edge forms) all display regular normal curves with single, clearly-defined modes as follows:

255ToofType

Rossen: Nanchoc Lithic Tradition

Tool types 2 (medium quadrilateral forms), 12 (pentagonal forms), 13 (incurved edges), and 15 (blocky forms with pentagonal cross-sections) have multiple edge angle modes:
ToolType

other measures, such as Type 6 (small semilunar forms) and Type 7 (thick semi-lunar forms), concentrate in the lower 16-30 degree edge-angle increment. Unmodified utilized flakes (Type 14) also concentrate in this same low edge-angle increment, distinguishing them from a majority of formal tool types. This suggests a curious combination of the most general, expedient lithics and the most specialized, curated lithics as both having relatively lower edge angles than the majority of tool types. Thus, on the basis of edge angle tendency, expedient utilized flakes may have had somewhat different functions than the majority of formal tool types. Tool types that have similar forms, but were differentiated only by overall size, also may be differentiated by edge angle increments. Type 1 tools (large quadrilateral forms) concentrate in a lower edge angle increment than Type 2 (medium quadrilateral forms) and Type 3 tools (small quadrilateral forms). Type 5 tools (medium semi-lunar forms) concentrate in.a different edge-angle increment than Type 6 tools (small semi-lunar forms). This suggests that functional differ.ences as well as size differences separate these size-graded varieties of tools with similar forms. Type 15 (blocky forms with pentagonal cross-sections) is the only type that concentrates in the high edge angle increment of 4660 degrees. These are relatively well-made tools whose function, probably woodworking, perhaps could not be served by casually-made, more expedient tools or unmodified flakes. The cutpoint used in Figure 19 obscures patterns among rarer tool types, and negated consideration of their relationships within the overall industry. Conversely, although edge angle increments (Table 9) obscure the bimodal edge-angle distributions of some tool types, some characteristics of rare tool types are more clearly represented. Types 18, 19, and 20 (small, medium, and large triangular forms, respectively) all concentrate strongly in the 31-45 degree edge angle increment. These types, along with Type 1 (large quadrilateral

Figure 19 thus allows a preliminary basis .for separating relatively specialized forms from more generalized, presumably multipurpose forms. With the exception of Type 15, with its unusually high edge angles, tool forms that display multiple edge angle modes generally have a much wider range of edge angles than those whose edge angles distribute more like normal curves with single peaks. Tool forms such as thick and long-handled types, ~hich tend to have substantial edgetrimming, have single edge angle modes. There thus appears to be a trend for more finely-worked tools to have relatively specific edge angles and, presumably, functions. Another, slightly different way to display edge angle data is to divide edge angles into increments and crosstabulate the resulting edge angle groupings with formal tool types. In Table 9, 15 degree increments were chosen. Instead of differentiating single peak and bimodal distributions, this table tends to mask bimodal edge-angle distributions. Instead, however, an incremental cross-tabulation effectively displays the more general, relative edge-angle tendencies of particular types. Some general observations related to Table 9 are as follows: A majority of tool types (n=12) concentrate in the 31-45 degree edge angle increment. A few tool types, considered relatively finely-made and specialized by

ANDEAN PAST 5 (1998) forms) specimens, could not be characterized in terms of edge angle tendency until the data were placed into increments. Edge angles, whether examined, through cutpoint scatterplots or incremental crosstabulations, do not help characterize some tool types. Types 4 (large semi-lunar forms), 8 (faceted pointed forms), 9 (non-faceted pointed forms), 10 (large core tools), and 17 (thick pentagonal forms) are all rare tool types with a wide diversity of edge angles that display no modes from either type of analysis. Because of their edge diversity and relatively well-made forms, including much shaping and trimming, these five types are viewed on a morphological basis as multiple-function, relatively curated toqls. Tool angle, or the angle of the main portion of the stone that supports the edge, is an important attribute, particularly for its relationship to edge angle. Figure 20 plots tpis general relationship. While. this figure refers to alllithics as an assemblage and not to actual relationships within individual tools, it does show a general trend where, below the 35 degree threshold, edge angles are associated with higher tool angles. Above that 35 degree threshold, edge angles are associated with lower tool angles. The two frequency peaks at 30 and 40 degrees are on opposite sides of this threshold. Examination of the relationship between the two lines in Figure 20 suggests that two high frequency clusters of artifacts exist, one group with lower edge angles associated with relatively higher tool angles, oriented to cutting and slicing, and a second group with higher edge angles and relatively lower tool angles, oriented more to scraping and planing. In order to further investigate this dichotomy and the relationship between edge and tool angles in the industry, the combination attribute "tool edge type" was computed. Tool edge type is a direct measure of the relationship between edge and tool angles at the individual artifact level. Because it combines aspects of the lithic edge with the supporting structure of the tool, it potentially can aid in separating tools used for heavier versus lighter

-256 tasks, providing another means of crosschecking conjectures on tool function made from general morphology and lithic use-wear analysis. Five tool edge types were defined. An edge is considered "very heavy" when the edge angle measurement is at least 10 degrees greater than the tool angle. An edge is "heavy" when the edge angle is from 1 to 9 degrees greater than the tool angle. "Medium' edges" denote tools where the two measurements are equal. A "fine edge" is when the edge angle is 1 to 9 degrees lower than the tool angle. Finally, a "very fine" edge denotes a tool where the edge angle is 10 degrees or more lower than the corresponding tool angle. Fine edges would presumably be used for cutting and slicing tasks (such as food preparation), while heavy edges would be oriented toward scraping, planing, and adzing tasks (such as woodworking). The degree to which a tool type tends to a particular tool edge type may thus suggest whether the type was more generalized or specialized in its use. That is, a tool type with a relatively even distribution of fine, medium, and heavy tool edge types would be considered more generalized than a tool type with a strong tendency toward one tool edge type. Table 10 is a cross-tabulation of tool edge type with tool type. There is a general tendency for formal, edge-trimmed tools to have heavier edge types, regardless of whether their edge angles are generally high or low. Unmodified utilized flakes are dominated by medium edges. This contrast indicates a basic difference in edge characteristics between unmodified flakes and edge-trimmed flakes. Edge trimming tends to produce a stronger edge. Seven formal tool types that show clear tendencies toward heavier tool edge types are Types 1 (large quadrilateral forms), 3 (smap quadrilateral forms), 5 (medium semi-lunar forms), 6 (small semi-lunar forms), 13 (incurved use edges), 15 (blocky forms with pentagonal cross-sections), and 26 (long-handled, short use-edge forms). Three additional tool types display weaker tendencies toward heavy tool edge types. Types 2 (medium quadrilateral forms), 4 (large semi-lunar forms); and 7 (thick semi-lunar forms) show a wider range

257of tool edge types, although there are still more heavy and medium tool edges in these categories. Two rare tool types, Types 16 (thick rectangular forms) and 17 (thick pentagonal forms) display strong dominance of medium tool edge types. Type 12, the distinctive pentagonal form, is unique among common tool types in displaying an almost even distribution of heavy, medium, and fine tool type edges. On the basis of this combination attribute, Type 12 tools may be considered multifunctional. Three rare tool types, Types 18, 19, and 20 (small, medium, and large triangular forms, respectively) also display an even distribution of heavy, medium, and fine tool edge types. These triangular forms, which are usually less edge-trimmed than other formal types, may also have had more general, multipIe functions. Debitage Analysis Other extensive analyses of unifacial industries have not separated tool forms and have instead concentrated on detailed analysis of debitage, the waste flake by-products of tool production (e.g., Stothert 1974). In contrast, a basic premise of this analysis has been that formal types may be extracted based on edge trimming and flake shaping. The analysis presented here has thus concentrated on key attributes of flake tool forms. However, a large collection of debitage (n=32,215) was also recovered. The primary analytical use of debitage here is to utilize total frequencies as percentage components of the reduction sequence, in order to demonstrate the general intersite consistency of reduction, and to further relate the sites into the same cultural system. A few summary tables (Tables 11, 12) of debitage (with all sites lumped) add further insights into the non-randomness of this unifacial industry. These patterns are similar at the individual site level. Figure 21 shows the computer-scaled histograms of flake length and width with superimposed normal curves. These two dimensions all show relatively high concentrations, compared to a normal curve, of flake dimen-

Rossen: Nanchoc Lithic Tradition sions around specific millimeter values. It has often been stated that more expedient lithic industries based on unifaces are amorphous and thus more randomly (less systematically) worked than more curated industry types, particularly those based on bifacial reduction (Bamforth 1986; Binford 1979; Henry and Odell 1989; Johnson and Morrow 1987; Torrence 1989). In terms of Andean unifacial in~ dustries, this point was most strongly stressed by Stothert (1974, 1988). Despite this, the present analysis, including this debitage figure, suggests a relatively systematic lithic reduction technology to the point of producing debitage that concentrates around specific di. mensional modes. The observation that debitage does not distribute randomly in terms of dimensions may be related to the earlier observation that cores are not "amorphous" but "multiple form" in character. This statement was meant to explain that, although not prepared in the classic sense, cores are.not totally amorphous and random when subjected to detailed analysis. IIi terms of both cores and debitage, this unifaciallithic industry is relatively systematic in reducing cores, usually to exhaustion, and producing recurring tool flake forms and relatively uniform debitage. One way to further explore the above observation in regard to debitage is to examine flake type, an attribute based on the relationship between flake length (as measured from the platform or bulb of percussion) and flake width (as measured perpendicularly to flake length). Table 11 quantifies an observation made during analysis, that an unusually high percentage of debitage (31.7%) has larger flake widths than flake lengths. Short, wide flakes are struck in order to control the distant edge angle and form, while long, thin flakes are struck for control of lateral edges (Speth 1972). The relatively high percentage of short, wide flakes thus indicates concern for control of the reduction process and production of particular edge angles on flake tools. Table 12, a crosstabulation of flake type with material type, shows that this concern with edge angle control occurs within most major material types. Among the four most common material types, basalt, diorite, tuff, and rhyolite, the percentage of short,

ANDEAN PAST 5 (1998) wide flakes only varies from 28.4% to 34.2% of the total debitage. Flake type, in terms of relative length and width, provides another expression of the relatively careful nature of lithic tool production in the NLT despite its unifacial nature. DISCUSSION Characteristics of the Nanchoc Lithic Tradition It is possible to outline some of the important defining characteristics of the NLT. Table 13 is a summary of the descriptive analysis, in terms of traits assigned to the various formal tool types. A quick perusal of this table illustrates the mixture of common versus rare, expedient versus curated, and generalized versus specialized tools present in the industry. Variation in edge angle mode, edge angle trend, and tool edge type tendency is also presented. This table serves to display concisely the hidden complexity present in this apparently simple, unifacial lithic industry. In conceptual terms, the NLT may' be generalized and summarized in four ways: 1. Relative morphological stability and consistency The most important attribute-variables of the NLT, including stage of reduction, material type, and tool type display general consistency when compared across sites and levels. When crosstabulated, stable relationships exist between attribute-variables in terms of relative percentages, tendencies, and modes. This consistency and stability extends to both frequent and rare tool and material types. Previous analyses have depicted unifacial lithic industries as amorphous, irregular, and unsystematic instead of consistent, regular, and systematic. It appears that the NLT is more regular and systematic than other analyzed Andean unifacial industries such as Mongoncillo, Siches, and Las Vegas (Dillehay et al. 1992; Malpass 1983; Richardson 1978; Stothert 1974, 1988). ,Despite this relative regularity, it must still be recognized that almost half of all recovered utilized lithics are unmodified flakes that were probably briefly

-258 used and discarded. Even formal tool types have been loosely defined on the basis of form (and sometimes, size) and a substantial amount of variability is present within types. Furthermore, despite trends in edge angles of particular types, there is continually a 20% to 30% minority of specimens in a given type that do not conform to the edge angle tendency. It is thus appropriate to refer to the, NLT as being relatively morphologically stable and consistent (especially for a unifacial industry) but certainly not as standardized in the sense of containing extremely exact replication of tool types and edges (Laville et al. 1980). 2. Expediency The NLT contains traits of both expedient and curated lithic industries. However, in general, this industry better fits the conceptual characterization of an expedient industry. Traits such as (1) its exclusively unifacial character, (2) the local abundance of raw materials, (3) the lack of tool retouch and rejuvenation, and (4) the high percentage of simple utilized flakes all fit the classic definition of expedient industries (Binford 1977, 1979). Additionally, expediency in lithics is often associated with semi- or total sedentism, which is probable at these sites, given the type of dwelling construction and other artifactual data such as the heavy presence of ground stone and perhaps the botanical remains (e.g., Binford 1990; Parry and Kelly 1987; Yerkes 1989:184). Other industry traits such as (1) edgetrimmed tool types with recurring forms, (2) exhaustion of most cores, and (3) cores having recurring multiple form~ instead of being amorphous do not fit the classic criteria of expediency. A few heavily used, relatively fmely-made tool types, although a minority of this industry, are more curated than other industry tool types. This industry thus may not be as purely expedient as other unifacial industries previously described in the literature. In terms of the expedient-curated dichotomy, the greatest paradox in the NLT is the apparent combination of highly exhausted

259cores and material conservation with local aboodance of raw material. Dillehay has suggestedthat the present-day aboodance of basalt cobbles in the steep, dry creek beds adjacent to the research sites presents a false illusion of past aboodance. He suggested that the heavy vegetation and flowing water that were present during the sites' occupation period produced a lower real availability (Dillehay, personal communication 1990). Even with this caveat considered, the Nanchoc area also contained boulder outcrops of the most commonly used lithic materials, and must have been moderately aboodant in raw material. Experiments with locally available materials revealed that it is very difficult to remove flakes from streambed cobbles, which must first be thoroughly soo-dried for at least two days (Dillehay, personal commooication 1994).. The absence of cobble cortex in the NLT indicates that the apparent aboodance of streambed material was not utilized, but instead, reduced chunks were brought to. the sites from boulder fields a few kilometers away. Thus, while raw material was aboodant, it was not plentiful in the sense of immediate availability that encouraged total expediency. In the cultural situation being described here, people utilized various environmental zones and engaged in subsistence intensification while selectively accepting and rejecting outside influences.. In such a situation, a lithic technology with traits of both expediency and curation, not totally conforming to simplistic conceptual dichotomies, should be expected. The concepts of expediency and curation, each with its own lithic, economic, and environmental implications, must be refined through further evaluation against other large archaeological collections. While the term "expediency" is a generally useful concept for characterizing this industry, the exceptions to the classic criteria demonstrate the need for flexibility within the expedient-curated conceptual dichotomy. In the case of the NLT, expediency and curation are relative terms instead of an absolute dichotomy, and the true degree of expediency in the industry depends on what comparison is being made.

Rossen: Nanchoc Lithic Tradition 3. Localization, Embeddedness, and Sporadic Exchange Aboodant local materials, especially basalt and tuff, have very consistent patterns in terms of stage of reduction, distribution, and tool and edge characteristics. Other local materials such as diorite and rhyolite are slightly more reduced. As a lumped group, exotic materials are a very small percentage of the overall industry (1.1%) and this low percentage maintains remarkable consistency across different sites and stages of reduction. These exotic materials were treated, in terms of reduction sequence, similarly to local materials and are rather evenly distributed across sites despite the presence of intersite activity differences. While it is probable that some non-local lithic material was procured through sporadic trade or exchange, it is possible that some exotic silexes and quartzes exhibit a pattern of "embeddedness" or "el1}beddedprocurement", if Binford's criteria for that process are accepted (Bamforth 1986; Binford 1979). That is, some exotic materials may have been procured by sending out individuals or small groups to collect them, or materials were procured incidentally or opportunistically in the course of other travel-related procurement. This is opposed to receiving exotic goods in well-organized trade or exchange, where the exotic material is expected to be more reduced in a site and is thus more valuable than local materials, as well as associated more with particular tool types and activities (ibid.). The possible combination of embedded contacts and sporadic exchange to procure a relatively small amooot of non-local lithic material reinforces a characterization of the Nanchoc Culture as locally developed with relatively few outside contacts (Dillehay et a/. 1989, 1992). From the perspective of chipped stone artifacts, the Nanchoc Culture cannot be viewed as one of cultural mediators, large-scale exchangers, or seasonally mobile or transhumant
people. .

ANDEAN PAST 5 (1998) 4. More distinct from the North-Central Andean-Panamanian "Unifacial Complex" than in conformity with it When comparing the NLT with other industries of the so-called Andean-Panamanian "Unifacial Complex" such as Chiriqui, Aguazuque, Las Vegas, Sangay, Cubilan, Siches, and Mongoncillo, important differences appear under the scrutiny of detailed analysis (Ranere 1975; Linares and Ranere 1980; Correal 1989; Stothert 1974, 1988; Porras 1988; Temme 1982; Richardson 1969; and Malpass 1983 respectively). Some of these industries are similar in that they appear to be associated with forested environments (with Mongoncillo being the possible exception, although Malpass [1983:38] believes that the lower Casma Valley was also forested). Many of these industries may have been associated with woodworking and diverse,. generalized economies. A few may also have been associated with at least partly horticultural lifestyles (e.g., Pearsall 1994; Piperno,l990; Richardson 1969:115; Stothert 1988), although the presence and relative importance of gardening in each case has not .been established. Despite the similarities, important morphological lithic differences exist in terms of stage of reduction, tool morphology (especially trimming and shaping), aniount of ground stone present, and relative morphological consistency of the industries. Other researchers have been generally frustrated in their attempts to find consistency within individual "Unifacial Complex" industries. Richardson (1969) came closest with his notched and denticulate tool varieties of the Siches Industry, although these were loosely defined in terms of edge characteristics and accorded great morphological variability within types. Stothert (1974, 1988) was furthest from finding consistency, and instead emphasized flake attributes, the lack of prepared cores, and, industry irregularity within Las Vegas and other, later unifacial industries of the Santa Elena Peninsula of Ecuador. Malpass' (1983; n.d. [c. 1984]) formulation of the concept of a "Unifacial Complex" of genetically-related though spatially discontinuous industries drew heavily on the studies of

-260
these and other industries (e.g., Linares and Ranere 1980; Ranere 1975). The result was a generalized characterization of unifaciallithic industries as (1) morphologically inconsistent and amorphous, (2) lacking specialized forms, e~pecially edge-trimmed tools, and (3) lacking internal complexity. As a case study, the NLT provides an exception to these generalizations, perhaps because of intergroup differences in' economic and social organization. CONCLUSIONS Unifacial tools were characterized by Binford (1977, 1979) as expedient, in being quickly made and discarded, as opposed to curated tools that were carefully-made, maintained and rejuvenated through longer uselives. However, detailed analyses of the NLT suggest that this unifacial industry contained both relatively expedient and curated tools for both generalized and specialized purposes. Many unifacial indFstries are associated with forested or semi-forested environments, and with some level of emphasis on plant exploitation. Some are associated with at least incipient levels of plant cultivation, if only, as in Panama, as a supplement to a continuing hunting-gathering lifeway. However, the question of whether there is a true "Unifacial Complex" of industries ,spread throughout tropical America (Malpass n.d. [c. 1984]) remains unanswered for several reasons. Primarily, the industries have been analyzed from varying levels of systematic detail, and very few have been analyzed in terms of their spatial distribution and associations with other artifact assemblages. Certainly, there are superficial similarities inherent in their very existence as unifacial industries, sometimes within similar montane forests and often having Preceramic Period dates ranging from 9000-5000 B.P. Yet, under close scrutiny, these industries appear to be as different as a group of bifacial industries might be in a similarly-sized region such as the Archaic Period eastern United States (Jennings 1978). For example, some trimming has been described for the Siches Industry of far northern coastal Peru (Richardson 1969, 1978), but published and unpublished descriptions of the

261industry, along with personal examination of Siches collections housed at the Museo Nacional de Antropologia, Arqueologia e Historia, Lima, have satisfied me that Siches tools are larger, bulkier, and less carefully-shaped, with more denticulate forms and much less removal of dorsal ridges than are characteristic of the NLT. Other unifacial industries such as Las Vegas of coastal Ecuador and Cerro Mangote of Panama have been described as heavily dominated by small, crude, irregular, unshaped flakes and highly amorphous cores (McGimsey et al. 1986-87; Stothert 1988). In contrast, the NLT is presented here as as relatively morphologically systematic. Based on the emerging data, Malpass has backed away rom the idea of the unifacial complex and a single cultural distinctiveness of. preceramic Andean unifacial industries (Michael Malpass, personal communication 1995). Another broad issue involving the NLT and its relationships may be mentioned. Are unifacial industries truly. discontinuous in space? Malpass (1983) and Richardson (1969) suggested that they were widely-spaced, but this may be a function of where detailed re-

Rossen: Nanchoc Lithic Tradition tion. In particular, the associations between unifacial lithics and com pollen in Panama (Smith 1980) and incipient horticulture in Ecuador (Stothert 1988) support the findings at Nanchoc. Stothert, in her most extensive statement on the Las Vegas lithics, has also taken a social relations approach to early intensification that would push the concept back to the Preceramic Period based on technologi~ cal indicators, including unifacial lithics (Stothert 1988:254-255). This potential relationship between unifacial industries and early intensification should be monitored in the course of future research. Unifaciallithic industries can be designed as part of a shift to a more plant-oriented economy. Detailed analysis may isolate technological diversity in these industries that reflects the activity diversity of broad-spectrum economies that often appear during the transition to food production (Piperno 1989). It has been argued elsewhere in detail that the NLT represents one)ine of evidence of the subtle beginnings of localized intensification during the Middle Preceramic Period (Rossen 1991:667-674). The totality of this argument depends on local paleoenvironmental reconstruction and non-lithic data collected over the course of the Zaiia-Niepos Project, including ground stone, faunal, human remains, entymological, ethnohistoric, and to a lesser extent botanical and pollen data. These data strongly suggest that the Middle Preceramic people of Nanchoc resided in an ecotone between semiarid, brushy, and tropical rain forest zones and exploited a wide variety of plants and animals (Dillehay and Netherly 1985; Dillehay et al. 1989; Koepcke and Koepcke 1958; Rossen 1991: Chapter 6; Raimondi 1940 [1874]; Simpson 1975; Vuilleumier 1971; Weberbauer 1945). The variety of the unifaciallithic tools in the NLT (and associated non-lithic assemblages) suggests a shift toward a diversifying economy, while the dominance of local lithic materials, midden development, structures, and burials at the sites suggest a localized permanence of settlement (Rossen 1991:648-654). Andeanists studying the Preceramic Period have long noted the rapid cultural changes of

search has occurred.- Investigationof other


areas of South America increasingly suggests that unifacial industries are much more widespread than previously supposed (Correal 1989; Porras 1988; Schmitz 1987). Unifacial Lithics and Cultural Evolution A central issue is whether unifacial industries are part of a general trend away from hunting-oriented economies with bifacial projectiles like Paijan in northern Peru (Chauchat 1988), toward greater economic diversity, localized permanence, and in some areas, intensification. That is, may unifacial industries be considered an archaeological indicator of intensification, or at least the shift toward different kinds of human-plant relationships which eventually led to food production in some areas? In this regard, the NLT only represents a single case study of perhaps an atypically complex unifacial industry. Researchers, however, have increasingly begun to describe similar associations between unifacial lithics and resource diversification and intensifica-

ANDEAN PAST 5 (1998) the Late Preceramic Period, with its advent of large ceremonial architecture, based on maritime and agricultural sedentism (Quilter 1991). In some cases, it has also been suggested that the origins of the intensification process must have occurred in subtler technological shifts of the preceding Middle Preceramic Period (e.g., Quilter et ai. 1991). Bender (1978) has asserted that the process of intensification begins with the addition of more versatile technologies and new dietary options, together representing an increased "productivity", or the potential to produce more even if increased production per land unit is not achieved. If Bender's premise is accepted, then the Nanchoc sites with their varied unifaces, diversified economy, and localized permanence appear to represent one cultural situation where an early subtle form of intensification foreshadowed the more dramatic cultural changes of the Late Preceramic Period. . Acknowledgements This paper builds on some twenty-live years of research in the Zafia Valley by the Zafia-Niepos Project,.founded and nurtured by Tom D. Dillehay and Patricia J. Netherly. The paper benefitted from numerous discussions with Tom, Patricia, William Y. Adams, Ramiro . Matos, Duccio Bonavia, and Peter Kaulicke. The thoughtful review comments of Karen E. Stothert were greatly appreciated. The lithics were drawn by Jimmy A. Railey. Analysis of the Nanchoc lithics was supported by a National Science Foundation dissertation improvement grant and the University of Kentucky Center for Computational Science. References Cited
Adams, William Y. and Ernest W. Adams 1991 Archaeological Typology and Practical Reality: A Dialectical Approach to Artifact Classification and Sorting. Cambridge: Cambridge University Press. Ahler, Stanley A. 1989 Mass Analysis of Flaking Debris: Studying the Forest Rather Than the Trees. In Alternative Approaches to Lithic Analysis, edited by Donald O. Henry and George H. Odell, pp. 85-118. Washington D.C.: Archaeological Papers of the American Anthropological Association.

-262
Bamforth, Douglas B. 1986 Technological Efficiency and Tool Curation. American Antiquity 51(1):38-50. Bender, Barbara 1978 Gardener-Hunter to Farmer: A Social Perspective. WorldArchaeology 10:204-222. Binford, Lewis R. 1977 Forty-seven Trips: A Case Study in the Character of Archaeological Formation Processes. In Stone Tools as Cultural Markers: Change, Evolution and Complexity, edited by R.V.S. Wright, pp., 24-36. Atlantic Heights, NJ: Humanities Press, Inc. 1979 Organizational and Formation Processes: Looking at Curated Technologies. Journal of Anthropological Research 35:255-273. 1980 Willow Smoke and Dogs' Tails: HunterGatherer Settlement Systems and Archaeological Site Formation. American Antiquity 44(3):4-20. 1987 Researching Ambiguity: Frames of Reference and Site Structure. In Method and Theoryfor Activity Area Research: An Ethnoarchaeological Approach, edited by Susan Kent, pp. 449-512. New York: Columbia University Press. 1990 Mobility, Housing, and Environment: A Comparative Study. Journal of Anthropological Research 46(2): 119-152. Boisvert, Richard A., B.N. Driskell, K.W. Robinson, S.D. Smith and L.F. Duffield 1979 Materials Recovered. In Excavations at Four Archaic Sites in the Lower Ohio Valley, Jefferson County, Kentucky, Volume 1, edited by Michael B. Collins, pp. 60-329. Lexington: Occasional Papers in Anthropology 1, Department of Anthropology, University of Kentucky. Bradley, B. A. 1972 Lithic Reduction Sequences: A Glossary and Discussion. In Lithic Technology: Making and Using Stone Tools, edited b~ E. Swanson, pp. 5-13. The Hague: Mouton. Brew, John O. 1946 The Uses and Abuses of Taxonomy. In The Archaeology of Alkali Ridge, Southeastern Utah, edited by J.O. Brew, pp. 44-66. Papers of the Peabody Museum 21. Cambridge, Massachusetts. Chauchat, Claude 1975 The Paijan Complex, Pampa de Cupisnique, Peru. Nawpa Pacha 13:85-96. Berkeley. 1978 Additional Observations on the Paijan Complex. Nawpa Pacha 16:51-65. Berkeley. 1988 Early Hunter-Gatherers on the Peruvian Coast. In Peruvian Prehistory, edited by Richard W. Keatinge, pp. 41-65. Cambridge: Cambridge University Press. Collins, Michael B. 1975 Lithic Technology as a Means of Processual Inference. In Lithic Technology, edited by E. Swanson, pp. 15-34. The Hague: Mouton. Correal U., Gonzalo 1989 Aguazuque: Evidencias de cazadoresrecolectores y plantadores en la altiplanicie de la Cordillera Oriental. Bogota: Fundaci6n de Inves-

263tigaciones Arqueol6gicas Nacionales, Banco de la Republica.


Crabtree, Don E.

Rossen: Nanchoc Lithic Tradition


Ford, James 1954 On the Concept of Types. American Anthropologist 56:42-53. Gardin, Jean-Claude 1984 Archaeological Constructs: An Aspect of Theoretical Archaeology. Cambridge: Cambridge University Press. Gould, Richard A., D. Koster, and A. Sontz 1971 The Lithic Assemblage of the Western Desert Aborigines of Australia. American Antiquity 36:149-169. ' Hassan, Fekri A. 1987 Re-Forming Archaeology: A Forward to Natural Formation Processes and the Archaeological Record. In Natural Formation Processes and the Archaeological Record, edited by David T. Nash and Michael D. Petraglia, pages 1-10. B.A.R. International Series 352. Oxford, England. Henry, Donald O. and George H. Odell, editors 1989 Alternative Approaches to Lithic Analysis. Washington D.C.: Archaeological Papers of the American Anthropological Association 1. Hyslop, John 1984 The Inka Road System. New York: Academic Press. Jennings, Jesse D., editor 1978 Ancient Native Americans. San Francisco: W.H. Freeman. Johnson, Jay K. and Carol A. Morrow, editors 1987 The Organization of Core Technology. Boulder: Westview. Koepcke, Maria 1954 Corte ecol6gico transversal en los Andes del Peru central con especial consideraci6n de las aves. Memorias, Museo de Historia Natural "Javier Prado" 3. Lima. Koepcke, Hans W., and Maria Koepcke 1958 Los restos de bosque~ en las vertientes occidentales de los Andes peruanos. Boletin del Comire Nacional de Proteccion a la Naturaleza 16. Lima. Krieger, A.D. 1944 The Typological Concept. American Antiquity 9:217-288. LaVallee, Daniele, Michele Julien, Jane Wheeler, and Claudine Karlin 1985 Telarmachay: Chasseurs et Pasteurs Prehistorique des Andes. Paris: Editions Recherche sur les Civilisations. Laville, Henri, Jean-Philippe Rigaud, and James R. Sackett 1980 Rock Shelters of the Perigord: Geological Stratigraphy and Archaeological Succession. New York: Academic Press. Linares, Olga F. and Anthony J. Ranere 1980 Adaptive Radiations in Prehistoric Panama. Peabody Museum Monographs 5. Cambridge, Massachusetts: Peabody Museum. Malpass, Michael A. 1983 The Preceramic Occupations of the Casma Valley, Peru. Unpublished Ph.D. dissertation, Department of Anthropology, University of Wisconsin-Madison.

1973 The Flintknapper's Raw Materials. In In


Search of Man: Readings in Archaeology, edited by Ernestene Green, pp. 230-234. Boston: Little, Brown and Co. Craig, Alan K.

1985 Cis-Andean Environmental Transects: Late Quaternary Ecology of Northern and Southern Peru. In Andean Ecology and Civilization: An Interdisciplinary Perspective on Andean Ecological Complementarity, edited by S. Masuda, I. Shimada, and C. Morris, pp. 23-44. Papers from the Wenner-Gren Foundation for Anthropological Research Symposium 91. Tokyo: University of Tokyo Press. Dillehay, Tom D. 1985. A Regional Perspective of Preceramic Times in the Central Andes. Reviews in Anthropology 12(3):193-205. Dillehay, Tom D., Gerardo Ardila Calder6n, Gustavo Politis, and Maria da Conceicao de Moraes Coutinho Beltrao 1992 Earliest Hunters and Gatherers of South America. Journal of World Prehistory 6(2):145204. Dillehay, Tom D., Herbert Eling, and Jack Rossen 1990 Informe de investigaci6n arqueol6gica del Proyecto Zafia-Niepos: la campafia de 1989.' Report submitted to the Instituto Nacional de Cultura, Lima. Dillehay, Tom D. and patrjcia J. Netherly 1983 Exploring the Upper Zafia Valley in Peru: A Unique Tropical Forest Setting Offers Insights into the Andean Past. Archaeology 36(4):22-30. 1985 Informe deinvestigaci6n arqueol6gica en el Valle de Zafia: campafia de 1984-1985. Report submitted to the Instituto Nacional de Cultura, Lima. Dillehay, Tom D., Patricia J. Netherly, and Jack Rossen 1989 Middle Preceramic Public and Residential Sites on the Forested Slope of the Western Andes, Northern Peru. American Antiquity 54(4):733-759. Dillehay, Tom D., Jack Rossen, and Patricia J. Netherly 1992 Ocupaci6n del Preceramico medio en la zona alta del valle de Zafia: inovaci6n 0 aculturaci6n? In Estudios de Arqueologia Peruana, edited by Duccio Bonavia, pp. 69-82. Lima: Fomciencias. Flannery, Kent V. 1969 Origins and Ecological Effects of Early Domestication in Iran and the Near East. In The Domestication and Exploitation of Plants and Animals, edited by P.J. Ucko and G.W. Dimbleby, pp. 73-100. Chicago: Aldine. Flenniken, J. Jeffrey 1985 Stone Tool Reduction Techniques as Cultural Markers. In Stone Tool Analysis: Essays in Honor of Don E. Crabtree, edited by Mark G. Plew, James C. Woods, and Max G. Pavesic, pp. 91-131. Albuquerque: University of New Mexico Press.

ANDEAN PAST 5 (1998)


n.d. [c. 1984] Early Lithic Technologies of Coastal Peru. Unpublished manuscript in possession of the author. Matos M., Ramiro 1975 Prehistoria y ecologia humana en las punas de Junin. Revista del Museo Nacional 41:37-74. Lima. McGimsey, Charles R. 1986-1987 Cerro Mangote and its Population. Journal of the Steward Anthropological Society 16(1 and 2):125-157. Morrow, Carol A. and Richard W. Jefferies 1989 Trade or Embedded Procurement?: A Test Case trom Southern Illinois. In Time, Energy and Stone Tools, edited by Robin Torrence, pp. 27-33. Cambridge: Cambridge University. Moseley, Michael E. 1983 The Good Old Days Were Better: Agrarian Collapse and Tectonics. American Anthropologist 85(4):773-799. Mueller, James W., editor 1976 Sampling in Archaeology. Tucson: University of Arizona Press. Nash, David T. and Michael D. Petraglia, editors 1987 Natural Formation Processes and the Archaeological Record. B.A.R. International Series 352, Oxford, England. Netherly, Patricia J. and Tom D. Dillehay 1985 Domestic and Public Contexts at the Cementerio de Nanchoc Site: RefIning Preceramic Public and Domestic Production. Paper presented at the 50th annual meeting of the Society for American Archaeology, Denver, May 1-5. 1986 Duality in Public Architecture in the Upper Zafia Valley, Northern Peru. In Perspectives on Andean Prehistory, edited by Daniel H. Sandweiss and D. Peter Kvietok, pp. 85-114. Ithaca: Cornell University Latin American Studies Program. Parry, William J. and Robert L. Kelly 1987 Expedient Core Technology and Sedentism. In The Organization of Core Technology, edited by Jay K. Johnson and Carol A. Morrow, pages 285304. Boulder: Westview'. Pearsall, Deborah M. 1994 Issues in the Analysis and Interpretation of Archaeological Maize in South America. In Corn and Culture in the Prehistoric New World, edited by Sissel Johannessen and Christine A. Hastorf, pp. 245-272. Boulder: Westview. Petraglia, Michael D. and David T. Nash 1987 The Impact of Fluvial Processes on Experimental Sites. In Natural Formation Processes and the Archaeological Record, edited by David T. Nash and Michael D. Petraglia, pp. 108-131. B.A.R. International Series 352. Oxford, England. Piperno, Dolores R. 1989 Non-affluent Foragers: Resource Availability, Seasonal Shortages, and the Emergence of Agriculture in Panamanian Tropical Forests. In Foraging and Farming: The Evolution of Plant Exploitation, edited by David R. Harris and Gordon C. Hillman, pp. 538-554. One World Archaeology 13. London: Unwin Hyman.

-264
1990 Aboriginal Agriculture and Land Usage in the Amazon Basin, Ecuador. Journal of Archaeological Science 17:665-677. Porras G., Pedro Ignacio 1988 Investigaciones arqueo/6gicas alas faldas del Sangay: tradici6n Upano. Quito: Centro de Investigaciones Arqueol6gicas, Pontificia Universidad Cat6lica del Ecuador. Price, T. Douglas 1981 Complexity in "Non-complex" Societies. In Archaeological Approaches to the Study of Com-' plexity, edited by Sander E. van der Leeuw, pp. 5497. Amsterdam: Universiteit van Amsterdam. Quilter, Jeffrey 1991 Late Preceramic Peru. Journal of World Prehistory 5(4):387-438. Quilter, J., B. Ojeda E., D.M. Pearsall, D.H. Sandweiss, J.G. Jones, and E.S. Wing 1991 Subsistence Economy of El Paraiso, an Early Peruvian Site. Science 251:277-283. Raimondi, Antonio 1940 [1874] El Peru, Volume 1. Lima: Escuela Tipografica Salesiana. Ranere, Anthony 1. 1972 Early Human Adaptation to New World Tropical Forests: The Viewfrom Panama. Unpublished Ph.D. dissertation, Department of Anthropology, University of California-Davis. 1975 Toolmaking and Tool Use among the Preceramic Peoples of Panama. In Lithic Technology: Making and Using Stone Tools. Edited by Earl Swanson. The Hague: Mouton. Richardson, James B. 1969 The Preceramic Sequence and Pleistocene and Post-Pleistocene Climatic Change in Northwestern Peru. Unpublished Ph.D. dissertation, Department of Anthropology, University of Illinois-UrbanaChampaign. 1978 Early Man on the Peruvian North Coast, Early Maritime Exploitation and the Pleistocene and Holocene Environment. In Early Man in America from a Circum-Pacific Perspective, edited by Alan Bryan, pp. 274-289. Occasional Papers 1, Department of Anthropology, University of Alberta, Edmonton, Canada. 1981 Modeling the Development of Sedentary Maritime Economies on the Coast of Peru: A Preliminary Statement. Annals of the Carnegie Museum 50(5):139-150. Rick, John W. 19S0 Prehistoric Hunters of the High Andes. New York: Academic Press. Rossen, Jack 1988 Informe de investigaci6n arqueo16gica en el valle de Zafia: la campafia de 1987. Report submitted to the Instituto Nacional de Cultura, Lima. 1989 The Nanchoc Culture: Preceramic Horticulturalists of the Upper Zafia Valley. Paper presented at the annual meeting of the Society for American Archaeology, Atlanta. 1991 Ecotones and Low-Risk Intensification: The Middle Preceramic Habitation of Nanchoc, Northern Peru. Unpublished Ph.D. dissertation, De-

265pamnent of Anthropology, University of Kentucky, Lexington. Rossen, Jack and Tom D. Dillehay 1994 Spatial and Temporal Transitions in the Zafia Valley, Northern Peru (8000-4000 B.P.). Paper presented at the annual meeting of the Society for American Archaeology, Minneapolis. Rossen, Jack, Tom D. Dillehay, and Donald Ugent 1996 Ancient Cultigens or Modem Intrusions?: Evaluating Archaeological Plant Remains in an Andean Case Study. Journal of Archaeological Science 23:391-407. Rouse, Irving 1939 Prehistory in Haiti: A Study in Method New Haven: Yale University Publications in Anthropology 21. 1960 The Classification of Artifacts in Archaeology. American Antiquity 25:313-323. Schick; Kathy D. 1987 Experimentally-Derived Criteria for Assessing Hydrologic Disturbance of Archaeological Sites. In Natural Formation Processes and the Archaeological Record, edited by David T. Nash and Michael D. Petraglia, pp. 86-107. B.A.R. Interna. tional Series 352. Oxford, England. Schiffer, Michael B. 1976 Behavioral Archaeology. New York: Academic Press. 1987 Formation Processes of the Archaeological Record Albuquerque: University of New Mexico Pre~. . Schmitz, Pedro Ignacio 1987 Prehistoric Hunters and Gatherers of Brazil. Journal of WorldPrehistory 1(1):53-126. Simpson, Beryl B. 1975 Pleistocene Changes in the Flora of the High Trop-icalAndes. Paleobiology 1(3):273-294. Smith, C. Earle 1980 Plant Remains &om Guitarrero Cave. In Guitarrero Cave: Early Man in the Andes, edited by Thomas F. Lynch, pp. 87-119. New York: Academic Press. Speth, John D. 1972 The Mechanical Basis of Percussion Flaking. American Antiquity 37:34-60. Stockton, E.D. 1977 Taxonomy at the Service of Prehistory. In Stone Tools as Cultural Markers: Change, Evolution and Complexity, edited by R.V.S. Wright, pp. 340-344. Atlantic Heights, NJ: Humanities Press, Inc. Stothert, Karen E. 1974 Lithic Technology of the Santa Elena Peninsula. Unpublished Ph.D. dissertation, Depamnent of Anthropology, Yale University. 1985 The Preceramic Las Vegas Culture of Coastal Ecuador. American Antiquity 50(3):613-637. 1988 La prehistoria temprana de la peninsula de Santa Elena, Ecuador: Cultura Las Vegas. Miscelanea Antropol6gica Ecuatoriana, Serie Monografica 10. Guayaquil: Museos del Banco Central del Ecuador.

Rossen: Nanchoc Lithic Tradition


Temme, Mathilde 1982 Excavaciones en el sitio preceramico de Cubilan (Ecuador). Miscelanea Antropo16gica
Ecuatoriana 2:136-163. Torrence, Robin, editor 1989 Time, Energy and Stone Tools. Cambridge: Cambridge University Press. Trujillo, Diego de 1948 [1571] Relacion del descubrimiento del Reyno del Peru, edited by R. Porras Barrenechea. Sevilla:

Americanos. Uceda C., Santiago 1986 Le Paijanien de la Region de Casma (perou): Industrie Lithique et Relations avec les Autres Industries Preceramiques. Unpublished Ph.D. dissertation, L'Universite de Bordeaux, France. 1987 El Paijanense en la Region de Casma. Trujillo, Peru: Universidad Nacional de Trujillo, Facultad de Ciencias Sociales. Vaughan, Patrick A. 1985 Use-Wear Analysis of Flaked Stone Tools. Tucson: University of Arizona Press. Vuilleumier, Beryl Simpson 1971 Pleistocene Changes in the Fauna and Flora of South America. Science 173:771-780. Weberbauer, Augusto 1945 El mundo vegetal de los Andes peruanos. Lima: Ministerio de Agricultura. White, Peter, N. Modjeska, and I. Hipuya 1977 Group Definitions and Mental Templates: An Ethnographic Experiment. In Stone Tools as Cultural Markers: Change, Evolution and Complexity, edited by R.V.S. Wright, pp. 380-390. Atlantic Heights, NJ: Humanities Press, Inc. Yerkes, Richard W. 1989 Lithic Analysis and Activity Patterns at Labras Lake. In Alternative Approaches to Lithic Analysis, edited by Donald O. Henry and George H. Odell, pp. 183-212. Washington D.C.: Archaeological Papers of the American Anthropological

Imprenta de la Escuela de Estudios Hispano~

A~ociation 1.

ANDEAN PAST 5 (1998)

-266

Figure 1. Location of upper Zaiia Valley, northern Peru (after Dillehay et al. 1989).

267-

Rossen: Nanchoc Lithic Tradition

"Down Valley to Qyot~n

3km

Figure 2. Map of Nanchoc area, showing south bank lateral quebradas. Dots representPreceramicsites. Site CA09-04is Cementeriode Nanchoc site, across river from Quebradade Las'
Pircas (after Dilleh~y et al. 1989). .

ANDEAN PAST 5 (1998)

-268

Figure 3. October 1969 (dry season) air photo of a portion of the Nanchoc Valley, including the (outlined) Quebrada de Las Pircas. (Photo courtesy Servico Aerofotografico Nacional [SAN].)

269'"

Rossen: Nanchoc Lithic Tradition.

Figure 4. Blow-upof outlinedarea in figure 3, showingOctober 1969 (dry season) air photo of the Quebrada de Las Pircas. Note the Nanchoc Valley floor at bottom of photo. Circles denote the locations of investigated sites. Dotted line denotes rechanneling of main rocky streambeddue to heavy El Nino rains of 1983-1984(Photo courtesy of SAN).

ANDEANPAST 5 (1998)

-270

Figure S. Schematicdiagram of the Quebrada de Las Pircas, with relative locations of the investigatedsites.

271-

Rossen: Nanchoc Lithic Tradition

Figure 6. North wall profile of North Block, site CA09. Note the increasing depth of eolian deposits on block's eastern side, toward the center of the site. Numbers denote the stratigraphy of (1) hard colluvial cap, (2) soft eolian deposits, and (3) sterile alluvial subsoil.

ANDEAN PAST 5 (1998)

-272

'" entrance

. . .

. .

... .

.
(j) <>

. .
1m .

S 1W64

I
dirt floor

Bedrock

outcrops

Rock

. _GrOundstone
adobes

Complete Posthole

Lizard
Hard-packed

Figure 7. Quincha hut of site CA09-27.

273-

Rossen: Nanchoc Lithic Tradition

RAW MATERIAL

Product

B~roduct

Primary Flakes

I
STAGE 1:
CORES, CORE TOOLS (rare)
., Primary Flakes, Exhausted Cores

,
STAGE 2: UNMODIFIED SECONDARY FLAKES (Sometimes Utilized)

Tertiary

Flakes

,
STAGE 3: TRIMMED SECONDARY FLAKES of Formal Tools)

(Proliferated

Stage

Figure 8. Schematic representation of the Nanchoc Lithic Tradition. (NLT) reduction sequence.

275-.

Rossen: Nanchoc Lithic Tradition

Figure 10. Type 7: Thick semi-lunarFlake Tools (n=57)


~ ~ ~ ~ ~

larger than Type 4 tools. made of large secondaryand occasionallyprimary flakes. parallel thinning flakes sometimesremoved along dorsal surface. relatively steep tool and edge angles. sometimesbroken work edge indicates heavy-duty use.

Metrics:
Dimensionsin millimeters Length Width Thickness Mean 49.9 29.6 12.2 Mode 43 22 10 Standard deviation 9.9 8.4 3.2

ANDEANPAST S (1998)

-276

Figure 11. Type 8: multi-facetedpointed secondaryflake tools (n=4)


~
~

rare.

pointed form with relatively thick, diamond-shaped,multi-facetedpoint (thicker than


remainder of tool). point shows ,pronounced wear.

Metrics:
Dimensionsin millimeters Mean 41.5 Mode Standard deviation

Width Thickness

Length

29.6
12.2

14.1
8.8

4.0

277-

Rossen: Nanchoc Lithic Tradition

'.

Figure 12. Type 10: core tools (n=8)


~

forms range from simplequadrilateralblocks with one or a few primary flakes removed
to more complex forms with several s~ndary flakes removed.

~
~

often with long, even pointed edge. has several use-edges.

Metrics:
Dimensions in

Mean 102.1 56.6

Mode 102

millimeters Length Width

Standard deviation 19.9

Thickness

40.0

31.0

9.9

ANDEANPAST S (1998)

-278

Figure 13. Type 10 core tool.

279-

Rossen: Nanchoc Lithic Tradition

Figure 14. Type 12: pentagonalsecondaryflake toolS(n=131)


~ ~

distinctivepentagonaloutline, often elongated. shaped by removing flakes along two to three edges, including both use and non-use
edges (this distinguishes this type from pentagonal variant of Types 1 to 3, along with use-edge location).

use-edge usually on upper left side of pentagon; (dorsal side up, striking platform at
bottom).

shallow edge angles.

Metrics:
Dimensions in

Mean 38.4 29.2 10.7

Mode 34 26 10

millimeters Length Width Thickness

Standard deviation 8.1 6.2 3.5

ANDEANPAST 5 (1998)

-280

Figure 15. Type 13: incurvedor 'notchedsecondaryflake tools (n=86)


~ ~ ~

appear in several forms, usually irregular and amorphous. have distinctiveconvex, incurved use-edgeor often, multiple use-edges. steep edge and tool angles.

Metrics:
Dimensions in

Mean 41.6 30.2 12.6

Mode 38 23, 31 8

millimeters Length Width Thickness

Standard deviation 10.1 9.7 4.7

281-

Rossen: Nanchoc Lithic Tradition

Figure 16. Cores of the NLT, including two exhausted specimens (top) and one non-exhausted specimen (bottom).

ANDEANPAST 5 (1998)

-282

Figure 17. Idealized schematic cross-sections of tool edge types: edge angle greater than tool angle forms a heavy edge (top); equal edge and tool angles form a medium edge (middle); edge

angle less than tool angle forms a fine edge (bottom).

283-

Rossen: Nanchoc Lithic Tradition

frequency

350 300 250 200 150 100 50

angle (degrees)
0,

10

20

.30

40

50

60

70

80

90

100

Figure 18. Line graph of edge angle frequenciesin the NLT.

ANDEAN PAST 5 (1998)

-284

++----+----+----+----+----+----+----+----+----+----+----+----+----+----+----+----++
100+ I I I I 80+ I I I I & D G & 60+ I I I I A N G L & 40+ I 1 I I 20+ I 1 I I 0+ 1 1 1 3 1 1 1 4 1 5 1 4 1 3 1 1 3 1 5 3 1 2- 1 2 1 1 1 1 1 .. 3 2 2 2 1 1 1 2 1 1 1 2 4 2 1 3 1 3 1 1 1 3 1 4 1 4 3 3 2 3 4 2 2 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 3 1 3 5 4 4 2 3 1 2 1 2 1 1 1 1 1 1 1 1 1 1 4 1 4 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 2 1 1 2 3 3 1 1 2 3 1 1 1 1 1 1 1 1 1 2 4 5 7 1 7 * 1 7 1 6 1 1 1 1 1 1 1 2 2 2 2 2 1 2 1 1 3 1 3 1 2 2 1 1 1 1 1 1 1 1 1 2 1 1 + I I I I + I I I I + I I I I + I I I 1 + I

I I I + 28

++----+----+----+----+----+----+----+----+----+----+----+----+----+----+----+----++
4 8 12 16 20 24

TOOL

T Y l' &

I I I I I I I
I I I I

2515 cases plotted

Cutpolnt-S

I 1 1 I 1 I
I I I I I

1- 5 - 1

26-30

- 6

6-10
11-15

":' 2 - 3

31-35 - 7 36-40 41-45 >45

8
*

16-20 21-25

4
5

- 9

Figure 19. Edge angles plotted against formal tool types, with a cutpoint of 5. Unmodified utilized flakes (Type 14) have been omitted due to their high frequencies. Clina! versus multiplepeak angle distributions are one method of discerning single from multiple function tools. Type 8 are multi-faceted pointed tools (boons) without true use edges, while Types 11,22,23, and 24 are unassigned numbers.

285-

Rossen: Nanchoc Lithic Tradition

frequency 450 400 350 300 250 200 150 100 50 0 o. Tool angle Edge angle

.
10 20 30

40

50

60

70

80

90

100

Figure 20. Edge and tool angle frequenciesof the NLT.

ANDEAN PAST 5 (1998)

-286

COUNT

MIDPOINT (mml

ONE

SYMBOL

ZQUALS

APPROXIMATELY

80 OCCURRENCES

20 1901 3450 3778 3857 2030 1428 1091 523 307 208 100 SS 38 16 9 6 6

o o 2

4.33 7.67 _fr :...... 11.00 * :..---.....--14.33 **************************************1******** 17.67 :...... 21.00 .-.--.-.....-.-....----.24.33 *_._..-.-...._._.27.67 *------_.-._-, 31.00 34.33 -*:37.67 1** 41.00 * 44.33 * 47.67 51.00 54.33 57.67 61.00 64.33 67.67 71.00
1 +

1 + 1 + 1 800 1600 2400 HISTOGRAM FREQUENCY

3200

4000

COUNT o
2 1089 7110 5736 2847 1242 527 178 66 23 4

MIDPOINT

ONE

SYMBOL

EQUALS

APPROXIMATELY

160

OCCURRENCES

(-I o
5 * 10 .**..**** 15 ************************************. 20 *...._._._._----_. Ir._ 25 * 30 *1* 35 * 40 45 50 55 60 65 70 75 80 85 90 95 1 + 1 + 1 + 1 + 1 o 1600 3200 4800 6400
..

****:*...............

o
1

o o o o o
1 o

1 8000

HISTOGRAM

FREQUENCY

Figure 21. Computer-scaled histogramsof flake length and width.

287-

Rossen: Nanchoc Lithic Tradition

Table 1. Radiocarbondates from the Quebradade Las Pircas.

Site

Provenience

Depth

Uncorrected date (Before Present)

Lab t

CA09-27

1985 test unit N 1 W 75, level 3 N 1 W 62, level 3 dwelling floor N 2 W 63, level 4 beneath floor 3 W 5, burial 1 pit bottom S 1 E 1, level 4

28 em

7950 +/-180

Beta12385

CA09-27

25 em
32 em

7630 +/-80

Beta30778

CA09-27

7690 +/-70

Beta30779

CA09-28

36 em
39 em

8210 +/-180*

Beta33523

CA09-28

8260 +/-130

Beta33524

CA09-52 CA09-52 CA09-52

N 10 W 4, level 2 N 10 W 6, level 4 1985 test unit N 22 W 10, level 3 S 13 W 7, level 4

17 em 33 em 27 em

8080.+/-70 7850 +/-140* 7920 +/-120

Beta30781 Beta33525 Beta12384

CA09-52

38 em

2270 +/-70

Beta30780

CA09-85

Test unit 1, south 1/2, level 4

36 em

8410 +/-140*

Beta33526

* denotes extended eount date (small samples)

ANDEANPAST S (1998) Table 2. List of formal tool types of NLT.

-288

large quadrilateral med~um quadrilateral small quadrilateral large semilunar med~um semilunar small semilunar thick semilunar multi-faceted pointed simple pointed large core ~entagonal ~ncurved use edge unmodified (untrimmed) utilized blocky thick rectangular thick pentagonal small triangular . 19 medium triangular 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18
20 21 25 26
1

large triangular amorphous 1 amorphous.2 long-handled,-short

use edge

289-

Rossen: Nanchoc Lithic Tradition

Table 3. Frequencies and percentages of chipped lithic raw material types (alllithics, all sites).

Material

Color

Mat Code

Frequency

Percent

basalt quartzite andesite *silex *silex *silex *silex limestone *silex rhyolite quartz *quartz *quartz *quartz *chalcedony
*silex

*silex qran!te mylonite *jasper *silex diorite qranite basalt *quartz *silex copper ore *jasper *silex *mica *silex *silex ..

black violet gray banded cream coffee dark qreen off-white qenera1 banded milky crystaline crystal var. colored white black black w coffee qray qray, veined
black w 'red

1 2 3 4 5 6 7 8 9 10 11
-1.2

13 14 15 16 17 18

1
20 21 22 23 24 25 26 28 29 30 31 32 33

qray cream, off-white green black, white flecks amber qray, red streaks dark qreen cream, black & red dark pink, violet silver maroon

black

yellow

16520 353 2658 41 25 13 70 2788 3 2298 88 50 2 48 24 67 1 104 4572 33 35 5576 1 7 3 11 126 16 1 1 3 1 -------

46.5

1.0
7.5 .1 .1 .0 .2 7.8 .0 6.5 .2 .1 .0 .1
.

.1

.2 .0 .3 12.9 .1 .1 15.7 .0 .0 .0 .0 .4 .0 .0 .0 .0 .0

------100.0

Total

35537

* denotes

material

type considered

exotic.

ANDEAN PAST 5 (1998)

-290

Table 4. Frequencyand percentagesof formal tool types (all sites).

Type Description

Type

Frequency

Percent

large quadrilateral medium quadrilateral small quadrilateral large semi-lunar medium semilunar small semi-lunar thick semi-lunar multi-faceted pointed non-faceted pointed core tools pentagonal incurved utilized unmodified blocky elongated rectangular large pentagonal small triangular medium triangular lar9 triangular amorphous1. amorphous2 small elongated biface

1 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19 20 21 25 26 27

25 148 70 28 120 82 57 4 8 8 131 86 1266 284 28 13 15 31 5 66 10 96 1 -------

1.0 5.7 2.7 1.1 4.6

3.2
2.2 .2 .3 .3 5.1 3.3 49.0 11.0 1.1 .5 .6 1.2 .2 2.6 .4 3.7 .0

-------

Total

2582

100.0

291-

Rossen: Nanchoc Lithic Tradition

Table S. Frequenciesand percentagesof core types (all sites).

Core Type

Code

Frequency

Percent

not exhausted, <10% cortex not exhausted, 10-40% cortex not eXhausted, 40-60% cortex not exhausted, 60-90% cortex exhausted, <10% cortex exhausted, 10-40% cortex exhausted, 60-90% cortex exhausted, >O% cortex

0 1 2 3 5 6 7 8

------772

59 77 19 5 376 195 40 1

------100.0

7.6 10.0 2.5 0.6 48.7 25.3 5.2 0.1

Total

ANDEAN PAST 5 (1998) Table 6. Core types crosstabulatedwith raw material types.

-292

CORE TYPE: COUNT ROIfPCTI 1 I


:::::~AL

I NOT 01 EXHAUSTD 11 21

I I 1 31 67 51 61 71 81 EXHAUSTD ROlf TOTAL

~-;
1 +

;~--:
7.6 9 7.4 I + I I
+ I I 1 3.2 I I

~:--:
10.4 16 13.2 I + I I
+ 1 I 3 9.7 I I 16.1

:--:
1.8 2 1.7 I + I I
+ I I 5 I 1

andesite

I I
+ I I

1 .8

m
51.1 I I
I I I I 22.6

---~:--: ~--;
22.6 35 28.9
4 26.7 13 41.9

:
I I
+ 1 I 1 I

327

I + I I
+ I I I I

6.4 7 5.8
1 6.7 2 6.5

1 + I I
+ I I 1 I

I 42.5 +
121 15.7 15

51 42.1
10

I I
I

exotics

66.7 7

I I 1

+--------+--------+--------+--------+--------+--------+--------+--------+
11l11estone 8 I I 31

1.9

+--------+--------+--------+--------+--------+--------+--------+--------+
rhyolite 10 1 I I I I I I I I I 6 8.5 I I 6 8.5 11 10.1 4 5.6 I I 1 I 1 I I I 2 9.5
76 9.9

4.0

1.4
3 2.8 2 2.8

I I I I I I I I I I
+

+--------+--------+--------+--------+--------+--------+--------+--------+
tutt 19 12. I 11.0 1 3 4.2 1 33.3 2 9.5
59 7.7

144 I 62.0 I I I I

I I

12 16.9

I I

2 2.8 1

I I

I I

71 9.2

4 3.7

43 I 39.4 - I 47 65.3 I I

34 31.2 11 15.3

I I I I

.9
5 6.9

I I I I I I

.9

1 I I 1

109 14.2

+--------+--------+--------+--------+--------+--------+--------+--------+
diorite 22 I I I I I I
+

+--------+--------+--------+--------+--------+--------+--------+--------+
copper ore 28 I I I I
+ 5

72 9.4

1 I 6 28.6 1 I
+

+--------+--------+--------+--------+--------+--------+--------+--------+
other 50 I I
+ 19 2.5

2 66.7 10 47.6 195

I I I I
+ 40

I I I I

.4

1 4.8

I I
+

21 2.7 770 100.0

+ COLUMN TOTAL

.6

375 48.7

25.3

5.2

.1

Core types:

o
1

. not 2 . not . not


3

. .
.

not exhausted, exhausted, exhausted,

<10' cortex 1040' 40-60' cortex cortex

eXhausted, not exhausted,

60-90' cortex >90\ cortex

5 - exhausted, 6 - exhausted, 7 - exhausetd,


8

<10' cortex

10-40\ cortex
40-60' 60-90\ cortex cortex

exhausted,

9 - exhausted.

>90' cortex (n-O)

293-

Rossen: Nanchoc Lithic Tradition

Table 7. Core types cross tabulatedwith form.

CORE

TYPE:

COUNT I
ROW PCT I

NOT

X H A U S T D

X H A U S T & D

I I FORM
amorphous 0 1 1

01

11

21
5 I 5.7 I 1 I 3.0 .1 I

I 31

ROW TOTAL 8I

51

61

--------+--------+--------+--------+--------+--------+--------+--------+--------+
3 1 9 1 3.4 I 10.2 I 4 I 30 I 29 I 4.5 1 34.1 1 33.0 I I 11 I 4 I I 33.3 I 12.1 1 I I 1 7 I 8.0 I 2 I 6.1 I I 1 I 88 1.1 I 11.4 I I I 33 4.3 1

71

+--------+--------+--------+--------+--------+--------+--------+--------+
spherical 1 I 7 I 8 I I 21.2 I 24.2 I I 1 I I

+--------+--------+--------+--------+--------+--------+--------+--------+
flat
(discoidal)

I 100.0 I
3

I
1 I

I
I I

I
I I

I
1 I

I
I 1

I
I I

.1
2 .3

+--------+--------+--------+--------+--------+--------+--------+--------+
conical 1 1 1 1 1 I 50.0 I 50.0 I

+------+--------+--------+--------+--------+--------+-------+--------+
(sub)triangular4 I

I
rectangular 5 I 1

7 I 13 I 5.9 I 10.9 I
7 I 9.1 I 13 I 6.8 I 5 I 6.5 I 15 I 7.8 I

3 I 2.5 I
3 I 3.9 I 4 1 2.1 I 3 I 1.7 I I I 1

1 61 I 29 I I 51.3 I 24.4 I
I 36 I 22 I I 46.8 1 28.6 1 111 57.8 83 47.4 39 I 20.3 I 50 I 28.6 I

6 I 5.0 I
4 I 5.2 1

I 119 I 15.4
I 77 10.0 1

+--------+--------+--------+--------+--------+--------+--------+--------+

+--------+--------+--------+-------- --------+--------+--------+
trapezoidal 6 I I 7 I I

+--------+--------+--------+-------pentOllgonOlll

10 I 5.,2 I 6 I 3.4 1 2 I 3.5 I 3 I

I 192 I 24.9 I I 175 22.7 57 7.4 28

--------+--------+--------+

13 1 20 I 7.4 I 11.4 I 5 I 8.8 I 2 I 4 I 7.0 I 2 I

+--------+--------+--------+-------semi-lunar 8 I 1 9 I 1 I 1.8 I I

--------+--------+--------+
I I I

+--------+--------+--------+--------+--------+--------+--------+--------+
10n9, thin, " 12 I 9 I

32 I 13 I 56.1 I 22.8 I

pointed
COLUMN TOAL

7.1 I
59 7.6

7.1 I
77 10.0 19 2.5

I
5 .6

I 42.9 I 32.1 I 10.7 I


376 48.7 195 25.3 40 5.2 1 .1

3.6
772 100.0

+--------+--------+--------+--------+--------+--------+--------+--------+

ANDEAN PAST 5 (1998) Table 8. Comparison of local versus non-local (exotic) materials by reductive stage.

:'294

Reductive stage

Local Materials

Non-Local Materials

All Lithics

Cores Tools utilized "Flakes Secondary Flakes Tertiary Flakes

757 2.2% 1288 3.7% 1245 3.5% 15069 42.9% 15716 44.7%

15 3.4% 32 7.2% 21 4.7% 164 36.8% 201 45.1%

772 2.2% 1320 3.7% 1266 3.6% 15233 42.8% 15917 44.7%

Totals

35127

446

35573

295-

Rossen: NanchOc Lithic Tradition

Table 9. Tool types crosstabulatedwith edge angle increments.

COUIIT I ROIl Pet' I


,

B DaB 0-15 I IS-3D

II N a L B I 31-45 I 4'-'0

115 DBGREE INCRBMBNUI I '1-15 I 14-90 I 91-105 I

ROlf TOTAL

I
I

fOOL r:P;
I 2

;
I

; ~~--; 41 ;.-::'I;---;;--;m--;--;
I 10.0 I 33.6 II4I~; II + 12.9 I 2.1

~;
2 50.0 I 91 32.1 I

;--; ;--;
25.0 I 25.0 I

;
I

;
I

;
I I + I I I I

24

1.0
140

:-~--m-;

t-------_.________. 31 31 UI I 4.4 I 3'.8 I


I I 31 10.7 I

f.-------_.
91 32.1 I 'I 21.4 I

I I + + 2 I I 2.9 I I t--------.--______. 11 I 3.' I I

5.'
68 2.1

28 1.1
115

;--m;--;-m;;--;
I 5.2 I 3'.5

;
I I

5.0

0; ;--;~;
I 40.0 I
29

'

;m_;;--; ;--;
14.8 I 2.' I
,

;--m;--;
I .9 I

4.'

1142~; II

36.3 I
f

13.8

;;--;--m;--; ;
I 2.5 I

II

80 3.2

.--------.
9 I I I I 2 50.0

I 18 I . 36.0

+
I I t 2 50.0

I I + I I

8 16.0

________.________.________. 50 I, I 2.0 + + + +

.________.________.______.._.
3 I 42.' I I I I I I I 7

.2

10 I, I
12

I I

1 I' 3 I 14.3 I 42.9 I

.3
5.1
85 3.4 1241 49.3 283 11.3 28 1.1 13

;
I t I I

;;--; ~;--;
1.8 1 1.2 I I I 32.' 22 25.9 I I I

QrJ
21 31.8

4 41.9

;
I t I I

;;--;
11.1

;__;

;___m__;129
I I I

13

I .8 I I t________.________.______.._. ZC I 8 I 1 I 3D.' I '.4 I 1.2 I

14

; ;;--;
I 7.5 I

15

;
I +

;
I

@J
44.3
3

550;---;;;--;~-.;;;--; ;;-.;
I 32.2 I 11.1 I 3.5 I

;--;
.1 I

;
I

; ;;--; ~;
I 14.5 IL::,;,:JI : 41

;;._;
32.5 I I
I + I I +

;;.-;

;--;

I fOOLmB
161

1.1 1111131
39.3

11.0 I 2.1 I '+:::::::::::.-t I I


I. + I I + I + 1 I +

I 44.4 I 14.3 I + + + 171 21 51 51 11 I 15.4 I 38.5 1 38.5 I 1.1 I + + t + + 181 I 51 II I I 33.3 I '.1 I + + + + + 191 I 91 11 I I 29.0 I 3.2 + + + + + 20 I I I 1 I I + + + + + 21121211 23116131 I 3.0 I 31.8 I 34.8 I 24.2 I ,4.5 + + + + 25 I I I I I I + + + + + I I + 2CI 27.4 I +

I +

.5
15

.,

+ 1 + I I + I +

+ I I + I I + 1 1.5 +

+ I I +

J1 1.2
5

.2
+ I I +

"
2.6
1

.3
3 3.2 + I I + 1 1.1 + I I +

+ 151 15.8 I +

95 3.8 1

21 I I I 1 I I I 1 1 100.0 I I + + + + + COLuMN 140 819 831 431 renAL 5.' 35.0 33.0 17.4

I I + 175 1.0 44 1.1

I 1 + , .4

I I +

.0
2515 100.0

Tool edge ~YP8.. -2 - VERYm:r.vr EDGlh ecI9. "'91.

-1 - m:r.vrEDGE. .dge 81>91. 1-' ag~. _n than tool "'gl. o - IlEDIUK EDGE. .dg. angl. tool 81>91. 1 FIlii:EDGE. .dg. "'91. 1-' d8gr_. 1... tban tool angle

. VERYFIlII: EDGE. 8dg. 2EDGz

.~ 1_n

10 agr_.

IIIOr. ~han ~ool "",gl.

COUII'1'I ROW PeT I COLPeT I I 'fOOL'l'YPE 12

ROW TO'rAL -2.001 -1;001 .001 1.001 1 -+ I I I 22 I 40 I 20 I I 17.2 I 31.3 I 15.6 I I 4.' 1 5.4 I 6.5 I + + + '18185 I 12.' I 28.2 I 10.6 I I 2.4 I 3.3 I 2.' I 2.001 --+ 18 I 14.1 I 5.7 I + '.4 2.5
142 11.5 45.1 36 12.8 11.4

81>91. .~ l...t
'1'YI1B

10 deg~.

1...

than tool angle

COUII'1'I ROIlPeT I C:OL PeT I I ,TOOL'1'YI1E , 1 I I I


2 I I

I 28 I 21.' I 4.0 + 131331111241 I 38.8 I 4.7

128 5.1

I
-..
c-n t-3 tII

ROW TOTAL -2.001 7 at.2 1.0


34 24.3

+
I I I I I I

+---+--+
I I I I I I 240 It.4 53.1 5'5 It.6 12.2

+
I I I I I. I 123' 0.4

I I

3.4

14

-1.001 .001 1.001 I I + + I 7 I 4 I 3 I I 2'.2 I 16.7 I 12.5 I I 1.5 I .5 I 1.0 I

2.001 + 3 I 12.5 I 1.0 I

24 1.0
140 5.6

+---+
15

291 23.5 41.6 108 38.4 15.5

I I I

415 33.5 56.5 54 It.2 7.4

I I I

151 12.2 0.0 28 10.0 '.1

I
I I I

I
I I I

_--+
I I I

I I I

1.0
1.0

~
281 11.2

+---+--+--+
I I It 13.6 I I 45 32.1 I I It 13.61

+
I 23 16.4

+
I I

+3

4.'

4.2 I

--+

6.1 I

+---+

6.2 I

7.3 I

+
68 2.7 28 1.1

+ +--+---+ + + 1613141111515128 I 10.7 I 14.3 I 39.3 I 17.' I 17.' I I .4 I ., I 1.5 I 1.6 I 1.6 1

1.1
13

I 17 I 12 I I 25.0 I 17.6 I I 2.4 I 2.7 I +--:--_+__ I 4 1 10 I 3 I I 35.7 I 10.7 I 1 1.4 1 .7 I + + 51331251241181151115 1 28.7 1 21.7 1 1 4.7 1 5.5 1 1 I _--+ 6 I 27 I 10 1 I 33.81 12.5 1

24 I 6 35.3 I 8.8 3.3 I 1.' + 5 I 5 17.' I 17.' .7 I 1.6 20.' 3.3 I I 15.7 5.8

I 'I I 13.2 I 2.' + I 5 1 17.' I 1.6 +

+I I

I
I

_+_--+
1 1 7 53.8

+___
1 I 2 15.4

I 1 + 1 I 1 +

17

3 23.1

1 1

1 I

1 7.7

1 I

.5
15

.4 I

--+__+__-+_+
6.7 .2 1 1

1.0 I

.3 I
I 1

.6 1
21 13.3 I .6 I

18151115121 I 33.3 I 1 .7 1

---+

33.3 .7

--+

I ,13.3 I .6

.6

-+
I I +
I 1 1 I I

21 I 26.3 I

1 13.0 I 1 4.8 I + + 'I 13 1 11.3 1 16.3 1

4.6 80 3.2 50 2.0 4

ItI

'I 51 71416131 I at.O I 16.1 I 22.6 I 12.' I It.4 I 1.3 I 1.1 I 1.0 I 1.3 I 1.' + + f - I
I I I 1 20.0 .1 ltI 28.8 2.7 I I I 1 20.0 .2 I I I 2 I 40.0 I .3 I I 1 1 20.0 1.3

1.2 5

20

+---+---+
7 I I 16 I 32.0 I

3.' I

2.2 I

2.'

+
I I

2.'

+---+
7 14.0

4.1 I
I I 211

.2
2.6

8 I 16.0 'I

10 I 20.0 I

'I 18.0 I

+I I

.,

I
12.1 2.6

+---+-'I I

2.3 I

1.8 1
2 50.0 I I

1 ---+---+

1.4 1

2.'

I
I I

2.2 I

+
i +

I I

812018111166 12.1 I 30.3 I 1.8 I 2.7 I

I I

16.7 3.5

2" I 50.0 I

I 1.2

+---+---+ 2512111111121
I 28.6 I 14.3 I 14.3

+
I 14.3

+
I 28.6

+
I

I +10

.3 I I

.4 I 1

I -+-__+
2 33.3 .6

I
I I I I

7 .3 '4 3.7 1

I 3 I I 50.0 I I .4 I' +I

I 1 I I 16.7 I 1.1 I I -+

I 1.2 1 -+

I .3 I .2 + +-_+_ 26 I 48 1 18 I 51.1 I It.l I 6.' I 4.0

+---+
I I I

.1 I I I 14 I I 14.' I I 1.' I

11

.3 I + 7 I 7.4 I 2.3 I
I I I

.6 I + 7 I 7.4 I 2.2 I
I I I

-+

27

I I 1

I I I

I 1 I 100.0 1.3

.0
2508 100.0
I N 1.0 0\

Table 10. Tool edge types.

I
COLUKII 'lO'fAJ.

I '" 27.' 452 18.0

--+ 734 U.3

+__ 308 12.3 315 12.6

297-

Rossen: Nanchoc Lithic Tradition

Table 11. Flake type frequencies.

Flake Type

Code Frequency -1.00 .00 1.00 5973 1164 11688

Percent 31.7 6.2 62.1

Flake Length < Flake Width Flake Length - Flake Width Flake Length > Flake Width

------Total 18825

------100.0

~
Table 12. Flake types crosstabulated with material typeS.

~ ~ rn
~
FlakeType Count I RowPet I Co1 pct I + + + + s11_ 21 I 3 I I 8 I gray I 27.3 I I 72.7 I I .1 I 1.1 I + + + + diorite 22 I 855 I 169 I 1539 I I 33.4 I 6.6 I 60.0 I I 14.3 I 14.5 I 13.2 I + ~--+ + + 23 I 1 I I I gra,nite green I 100.0 I I I Row Total
,11

FlakeType Flake '1'ypia Count I Count I 110" Pet I Row Ro" Pct I 110" Col Pct I '1'otal Col Pct I I -1.001 .001 1.001 I -1.001 .001 1.001 + + + + I -1.001 .OO~ 1.001 baealt 1 I 3313 I 625 I 5999 I 9937 + + + + I 33.3 I 6.3 I 60.4 I 52.8 quartz 11 I 11 I 4 I 40 I 55 I 55.5 I 53.7 I 51.4 I milky I 20.0 I 7.3 I 72.7 I .3 + + + + I .2 I .3 I .3 I quartzite 2 I 33 I 8 I 127 I 168 + + + + I 19.6 I 4.8 I 75~6 I .9 quartz 12 I 5 I 2 I 26 I 33 I .6 I .7 I 1.1 I cryatalline I 15.2 I 6.1 I 78.8 I .2 + + + + I .1 I .2 I .2 I andesite 3 I 414 I 119 I 1335 I 1868 + + + + I 22.2 I 6.4 I' 71.5 I 9.9 quartz 27 14 I 10 I 3 I 14 I I 6.9 I 10.2 I 11.4 I var colo~ed .1 I 37.0 I 11.1 I 51.9 I + + + + sU_ I .2 I .3 I .1 I 4 I 5 I 1 I 25 I 31 + + + + banded I 16.1 I 3.2 I 80.6 I .2 chalcedony 15 I 6 I I 6 I 12 I .1 I .1 I .2 I I 50.0 I I 50.0 I .1 + + + ~-+ ai18x I .1 I 1.1 I 13. 5 I 1 I 1 I 11 I cnam + + + + I 7.7 I 7.7 I 84.6 I .1 silex 44 16 I 10 I 2 I 32 I I '.0 I .1 I .1 I black I 22.7 I 4.5 I 72.7 I .2 + + + + sil_ I .2 I .2 I .3 I 6 I 2 I 141 6 + + + + coffee I 33.3 I I 66.7 I .0 granite 78 18 I 20 I 3 I 55 I I .0 I 1.0 I gray I 25.6 I 3.8 I 70.5 I .4 + + + + s11_ I .3 I .3 I .5 I 7 I 11 I 1 I 20 I 32 + + + + elk g~een I 34.4 I 3.1 I 62.5 I .2 tuff 19 I 708 I 106 I 1256 I 2070 I .2 I .1 I .2 I + + + + I 34.2 I 5.1 I 60.7 I 11.0 lu.atone I 11.9 I 9.1 I 10.8 I 8 I 391 I 81 I 766 I 1238 + + +~ + I 31.6 I 6.5 I 61.9 I 6.6 jasper 14 I 6.6 I 7.0 I 6.6 I .20111 1131 Abl " red + + + + I 7.1 I I 92.9 I .1 s11_ 2 9 I 1 I III I .0 I 1.1 I + + + + ~neral I 50.0 I I 50.0 I .0 I .0 I 1.0 I + + + + rhyolite 10 I 150 I 33 I 346 I 529 I 28.4 I 6.2 I 65.4 I 2.8 I 2.5 I 2.8 I 3.0 I
'1'otal

CI8

\0 \0

.1 2563 13.6 1 .0
4

I
baaalt snowflake

.0

I
+ I 3 I 75.0 1.0 + I 2 I 100.0 '1.0 + I 6 I 100.0 1.1 +

I
+ I I I + I I I + I I I + I I + I I I +

quartz amber

s11_ gray " red

+ + 24 I 1 I I 25.0 I I .0 I + + 25 I I I I I I + + 26 I I I I I I + +

.0
2 .0 6 .0

copper ore 28 I 12 I 22.6 I .2 + jaaper 29 I 4 c~e..bl red I 80.0 I .1 + Column 5968 '1'otal 31.7

I
I I + I I I +

5 I

36 I

9.4 I 67.9 .4 I .3 + I 1 I 20.0 1.0 + 1163 11671 6.2 62.1

53 .3 5 .0
18802 100.0

+---+

I tV \0 OQ

tV \0 \0 I

Table 13. Summary

of traits associated with formal tool types.

Type

FODII

Common/ Exped/

Rare

Curate

Edqe Ang.l.e Edqe Angle General/ Trend Mode Specialized

TET*
tendency

Activity

lq quadrilat

---20, 40 20, 40
35, 40-45

2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19
20

med quadrilat sm quadrilat


19 semilunar

med semilunar

sm semilunar thick semilunar C point multi point simple

C R R C R R R R C C
C

19 core
pentagonal incurved unmod util

blocky
thick rect thick pent sm triang med triang
19 triang

21 amorph 25 amorph
26 19 handled

C R R R R R C R
C

E E E E E C C C C C C E C C
E

16-30 31-45 31-45

----

40-45 35 30

-------

31-45 16-30 16-30

---20, 40 40, 50

----------

-------

60, 75 35-40

E E E E E
C

---------30-35

----

31-45 31-45 16-30 46-60 31-45


31-45

G G G G S S S S S S G G G S S G
G
.

VH-H VH-M VH-M VH VH-M VH-M

----------

slicing-cutting slicing-cutting slicing-cutting slicing-cutting slicing-cutting slicing-cutting

cutting
perforating perforating scrapingjplaning, hoeing slicing-cutting

VH-H M VH-M M VH M M
VH-M

planing,woodworking
light slicing-cutting scrapingjplaning, wood

:::a
CooS
. .

cutting cutting
light slicing-cutting
light slicing-cutting

31-45 31-45
-

G G
G

---------VH

light slicing-cutting

----

31-45

VH-M

------

31-45
31-45

G
S

-----scraping-planing, wood

C")

40-45

:
r:;'

[
* Tool edge type.
I

g:
;:s

LITHIC PROVENIENCE ANALYSIS AND EMERGING MATERIAL COMPLEXITY AT FORMATIVE PERIOD CHIRIPA, BOLIVIA

David L. Browman
Washington University

- St. Louis

Introduction Technical analyses of lithic artifacts from the Formative Period component of Chiripa, on the south shore of Lake Titicaca, Bolivia, indicate a number of items of non-local provenience. Among the items originating from considerable distances, up to 300-500 Ian from the site, are copper ores from the arid Pacific coast, obsidian from the punas of Arequipa and Puno, sodalite from Cochabamba, and basalt from Lake Poop6. Construction stone employed in the facing. and temple walls, weighing up to 4.5 tons, may have been extracted from quarries as far as 80 Ian distant by water. The identification of the systems by which the Chiripa inhabitants obtained these materials may serve to clarify the integration of the Chiripa polity with its neighbors, and its contributions to the development of the subsequent Tiwanaku federation.

tions. The Formative Period occupation of the site may be defined by 28 dates from 5 radiocarbon labs and 1 TL lab, listed in Table 1. Christine Hastorf and her students have recently begun a new series of excavations at the site. Additional dates no doubt will be forthcommg. The ceramics from Kidder's excavation were analyzed by K. L. Mohr for her Master's thesis. She proposed three phases: Pre-Mound or Sub-House level, 1400-900 b.c. (uncalibrated); Lower House level 900-600 b.c.; and-Upper House level 600-100 b.c. (Layman and Mohr 1965:200; Mohr 1966:3). Recently, she has renamed these phases Early, Middle, and Late Chiripa (Mohr"Chavez 1988:18).

Chiripa ceramics, when fIrst reported, were unique in having fiber temper, mainly ichu grasses (Stipa sp. and Festuca sp.). Thus, for many years, fiber tempered wares in the south-central Andes were characterized as The setting and chronology being derived from Chiripa, an interpretation Chiripa is an Upper Formative, or an Ininow fortunately mainly abandoned. In our excavations, we identified a non-fiber temtial Period and Early Horizon phase occupation and temple site on the south end of Lake pered ware component pre-dating the fiber Titicaca in Bolivia, at 3835 m or 12,580 ft temper phases. New phase names taken from elevation (Figures 1, 2, and 3). During the the local landowners of the site were emlater Tiwanaku federation period, the site was ployed to facilitate distinction between prere-utilized. Initial test excavations were con- . fiber tempered and fiber tempered phases and ducted by W. C. Bennett (1936); A. Kidder II to include other new archaeological elements in their definition. (1956) carried out subsequent excavations; and the current report is based on a joint American-Bolivian project in 1974-75 diThe traits defining these phases are derected by the author and Gregorio Cordero rived from ceramic (Browman 1980), arMiranda (Browman 1978a, 1989a, 1989b). chaeobotanical (Browman 1989b), zooarchaeological (Browman 1989a), and architectural The first series of radiocarbon determina(Browman 1978a) criteria. For example, the tions for Chiripa was a suite of 14 dates subCondori phase (1350-850 b.c.) is defmed in mitted by Kidder from his excavation units in part by such ceramic traits as grit-tempered the upper levels of the mound (Ralph 1959:56- (sand and crushed sandstone with micaceous 57). An additional 14 radiocarbon assays and inclusions) ware, neckless ollas, rare decora10 thermoluminescence determinations have tion of red slips and occasional incised debeen run on samples from the 1974-5 excavasigns, and vessels with round bases and basal ANDEAN PAST 5 (1998):301-324.

ANDEAN PAST 5 (1998) lugs. The shift from Condorill (1350-850 b.c.) to LluscolIl (850-650 b.c.) can be identified by new vessel and temper categories, but it is equally as evident in the archaeobotanical record (Browman 1989b:142), where a second chenopod species is added (provisionally identified as caniwa as well as quinoa), amaranth first appears, and some more minor weedy taxa first occur. It can also be identified zooarchaeologically as part of a pattern displaying a shift to greater dependence on lacustrine species. Fiber temper, which may first appear late in Condori/IB, becomes a significant ware in LluscolIl (850-650 b.c.), along with the addition of such traits as ring bases and flat-bottomed vessels. Classic Chiripa style decorations typify Mamani/IIIA (650-350 b.c.), the period of first temple construction and the appearance of the double wall . bin-storage houses; and at the end of MamanilIlIB (350-50 b.c.), Tiwanaku I-like ceramics are first identified. Correlation . between 'Mohr Chavez's phases and the Condori, Llusco and Mamani phases is aided by the fact that our project deliberately placed excavations (Unit II, Figure 2) below Kidder's House 2 and 3, and Lower and Sub-lower Houses 1 and 2, which are the basis of Mohr's three ceramic phases. Eight of the radiocarbon determinations for Condori and Llusco phases were derived from samples in Unit II under the Sub-lower houses. The assignment of the "pre-mound" Pennsylvania dates as CondorilIB is based on their approximate stratigraphic location in this same excavation unit. Inspection of the ceramics illustrated by Mohr (1966) suggests that her Early (or Pre-Mound), Middle (or Lower House), and Late (or Upper House) Chiripa ceramic phases are roughly equivalent to CondorilIB, LluscolIl-Mamani/IIIA, and Mamani/ IIIA-IIIB.
.

-302 southern lake basin Formative Period sites such as Chissi and Titimani also have subterranean temples. The mound began as a rather typical sort of accretional habitation site, where centuries of refuse accumulation along with the disaggregation of adobe and/or tapia mud construction resulted in an accumulated deposition rising three to four meters above, the surrounding fields. Evidence from our test cuts, supplemented by data from Bennett's and Kidder's work, indicate that the north side, and at least portions of the east and west sides of this mound (if not the entire mound), were later faced and enclosed with a stone wall, resulting in a roughly rectangular ground plan configuration (Figures 2 and 4). This event occurred near the beginning of Mamani/IIIA, based on evidence from one of our test units which cut the north wall (Unit I, Figure 2). Half a meter on the inside of the base of this wall is the oldest radiocarbon assay of 1530 b.c. :!::180, while the stratum on the exterior associated with the wall construction trench has a TL assay of 640 b:c. :!:: 390 and a 14Cassay of350 b.c.:!::155. My initial interpretation was that the visible subterranean temple (Figures 3 and 5) was constructed at this point as well. Much of the construction stone from Chiripa has been robbed and recycled in lat~r sites in the area. Opportunistic samples were taken from the remaining stone, and submitted to Dr. Ernest Ehlers for x-ray diffraction (XRD) and thin-. section microscopic analyses. Three different building stone materials were identified: andesite, limestone, and sandstone. No geologic origin voucher specimens were taken, as at that point we had no idea where quarries might be located. Andesites were rare in our sample and were confined to only a few small-sized quarried stones. Three major areas of andesites are near the "Little Lake" or Lago Pequeno section of Lake Titicaca. There are also some minor outcroppings south of Tiwanaku (Figure 1). Based on the specific inclusions and characteristics of the specimens submitted for analysis, Dr. Ehlers indicated that they were most similar to the andesites described for a source near the town of Copacabana (Newell

Building stone and construction phases Chiripa architecture is significant, in terms of our current knowledge of Titicaca basin Formative villages, in having a rectangular stone-faced mound on which first domestic dwellings, and later a subterranean temple were constructed. It is not unique: other

3031949; Mogrovejo Terrazas 1970). While visual inspection of the map suggests that, as the crow flies, it is only 30 Ian to this andesite source, because of local topographic features, the stone source is most readily reached by a lake voyage of about 80 Ian to a quarry on the north side of the peninsula. There is presently no evidence that andesite was transported by the more direct route over the intervening high hills. Analyses of andesites associated with Ponce Sangines's Tiwanaku III (a.d. 133-374) and IV (a.d. 375-724) phases at the site of Tiwanaku indicated that these construction stones also were quarried from the Copacabana peninsula, from the Yunguyo and Copacabana area sources illustrated in Figure 1 (Mogrovejo Terrazas 1970:251; Ponce Sangines and Mogrovejo Terrazas 1970:274). Ponce Sangines (1970a:75, 142, 146, 1971:90) reports blocks as heavy as 11 to 16 tons were first rafted by balsas (lake boats made from bundles of totora, a cattail-like reed) from Copacabana, through the straits of Tiquina, to the prehistoric lake port of Iwawe some 95 km by water, and then dragged another 22 Ian overland from Iwawe to Tiwanaku. Movement of these massive blocks on and off the balsas, without .capsizing, was difficult; several ashlars, lost in off-loading, help define the quay and pier of the Iwawe port.
.

Browman: Formative Period Chiripa sampled were 4.5 and 3.3 tons, while the two largest limestone ashlars sampled were 2.5 and 1.7 tons. These are the largest remaining stones in the temple floor plan map (Figure 3). Weights were determined by computing the volume of the whole ashlar, multiplied by the specific gravity determined from a sample of that ashlar. There are three major sandstone formations around the "Little Lake" (Newell 1949), as illustrated in Figure 1: the Cabanillas (of which only a portion near Cumana Island is included on the map), the Taraco (along the Taraco peninsula just south of Chiripa), and the Puno (along the south side of the Little Lake). Because of the proximity of the Taraco Formation to Chiripa, I initially assumed that all the sandstones would prove to be derived from this source. Based on inclusions in the samples, however, Ehlers indicated that in addition to Taraco materials, sandstones from the Puno Formation were utilized as well, a source which for Chiripa would be most efficiently exploited via water transport. Significantly, much of the sandstone employed later at Tiwanaku (a.d. 500-1100) also came from the Puno Formation (Avila 1971:226; Castanos 1971:212; Urquidi 1971:234). Limestone construction blocks from Chiripa were quarried at' some distance, because there are no nearby sources on the peninsula. XRD and thin-section microscopy of the limestone samples submitted to Dr. Ehlers for analysis indicated that these limestones were all from the Copacabana Formation. This Formation crops out along a fault line which abuts the south side of the Straits of Tiquina on the Copacabana Peninsula and runs south through Cumana Island. Because Formative Chiripa ceramics had been reported by Bennett and others on sites on Cumana Island, my original interpretation was that limestone was quarried on the island, and transshipped by the balsas to the shores below Chiripa. But Chiripa wares also have been identified from a number of sites on the Copacabana peninsula by Gregorio Cordero Miranda, John Hyslop, Karen Mohr Chavez, Sergio Chavez, Charles Stanish, and others. Neither area of limestone outcropping can be specifically identified as

Limited observation of the shoreline directly in front of Chiripa failed to reveal any evidence of similar off-loading patterns. There is a linear levee or aqueduct which runs from the mound area north to the lake (Graffam 1990:138-150). This seems to be the logical route for transport of the stone from the lake to the temple. If dock facilities existed at Chiripa, they may have been somewhat removed from the immediate foreshore, or may have existed at a different lake level; we know that in the last 50 years, Lake Titicaca levels have varied more than 5 meters. The most common building stones remaining at Chiripa were limestone and sandstone. Centuries of stone-robbing prevent accurate reconstruction of importance. The two largest sandstone ashlars opportunistically

ANDEAN PAST 5 (1998) the source at this point, but access to either requires water transport of the construction stone. Ponce Sangines (1970a:62) reports that limestone employed at the site of Tiwanaku comes from this same Formation, which also occurs at some isolated outcrops in the Pampa Koani and Tambillo area (two of which are shown in Figure 1). Thus the Tiwanaku masons seemed to have continued exploitation of the same sources of limestone as the Chiripa masons. While some of the sandstone may have come from deposits directly behind Chiripa, other sandstone blocks came from Puno Formation deposits 20-30 km by water to the southwest. The limestone blocks came from Copacabana Formation sources 20-40 km by water to the northeast and the andesite appears to have derived from a quarry to the north which is most readily reached by a water route of 80 km in length. These data indicate that the masons of the semi-subterranean temple at Chiripa did not have a single.preferred source, but rather secured building stone from a wide number of sites around the "Little Lake", no doubt in part based on ease of access to the shore. At this point in the analysis, it seemed that the pattern of quarrying andesite, sandstone, and limestone blocks of substantial size had first" been initiated during the Mamani IlIA or Classic Chiripa phase, and continued unchanged on into Tiwanaku Period constructions. As work continued, it became possible to differentiate some construction techniques of Chiripa masons from those of Tiwanaku masons.1 The Mamani phase retaining wall on the north side of the mound, as exposed in our test cuts, as well as in Bennett's work, consisted of relatively small, partially coursed rectangular blocks (Figure 4). In unpublished
.

-304 Bennett and Kidder, there was evidence of a later Tiwanaku Period retaining wall, with vertical ashlars set at intervals, with the smaller rectangular stone laid between them (Figure 5), very typical of "Classic" Tiwanaku architecture at the Tiwanaku type site. (By 1974, utilization of the site by the current village residents as an adobe fabrication "mine" had left the stones from this wall in a dis-' turbed array.) The extant, visible semi-subterranean temple in the center of the Chiripa mound, measuring 21.3 m by 22.6 m, employed vertical ashlars along with horizontally laid smaller stones. Most of the 1,450 cubic meters of fill removed from this temple was disturbed, but near the base of the walls we recovered some intact zones and floors of Tiwanaku III refuse, indicating a Tiwanaku III (a.d. 133-374) Period. of construction for this version of the temple. During cleaning of the temple floor, we sank a test cut in the northeast comer to assess floor construction sequences. Much to our surprise, we found evidence of an earlier, deeper temple wall set back almost half a meter from the later Tiwanaku phase temple walls (Figure 6). Evidence of this earlier temple also appears in profiles of Bennett's test trench. Steel rod probes at half a dozen places around the perimeter of the Tiwanaku III phase temple suggest that this earlier structure was probably similar in outline to the Tiwanaku phase building, but slightly larger, roughly 23 m by 24.5 m. Ceramics recovered from the test cut in the northeast comer indicated the earlier structure was Classic Chiripa or Mamani phase. Alan Sawyer (1981), who worked as a member of Kidder's project, reported that portions of the walls of the upper Chiripa houses uncovered in their excavations had been leveled off, and the houses packed with backfill from temple construction. These data indicate that the second, Tiwanaku phase, temple was constructed after the abandonment of the mound, with the mound top leveled off and a new retaining wall constructed around the exterior. The evidence of two distinct periods and styles of temple construction and mound facing raised the question whether the magnitude

photographswhich GregorioCorderoMiranda
provided, as well as in the excavation cuts of
IClassic Tiwanaku style temple construction involves vertical ashlars with various types of coursed infilling. For examples, see Manzanilla's work (1992) at the Akapana, or Ponce's various publications on the Kalasasaya.

305of long-distance transport of massive building stone initially posited for Mamani phase in fact pertained to that period or to the later Tiwanaku rebuilding. Most of the large ashlars of limestone and sandstone that we had sampled were the vertical ashlars of the typical Tiwanaku style construction. On the other hand, a 0.6 ton, Pajano style sandstone monolith (Browman 1978b; Cordero Miranda 1977), characteristic of Mamani phase sites around the Little Lake, was found recycled as a vertical ashlar in the later Tiwanaku style temple, with its decorated side buried facing the wall fill and its undecorated backside showing. Thus we know that at least some large sandstone ashlars were quarried in Mamani III times. The only andesite pieces which can securely be placed as Mamani phase are small (under 20 kg); and unfortunately, none of the large limestone ashlars we sampled could be definitively associated with the earlier constructions. Nonetheless, the Mamani phase Chiripa residents were quarrying substantial quantities of stone. Although most of the wall stone subsequently has been robbed, the few remaining sections of the Tiwanaku Period temple indicate that approximately 200 tons of stone were required for that temple building. If the earlier temple is of the same magnitude, as our probes suggested, then it also required roughly the same quantity of quarried stone. The Mamani Period exterior facing walls of the mound may have required as much as 5 to 10 times more stone, depending on height and dimension estimates, and whether they enclosed all four sides or not. A similar quantity should be projected for the Tiwanaku Period facing walls. A revised construction sequence model would suggest that the Mamani Chiripa builders at 600 b.c. were principally using sandstone, along with minor amounts of other building stone. The Tiwanaku III Period Chiripa area residents, who rebuilt the temple ca. a.d. 300-400, chose or found themselves forced to secure the large vertical ashlars from more diverse sources. Limestone blocks came from the Copacabana Formations 40 Ian to the east. Some sandstones were acquired as far as 30 Ian away from the Puno Formations at the southwest end of the Little Lake. In addition,

Browman: Formative Period Chiripa local sandstones from the hills behind the site were used, and andesites likely from the Copacabana quarry 80 km by water to the north were also employed. Because the style of construction of the second subterranean temple at Chiripa is similar to the Tiwanaku III phase Semisubterranean Temple at the prestigous center of Tiwanaku, it is not surprising that the materials utilized closely match those previously identified for the developing capital. The sandstone from the Puno Formation, which is closer to the Tiwanaku center, might have been brought to Chiripa by masons and their crews coming from Tiwanaku to supervise the temple's construction. Metal and ores A total of 31 gold, copper, and tin-copper bronze metal and copper-mineral items were recovered during our Chiripa excavations. Silver is known from Formative Period levels at Tiwanaku but was not recovered in our test units. Although gold has been reported from Formative Period sites elsewhere in the south central Andes as early as 2,000 b.c., the only gold from stratigraphic contexts found in our Chiripa work came from Tiwanaku III contexts; Bennett (1936) also reported finding gold in Tiwanaku Period burials at the site. Copper ore and copper artifacts, however, were recovered from the earlier Chiripa phases: two items from Condori IB contexts (1250-850 b.c.) and one from Mamani IIIB components. In addition, Bennett (ibid:433) excavated a Classic Chiripa phase burial which had copper, stone, and bone beads around each ankle. Thus, we have good evidence for copper metal use during the Formative Period. Gold and copper metal artifacts, as well as a number of the turquoise and sodalite items, were recovered from the 1,450 cubic meters of disturbed temple fill excavated. While 95% of the artifacts from the temple fill were Mamani or Classic Chiripa, the other 5% included Tiwanaku III, IV and V, Inca, Colonial, and even 20th century artifacts. It is tempting to assume that because most diagnostic artifacts in the temple fill are Mamani phase-related, it is therefore likely that most of the metal arti-

ANDEAN PAST 5 (1998) facts could also be Mamani phase. However, the disturbed nature of the deposit makes this an uncertain proposition. During recent historic periods, copper was extracted from several mines in the Titicaca Basin. Four mines are known in the immediate Chiripa/Tiwanaku vicinity (Ponce Sangines and Mogrovejo Terrazas 1970:220-223), so the early occurrence of copper in the Chiripa site seems reasonable. The ores from the local mines are carbonates including azurite and various copper sulfides. However, by XRD and microscopy, Ernest Ehlers identified copper ore samples from the Condori IB and disturbed temple fill as being brochantite, and he also identified as brochantite a Formative specimen from Kidder's excavation provided by Alan Sawyer at my request (Ehlers, report of June 15, 1976). Brochantite is aform of copper mineral which only occurs in extremely arid areas. In reviews of the geologicalliterature, both Ehlers (personal communication, June 15, 1976) and Heather Lechtman (personal communication, April 25, 1977)'reported that the closest known sources for this mineral are along -the Pacific Coast, from a series of mines in the north Chilean desert along the Tarapaca-to-San-Pedro-de-Atacama axis, as well as in some of the southern Peruvian mines in the Arequipa area, although it as not as typical of the Peruvian mineral beds. There also is evidence of exploitation of local copper ores at Chiripa. A Tiwanaku Period ore sample from our excavations at Chiripa was identified as azurite by Ehlers. One specimen of azurite ore was identified among the Chiripa samples from Kidder's excavation, supplied to Ehlers at our request by Alan Sawyer. In an earlier XRD analysis on a sample from Tiwanaku, malachite was identified (Sawyer, personal communication, January 29, 1975). These findings suggest that local Titicaca basin ore sources were being exploited by the first centuries A.D., if not earlier. With copper ores readily available in the immediate area and exploited during Tiwanaku and later phases, the occurrence of apparent Pacific coast copper ores, such as brochantite and antlerite, seemed almost a "coals to Newcastle" situation. The associa-

-306 tion of brochantite and antlerite with the Formative occupations, with local ores appearing only in the later phases, suggested a possible initial development of this technology in northern Chile, with a subsequent adaptation of the technology by the altiplano cultures. This interpretation seems to be reinforced by Ruppert's (1982, 1983) identification of the known turquoise sources as also North Chil-' ean. Analyses conducted on our other Bolivian samples replicated the pattern observed at Chiripa. We secured two copper ore samples from the Formative Period sites of Chullpapata and Santa Lucia in Cochabamba while we were collecting comparative ceramics specimens from cuts made by William Kornfield, and two additional samples were acquired from the Putuni and Pumapunku structure areas ~t the Tiwanaku site. The ore sample from Chullpapata included a mix of brochantite and antlerite, and the sample from Santa Lucia was brochantite. Antlerite, .like brochantite, was only known from arid areas such as the north Chilean coast according to the sources available to Lechtman and Ehlers in 1975. Hence the two ore samples from the Cochabamba area appeared to reinforce a possible pattern of initial association of Chilean ores with Formative Period sites in Bolivia. The ore sample from the Putuni temple/palace area of Tiwanaku included antlerite, brochantite, and cuprite, while the second sample from the Pumapunku temple/palace area was brochantite. These data, taken in conjunction with the information from Chiripa, suggested that if antlerite and brochantite were in fact most typical of copper sources from the arid north Chilean coast, that as early as 1200-1000 b.c., Chile was a significant supplier of such ores to the Bolivian altiplano and valleys, and the coastal sources continued being important until a.d. 500 if not later. Locally available Titicaca basin ores, such as azurite and cuprite, at least in our extremely small sample, do not seem to have begun to be extensively exploited until perhaps as late as 100 b.c.-a.d. 100, a surprising situation, considering Ponce Sangines's argument in several publications that an indigenous copper smelting industry

307was well entrenched in this part of the altiplano by 1200 b.c.2 After presenting this model of Formative utilization in the Bolivian altiplano of copper ores from North Chile in several papers, I now have new evidence which will require modification of this hypothesis. While brochantite and antlerite are most typical of Chilean mines 300 to 500 km southwest of Chiripa, at least brochantite is found as a less common copper mineral at the Bolivian mine of Corocoro, only 100 air km south of Chiripa.3 Formative
2Much of Carlos Ponce Sangines'model of extensive
.

Browman: Formative Period Chiripa copper ores of brochantite could thus derive either from the coastal Chilean-Peruvian sources or the Corocoro area, although the Formative Period antlerite ores presumably indicate coastal origin. Thus Chiripa might be obtaining at least part of its copper minerals from the Corocoro source area, but it appears that the Formative Cochabamba sites might still be characterized as securing copper mate,:, rials from Chile. Beads and pendants: cbrysocolla, sodalite and turquoise Of the 24 beads and pendants recovered in our excavations, 16 were blue and green, mainly copper-mineral based stone. This blue-green mineral group was initially divided into three categories in the field: malachite, turquoise, and sodalite. These artifacts remain in the national museum in La Paz. However, some of the items were damaged by the workmen during excavation (the National Museum-trained crew insisted that Bolivian regulations required picks and shovels, rather than trowels), and the resulting small chips recovered were thus available for analysis. True "chemical" turquoise is a specific crystalline compound of copper, but "cultural" turquoise may include a wide range of other copper-bearing blue and green stones, such as malachite, azurite, chrysocolla, and the like. Four turquoise artifacts were recovered in our Chiripa excavations; the three artifacts for which samples were available were identified as true chemical turquoise. No published identification or characterization of Andean turquoise sources was available for the geological consultants of the project when they
Cuprita), which is not linked to intrusive activity but generated in red clastic sediments by migration of cennate brines. Moreover, brochantite is reported as an 'oxidation production of chalcocite at Corocoro. "So you have two options: brochantite from Corocoro or brochantite from Chile/Southern Peru. The former seems more likely, although pre-Colonial exploitation of oxide copper (including brochantite) was widespread at Chuquicamata. I'm not sure about Cerro Verde."

copper utilization by 1200 b.c. isJ:>asedupon material which he identified as "escoria" or copper slag at the site of Wankarani, in a level dated between 1210 b.c. and 800 b.c. However, Marc Bermann (1995) reports that. from his Wankarani culture site .at La Joya, he submitted 34 samples of "scoria" apparently identical to Ponce's. material, which Bermann thus assumed was copper slag. Test results indicate that none of these samples proved to have any copper. Rather they appear to be a type of natural pyroclastic tufa. Thus the basis for an extensive early copper smelting industry in the early formative period of .the Bolivian altiplano seems now very much open to question. 3Richard H. Sillitoe, Consulting Economic Geologist of London, who has written extensively on the copper industry of the Andes, has offered some fine-tuning suggestions. He writes (Personal communication, February 22, 1992): "Brochantite and antlerite only form by oxidation of chalcocite in the presence of pyrite. Chalcolite normally forms by supergene enrichment in arid to semiarid climatic regions. The major chalcocite zones (blankets) in northern Chile (e.g. Chuquicamata) and Southern Peru (e.g. Cerro Verde in Arequipa) were generated in the Oligocene-mid Miocene (25-15 m.y.a.), and were then fossilized by intensifying aridity (hyperaridity). The brochantite and antlerite formed in oxidized zones above, and at the same time as, the chalcocite blankets. "Hydrothermal copper deposits related to intrusive activity on the Bolivian altiplano (e.g. La Joya) were only emplaced after the cessation of this major chalcocite-forming event, so are unlikely to contain brochantite and antlerite. However, chalcocite (with local pyrite) formed as a bynogene (primary) ore mineral at Corocoro (and nearby smaller deposits, e.g. Chacritas,

307was well entrenched in this part of the altiplano by 1200 b.c.2 After presenting this model of Formative utilization in the Bolivian altiplano of copper ores from North Chile in several papers, I now have new evidence which will require modification of this hypothesis. While brochantite and antlerite are most typical of Chilean mines 300 to 500 km southwest of Chiripa, at least brochantite is found as a less common copper mineral at the Bolivian mine of Corocoro, only 100 air km south of Chiripa.3 Formative
2Much of Carlos Ponce Sangines' model of extensive copper utilization by 1200 b.c. is J:>ased upon material which he identified as "escoria" or copper slag at the site of Wankarani, in a level dated between 1210 b.c. and 800 b.c. However, Marc Bennann (1995) reports that from his Wankarani culture site at La Joya, he submitted 34 samples of "scoria" apparently identical to Ponce's material, which Bermann thus assumed was copper slag. Test results indicate that none of these samples proved to have any copper. Rather they appear to be a type of natura~pyroclastic tufa. Thus the basis for an extensive early copper smelting industry in the early formative period of the Bolivian altiplano seems now very much open to question. 3Richard H. Sillitoe, Consulting Economic Geologist of London, who has written extensively on the copper industry of the Andes, has offered some fine-tuning suggestions. He writes (Personal communication, February 22, 1992): "Brochantite and antlerite only fonn by oxidation of chalcocite in the presence of pyrite. Chalcolite normally forms by supergene enrichment in arid to semiarid climatic regions. The major chalcocite zones (blankets) in northern Chile (e.g. Chuquicamata) and Southern Peru (e.g. Cerro Verde in Arequipa) were generated in the Oligocene-mid Miocene (25-15 m.y.a.), and were then fossilized by intensifying aridity (hyperaridity). The brochantite and antlerite fonned in oxidized zones above, and at the same time as, the chalcocite blankets. "Hydrothennal copper deposits related to intrusive activity on the Bolivian altiplano (e.g. La Joya) were only emplaced after the cessation of this major chalcocite-fonning event, so are unlikely to contain brochantite and antlerite. However, chalcocite (with local pyrite) fonned as a bynogene (primary) ore mineral at Corocoro (and nearby smaller deposits, e.g. Chacritas,

Browman: Formative Period Chiripa copper ores of brochantite could thus derive either from the coastal Chilean-Peruvian sources or the Corocoro area, although the Formative Period antlerite ores presumably indicate coastal origin. Thus Chiripa might be obtaining at least part of its copper minerals from the Corocoro source area, but it appears that the Formative Cochabamba sites might still be characterized as securing copper materials from Chile. Beads and pendants: cbrysocolla, sodalite and turquoise Of the 24 beads and pendants recovered in our excavations, 16 were blue and green, mainly copper-mineral based stone. This blue-green mineral group was initially divided into three categories in the field: malachite, turquoise, and sodalite. These artifacts remain in the national museum in La Paz. However, some of the items were damaged by the workmen during excavation (the National Museum-trained crew insisted that Bolivian regulations required picks and shovels, rather than trowels), and the resulting small chips recovered were thus available for analysis. True "chemical" turquoise is a specific crystalline compound of copper, but "cultural" turquoise may include a ~de range of other copper-bearing blue and green stones, such as malachite, azurite, chrysocolla, and the like. Four turquoise artifacts were recovered in our Chiripa excavations; the three artifacts for which samples were available were identified as true chemical turquoise. No published identification or characterization of Andean turquoise sources was available for the geological consultants of the project when they
Cuprita), which is not linked to intrusive activity but generated in red clastic sediments by migration of cennate brines. Moreover, brochantite is reported as an oxidation production of chalcocite at Corocoro. "So you have two options: brochantite from Corocoro or brochantite from Chile/Southern Peru. The fonner seems more likely, although pre-Colonial exploitation of oxide copper (including brochantite) was widespread at Chuquicamata. I'm not sure about Cerro Verde."

ANDEAN PAST 5 (1998) conducted their studies in 1975; thus a place of origin could not be determined. However, since that time, Ruppert (1982:73, 1983:102) has conducted a series of trace element analyses and believes that the cultures of Northwest Argentina and Bolivia, including Tiwanaku, obtained the greater part of their turquoise from deposits in northern Chile: from the Chuquicamata area, including the mines of Chuquicamata, EI Abra, and EI Salvador; from a presumed mine east of Arica; and from still unknown other Chilean mines. The Chiripa materials no doubt were obtained from these same North Chilean turquoise mines. The class of six items field-categorized as "malachite" proved to be a variety of copperbased minerals other than malachite, when appropriate thin-section and XRD analyses where performed on samples submitted. Two of these samples were identified as chrysocolla -- one from Mamani/IIIB strata deposits, and the other from the temple fill. The initial chemical analysis of the chrysocolla suggested that the mineral was in two phases, most typical of an artificial process, such as faience fabrication in Egypt. . This tentatively was proposed by Ehlers as possible evidence for a previously unreported technology, but subsequent analyses by Zimmerman and Ehlers indicated that the samples had not been heated prehistorically, and that the two phases were due to natural processes. Another sample of the field category "malachite" was identified as chlorite and calcite, and a fourth as cuprite. Fragments were not available for the other two items, so the complete re-definition of the field category "malachite" may be even broader. In order to assist in verifying the six sodalite artifacts collected for evaluation, I obtained a voucher piece of sodalite from the Cerro Sapo mine in Cochabamba, which was used for the type material against which Ehlers tested the sodalite category materials by XRD. The two sodalite fragments tested from our Chiripa excavations (one from a Mamani Period stratum and the other from temple fill) were from Cerro Sapo. A sodalite sample provided by Alan Sawyer from Kidder's Chiripa excavations also was determined

-308 to be from Cerro Sapo. Bennett (1936:433) reported excavating a flexed burial from Classic Chiripa deposits, "around each ankle were beads of lapis-lazuli, bone and copper". As sodalite is locally mis-named lapis-lazuli, presumably these anklet beads also are sodalite. Ehlers analyzed three additional sodalite samples that we had collected: one each from' the Tiwanaku phase deposits of Tiwanaku, the Tiwanaku phase deposits at Lukurmata, and the Formative Period midden of Chullpapata, Cochabamba. XRD tests on these samples indicated Cerro Sapo origin. Sawyer (personal communication, January 29, 1975), has documented sodalite from in his collections from Tiwanaku through XRD, and Ponce Sangines has reported in a number of publications on other samples of Cerro Sapo sodalite from Tiwanaku. At this point, all chemically identified samples of sodalite from the Bolivian altiplano have proved to be from the Cerro Sapo source. Ruppert (1982:74, 1983:103) reports that the sodalite from prehistoric sites near Andahuaylas and in the Asia Valley in Peru, as well as the Quiani complex of Arica, Chile, are also from Cerro Sapo; some of these samples are associated with deposits believed to date back almost four millennia, suggesting that the trade in sodalite from Cerro Sapo pred~tes the origins of Chiripa by half a millenmum or more. Trade in Cerro Sapo sodalite items thus has a long duration in the Bolivian altiplano. Sodalite was exchanged on a continuing basis between the Cerro Sapo source in the ' Ayopaya area of Cochabamba 225 air km east of the Titicaca basin and other Andean centers for minimally 2,500 years (based on Chiripa data) and more than 4,000 years (based on external reports), and continued to be a significant jewelry item through Tiwanaku V occupations or later, or at least as late as a.d. 1100. Turquoise, chrysocolla, and other copper mineral-based artifacts may be part of the evidence relating to other exchange networks, such as with the Arica area in northern Chile 300 air km southwest, or the Chuquicamata area of Chile 500 air km south-southwest. Although the earliest stratified samples of turquoise, sodalite, and chrysocolla are only from

309the Mamani component deposits (650-50 b.c.), the fact that copper ores were securely identified from the Condori/IB levels (1250-850 b.c.) indicates that the Titicaca shore inhabitants were participating in the trade in blue and green mineral beads and pendants by then, if not earlier as suggested by evidence from other localities. Tools: obsidian and vitreous basalt Chiripa Formative Period projectile points, knives, drills, scrapers, and other stone tools were manufactured from a variety of cryptocrystalline materials, including various cherts, quartzite, and obsidian. Stone hoes and an adze-like form were manufactured from vitreous basalt. Stone bowls, mortars, pestles, and other grinding stones were manufactured from several other stones, whose identification in the field was not secure; these items remain in Bolivia. Only the materials of volcanic origin, obsidian and basalt, can currently be identified as to source. Samples of two of the 16 artifacts of vitreous basalt, a hoe from Condori/IB (1250-850 b.c.) and a hoe fragment from Mamani/IIIB (350-50 b.c.), were taken by Gregorio Cordero Miranda, Director of the Museo Nacional de Arqueologia, a.n:dco-director of the project, and Luis Girault, of the Centro de Investigaciones Arqueol6gicas en Tiwanaku, for petrographic analysis in the national museum laboratory in La Paz for verification of field identification. Cordero and Girault reported (personal commuriications, June 1975) that . these two artifacts were confirmed to be vitreous basalts from Cerro Querimita, in the Lake Poop6 area, Oruro.4 Marquez et aZ. (1975) and Ponce Sangines (1970b) report Cerro
4Petrographic laboratory procedures for identification of the basalts at the National Museum in 1974 are the same as those described by Marquez et ai. (1975). Characterization was limited to a few elements. While this is less extensive than current NAA trace element analyses, recent NAA work reported for Cruro has shown these petrographic analyses, so far, to be quite adequate for the characterization of Cerro Querimita as contrasted to other possible vitreous basalt sources.

Browman: Formative Period Chiripa Querimita vitreous basalts utilized as hoes from Wankarani through Imperial Tiwanaku (or later) Periods, or from as early as 1200 b.c. through at least a.d. 1100, from sites in La Paz and Oruro departments. In the Little Lake basin, these basalts previously had not been recovered in any dated contexts earlier than ca. 300 b.c. The Condori and Mamani samples suggest that the vitreous basalt industry of the Lake Poop6 area was involved as a significant component of a long distance trade network with the Titicaca basin (and probably also with Cochabamba). Vitreous basalt hoes were observed during our surveys of the Formative mounds at Chullpapata and Santa Lucia in association with materials dating from as early as 1200 b.c. Obsidian materials from Chiripa were analyzed by Ehlers, Richard Burger, and Larry Haskin. Inopportunely, Ehlers's analysis of the first sample (C5) in 1976 using X-ray fluorescence (XR.F) was before Burger and Asaro's seminal article. on Peruvian and Bolivian obsidian was published (1977, 1979). The XR.F-determined composition of this sample proved to be different than the XRF analyses published earlier by Avila (1975a, 1975b) for Bolivian obsidians. Hence Ehlers reported that sample C5 was from a different obsidian source than the Formative Sora Sora or later Tiwanaku obsidians that Avila had analyzed. But Burger and Asaro subsequently observed (1977:6) that "there are disturbing differences between our XR.Fresults and those of Avila, although the samples analyzed from Tiwanaku and Sora Sora were taken from the same obsidian fragments used in the study of Avila." Thus the non-comparability of Ehlers' and Avila'sXR.F results may be methodological, rather than the documentation of a new obsidian type. Burger and Asaro characterized 21 samples of obsidian from Bolivia by XR.F and NAA (neutron activation analysis): 3 samples from Formative Sora Sora, 2 samples from Kallamarka (a Tiwanaku administrative site with Tiwanaku III, IV, and V materials) and 16 samples from Tiwanaku itself. These samples proved to be from a source they define as the Titicaca Basin Type. This obsidian type

ANDEAN PAST 5 (1998) also is very common in sites from Arequipa and Puno Departments in Peru. Because this obsidian was particularly common along the Sicuani-Puno axis, and was utilized in Peru as early as 1200 b.c. at Pikicallepata, and as late as A.D. 1500 at Sillustani, Burger (Burger and Asaro 1977:37) posited a possible origin somewhere in Puno Department. Subsequently Burger (personal communication, April 21, 1980), suggested that "I now think that the actual mine may be in Arequipa". This prediction has been borne out as Sarah Brooks recently has located the Titicaca Basin Type source at a quarry in the Colca valley in Arequipa.5 Burger subsequently tested three additional samples (C1, C2, and C3) from the Chiripa project, using XRF. Two of the samples(C2 from a Mamani IIIB component, 35050 b.c., and C3 from disturbed' temple fill) proved to be of the Titicaca Basin Type; the third sample, C1, from a Condori IB component (1250-850 b.c.) was of the Tumuku Rare Type (Burger 1977). In his 1977 technical report, Burger observed that the Tumuku Rare Type is not infrequent in sites in the Chucuito area and other parts of the southern highlands, and should therefore be considered a major rather than a rare type. The source of the Tumuku Type was suggested to be near the PeruBolivia border on the west side of Lake Titicaca in the 1977 report; in his 1980 communication (Burger, April 21, 1980), he refined that hypothesis, suggesting that "the Tumuku Rare
5 After the original draft of this paper was submitted, Sarah Osgood Brooks (personal communication, June 19, 1995) reported discovering two obsidian quarries in the Colca Valley in 1994, while conducting research for her Ph.D. dissertation (Brooks 1998) at the University of Wisconsin-Madison. NAA analyses of the obsidian samples conducted by Michael D. Glascock, Missouri University Research Reactor, indicated that the Cotallalli Quarry samples are "a perfect match" for the Titicaca Basin-type of obsidian (Glascock to Brooks, August 22, 1994). For further infonnation, see Brooks et ai. (1997:449). Richard Burger and colleagues have also identified the Colca Valley as the source of the "Titicaca Basin"-type obsidian. See Burger et ai. (1998) in this volume.

-310

Type is especially common in the Chucuito archaeological sites tested over the last few years and I suspect that its source should lie in that area." The fifth obsidian analysis was conducted on a flake recovered from a Condori IB component (1250-850 b.c.) float sample. Haskin, analyzed this sample, C4, using NAA. The results from this NAA work (Table 2) were unlike any of the NAA analyses by Burger and Asaro, suggesting a previously unknown source. Thus stratified Chiripa Formative Period obsidian artifacts come from at least three different sources.6
6Michael D. Glascock of the Missouri University Research Reactor (MURR) has been conducting the analyses of obsidian from various Titicaca basin sites for Martin Giesso. Glascock recently reviewed my obsidian data from Chiripa as part of the background for that project. While he concurs that samples C2 and C3 are from the Titicaca Basin Type source, with respect to the identification of C 1 being TUmuku Type, he notes the paucity of infonnation in the literature on the characterization of this type, and thinks that we need more supporting data for a sure detennination. Sample C4 does not match any of the types he had tested to that point (Glascock, personal communication, March 4, 1994). Sample CI derives from the probable Tumuku Rare Type, from a source possibly in the Puno region; samples C2 and C3 derive from the Titicaca Basin Type from the Colca valley source in Arequipa; C4 comes from a previously unidentified source. Thus there are clearly 3 different obsidian sources. The XRF work by Ehlers on sample C5 did not match any of the XRF data published by Avila (1971, I975a, 1975b), so he suggested that this sample was an unknown. As noted, other labs have had problems replicating the results of Avila. I do not have a copy of the XRF profile by Ehlers, so it is not possible to compare it with more recent studies. Thus, I cannot say whether it is yet another previously undefmed source or not, and hence whether there are three or four different obsidian sources represented in our excavation work at Chiripa. Glascock and Giesso (1994) have also recently conducted some additional analyses of Titicaca Basin obsidians. They identified 10 different obsidians: 5 fairly common, and 5 with one example only. One of the common types is Tiwanaku A, with no known source, which appears from visual inspection to most likely be from the same source as Burger's Tumuku Rare Type. Sora Sora obsidian, a common type in later

311Burger (1977) observed that "the discovery of Tumuku Type obsidian in the early layers of Chiripa had not been anticipated. It suggests that early patterns of obsidian utilization near Lake Titicaca may have been very different than in later times." The subsequent defmition of yet another new type of obsidian in the Condori layers supports this proposition. It appears that initially Chiripa may have been utilizing more local sources (the Tumuku source and the new type), and that later, exploitation of these sources was supplanted by the Titicaca Basin source, 325 air Ian west in the Colca valley of Arequipa Department, Peru.Chiripa may well have been one of the nodes on a network, which could have involved Tumuku and Titicaca Basin Type obsidians passing south to Lake Poopo in exchange for items such as Cerro Querimita vitreous basalt and Uyuni salar salt blocks mov-

Browman: Formative Period Chiripa Paz in April of 1995 are currently in progress.8 Ehlers tested one archaeological sample from a Condori/IB component (1250-850 b.c.) and Gundersen tested a second sample from a Mamani/IIIB component (350-50 b.c.). The first sample, examined by Ehlers in 1976 using XRF techniques, was tentatively identified as katawi, a calcium carbonate preparation, frequently consumed with chenopod grains. The identification is listed as "tentative" because in 1976 I had no expectation of recovering such materials in prehistoric context, and thus had limited comparative materials. The second sample studied by Gundersen in 1987 using both XRD and XRF was more securely identified as. katawi (see discussion in Browman and Gundersen 1993). While katawi may have been derived from local sources, many of the other comestible earths consumed in the Titicaca basin in historic and current times are not local, and may originate from areas up to 500 Ian distant. The katawi evidence appears to establish the antiquity of geophagy in the Andes at least two millennia earlier than previously documented. Final comments Locational patterns based on elemental analyses of various mineral artifacts suggest a substantial area for resource exploitation. Some of the temple building stone, up to 4.5 tons in weight, came from a quarry source more than 80 Ian north by water; some early copper ores appear to originate from mines of the Pacific Coast region of southern Peru or northern Chile, 300 to 400 air Ian west and southwest; turquoise is identified as coming from the Chuquicamata mines 500 air Ian south-southwest or from a hypothesized mine in the Arica area 300 air Ian southwest; sodalite artifacts came from Cerro Sapo mines 225 air Ian east in the Ayopaya area; obsidian came from at least 3 sources, one of which has
8A new series of earth samples was collected in the special pharmaceutical markets in La paz and Oruro in April of 1995. Dan Kremser, Microprobe Specialist, Department of Earth and Planetary Sciences, Washington University, is conducting XRD, EDS, and other analyses of these samples.

ing north.7
Comestible earths

Two samples of comestible earth, one from a broken pot and one from a feature, were recovered nom the Formative Period levels at Chiripa. With the aid of James Gundersen, several of the more than 25 comestible and medicinal earths utilized in ethnohistoric and modem Titicaca basin households have been chemically characterized (Browman and Gundersen 1993); additional work on samples collected in the local markets in Oruro and La
basin sites, is not present in our Chiripa sample. Giesso secured three additional samples from Chiripa (cultural associations unknown) which all were identified as the Titicaca Basin type. While these new samples do not change the number of obsidian sources known for Chiripa, they indicate that, as with other Titicaca Basin sites such as Lukunnata, Tiwanaku, and Khonko Wankani, the Titicaca Basin type seems to be the most commonly exploited source. 7Waldo Avila (personal communication, July 1975), believed he could identified Titicaca basin obsidian types from "Mound culture" (e.g., Wankarani) sites in Oruro. Work in progress by Martin Giesso will no doubt help in the evaluation of this model of obsidian trade.

ANDEAN PAST 5 (1998) been identified as a quarry 325 air kIn west in the Colca valley of Arequipa Department, Peru; basalt hoes originated from Cerro Querimita 325 air kIn south on the southwest shores of Lake Poopo, and so on. A regional system in scarce resource acquisition appears to functioning across the altiplano during the second and first millennia B.C. and was part of the basis for integrating the local level political elements.9 This exchange structure, which is argued to have involved movement of goods by balsa boats and llama caravans, must be included in any study examining the evolution of the complexity of Titicaca Basin societies. Llama caravans are the local mechanism through which the lighter weight mineral items identified here most likely were mobilized, e.g., the copper ores and m~tal; sodalite, turquoise, and chrysocolla jewelry; obsidian hunting, shearing, and butchering tools; basalt agricultural hoes; and at least some of the medicinal and comestible earths. Llamas were important in the Formative phases at the Chiripa site: a hig4 frequency of llama bones were found in all levels (Browman 1989a; Kent 1982); llama dung was an major fuel source (Browman 1989b); several hundred bone weaving tools were recovered, suggesting a major wool textile industry; and bone

-312

pack-harness toggles were retrieved, indicating llama pack usage. The larger building stone of the temples and retaining walls at Chiripa appears to have been transported via a second mechanism, that of balsa totora-reed boats. Zooarchaeological analyses of the Chiripa assemblage reveal an increasing focus on lacustrine resources, in-. cluding aquatic birds, various fishes, and harvesting of comestible algae and duckweeds, during the first millennium B.C., indicating intensified utilization of balsas for extraction of materials. Analyses by Horn (1984) have shown that the specific species of fish and aquatic birds recovered in the Chiripa zooarchaeological samples are the same as those taken by the contemporary Um fishing cooperatives on Lakes Titicaca and Poop6 that he studied. The mineral goods discussed all appear to have been traded at least as early as 12001000 b.c. (the limit of project evidence available - see footnote 9), ahd to have maintained a significance in the economy of the altiplano outlasting anyone particular regional political group. These mineral items continued to retain importance as culturally-defined status markers (such as the sodalite, turquoise, chrysocolla, and metals), or as necessary raw materials for the primary ,exploitation of the local resource base (such as the obsidian and basalt). They persist over long periods as important trade goods, and some of them may possibly even develop into mercantile commodities during Tiwanaku federation phases. The acquisition of construction blocks of sandstone, limestone, and andesite seems to mirror these long-standing regional patterns, at a smaller scale. There was a long history of quarrying and transporting building stone from quarries around the Little Lake, as revealed by the origins of construction materials from Chiripa, as well as other sites. At Chiripa, the earliest evidence comes from a few samples in the Mamani phases. Such transport becomes a particularly notable component of the subsequent Tiwanaku phases both at Chiripa and other sites (Ponce 1970a, 1971, Ponce and Mogrovejo 1970), and con-

9Reconstruction of the magnitude, duration, and function of the regional resource procurement networks is deliberately limited to the evidence from the Chiripa excavations. Climatic evidence, discussed in a paper currently in preparation, suggests the potential for a rather dramatic reorganization of the Titicaca Basin about 4,000 years ago. In other areas of the Andes, preceramic resource procurement systems have greater time depth, and from studies on the Archaic Period in the southern sierra of Peru, might well be extended here. The evidence is not yet available to ascertain what impact the reorganization of the Titicaca basin economies four millennia ago might have had on the caravan. systems. During the first millennium B.C., a number of local level polities developed at places like Sillumoqo, Chiripa, Chissi, Tiwanaku, and the like. The explication of the political interaction between these polities is beyond the scope of this paper.

313tinued on through the Inca Period. In a sense, one could argue that the activity continued into the early Colonial and Republican Periods, because the built stone ruins of the preColumbian sites were employed as stone sources for the impressive Catholic cathedrals at places like Laja and La Paz, the cobble stones and foundation stones for residences in La Paz, and later even as construction stone for the late 19th century railroad bridges. The current understanding of ceramic patterns of the Bolivian altiplano points to a sharing of specific ceramic vessel shape, decoration, and fabrication traits among discrete clusters of communities around the Lago Pequeno during the first millennium B.C., which appear to define a series of polities. One of the more general traits associated with Chiripa ware is fiber temper; because fib~r temper was first identified in the archaeological record at Chiripa by Bennett, the occurrence of fiber temper wares from as far south as Northwest Argentina, east in Cochabamba, and west in northern Chile and far southern Peru, has often been linked with {::hiripa-- but at this point fiber temper is such. a general Titicaca basin Formative trait that simple observation of its occurrence is not sufficient to document an actual linkage with the Lago Pequeno polity. Rather, the accident of archaeological discovery which led to fiber tempered wares first being described at Chiripa has led to undemonstrated lists of associations. On the other hand, the trace elements analysis of the Chiripa mineral samples demonstrates early relations between Chiripa (Figure 7) and sites to the east in Cochabamba (sodalite), to the south at Corocoro (copper) and as far as Lake Poop6 (basalt), to the southwest in northern Chile (copper minerals, turquoise), and to the west at Colca in southern Peru (obsidian). These associations allow us to verify and refine some of the linkages previously predicated on ceramic arguments, and to begin to outline the basis of the political economy resulting in the subsequent Tiwanaku hegemony.

Browman: Formative Period Chiripa Acknowledgements The technical analyses upon which the arguments in this article are based were provided by a number of specialists, whom I acknowledge with great gratitude. XRD and thin-section analysis of construction stone samples were conducted by Ernest G. Ehlers, Department of Geology and Mineralogy, Ohio State University. XRF and NAA analyses of obsidian samples were conducted by Ehlers, Richard L. Burger, Department of Anthropology, Yale University, and Larry A. Haskin, Department of Earth and Planetary Sciences, Washington University. XRF, XRD, TL, and SEM microprobe analyses of copper, turquoise, sodalite, and chrysocolla materials were conducted and evaluated by Ehlers, Heather N. Lechtman, Department of Metallurgy and Material Sciences, M.LT., and David W. Zimmerman, Center for Archaeometry, Washington University. XRF, XRD, and EDS analyses of comestible earths were done by Ehlers, 'James N. Gundersen, Department of Geology, Wichita State University, and Daniel Kremser, Department of Earth and Planetary Sciences, Washington University. These analyses were conducteq intermittently as I provided samples to these consultants from 1976 through 1995; thus results represent the "state of the art" for that analytical technique when conducted. The integration and interpretation of the technical analyses has been done solely by Browman. While none of this paper would have been possible without the superb assistance and contributions of these colleagues, all syntheses and the responsibility for any errors resulting from them, are mine alone. The excavation work for this project was funded under NSF grant BNS 7401258. Support for radiocarbon and thermoluminescence assays, and for part of the technical analyses, has been provided by three Washington University Faculty Research grants. Ehlers, Gundersen, Haskin, Kremser, and Zimmerman provided additional technical support from their own programs, without which assistance this paper would have been the poorer. In

ANDEAN PAST 5 (1998) Bolivia, the help provided by Carlos Ponce Sangines, Director of the Instituto Nacional de Arqueologia, and Gregorio Cordero Miranda, Director of the Museo Nacional de Arqueologia and Bolivian co-director of the Chiripa project, eliminated many logistical problems and allowed the smooth functioning of the excavations. The assistance of Waldo Parra Velasco on Figure 2, Clark L. Erickson on the development of Figures 4 and 5, and Carol Diaz-Granados Duncan on the re-drafting all figures, is gratefully acknowledged. References Cited
Avila Salinas, Waldo 1971 Estudio comparativo por difracci6n de rayos X de las areniscas de Pumapunku. In Procedencia de las areniscas utilizadas en el Templo precolombino de Pumapunku (Tiwanaku), by Carlos Ponce Sangines, Arturo Echazu, Waldo Avila Salinas, and Fernando Urquidi Barrau, pp. 219-228. La Paz: Academia Naciona1 de Ciencias de Bolivia, Publicaci6n 22. 1975a Elementos trazas de algunas obsidianas bolivianas. La Paz: Centro de. Investigaciones Arqueol6gicas, n.s. Publicaci6n 4, Instituto Nacional de Arqueologfa. 1975b Amilisis espectrograftco semicuantitativo de algunas obsidianas de Bolivia, Argentina, PerU y Chile. La Paz: Centro de Investigaciones Arqueologicas, n.s., Publicaci6n 14, Instituto Nacional de Arqueologfa. Bennett, Wendell C. 1936 Excavations in Bolivia. Anthropological Papers, American Museum of Natural History 35:329-507. Bermann, Marc 1995 Formative Period Settlement Hierarchy and Political Economy in La Joya, Oruro. Paper presented at the 60th annual meeting, Society for American Archaeology, Minneapolis, Minnesota, May 1995. Brooks, Sarah O. 1998 Prehistoric Agricultural Terraces in the Rio Japo Basin, Colca Valley, Peru. Unpublished Ph.D. Dissertation, Department of Geography, University of Wisconsin-Madison. Expected December 1998. Brooks, Sarah 0., Michael D. Glascock and Martin Giesso 1997 Source of Volcanic Glass for Ancient Andean Tools. Nature 386:449-450 (April 3, 1997 issue). Browman, David L. 1978a The Temple of Chiripa (Lake Titicaca, Bolivia). In El Hombre y la Cultura Andina, Volume 2, edited by Ramiro Matos Mendieta, pp. 807-813. Actas y Trabajos del III Congreso Peruano del Hombre y la Cultura Andina. Lima: Editora Lasontay.

-314
1978b Toward the Development of the Tiahuanaco (Tiwanaku) State. In Advances in Andean Archaeology, edited by David L. Browman, pp. 327-349. The Hague: Mouton Press. 1980 Tiwanaku Expansion and Altiplano Economic Patterns. Estudios Arql.(eo/6gicos 5: 107-120. 1989a Origins and Development of Andean Pastoralism: an Overview of the Last 6,000 Years. In The Walking Larder: Patterns of Domestication, Pastoralism, and Predation, edited by Juliet Clut-, ton-Brock, pp. 256-268. London: Unwin Hyman. 1989b Chenopodium cultivation, lacustrine resources and fuel usage at Chiripa, Bolivia. In New World Paleoethnobotany, edited by Eric E. Voight and Deborah M. Pearsall, pp. 137-172. Special Issue, Missouri Archaeologist 47. Columbia: Missouri Archaeological Society. Browman, David L. and James N. Gundersen 1993 Altiplano Comestible Earths: Prehistoric and Historic Geophagy of Highland Peru and Bolivia. Geoarchaeology 8(5):413-425. Burger, Richard L. 1977 X-Ray Fluorescence Analysis of Obsidian Artifacts from Chiripa, Bolivia. Technical report on file, Department of Anthropology, Washington University, S1.Louis. Burger, Richard L. and Frank Asaro 1977 Trace Element Analysis of Obsidian Artifacts from the Andes: New Perspectives on Pre-Hispanic Interaction in Peru and Bolivia. Berkeley: Lawrence Berkeley Laboratory, Report LBL-6343. 1979 Analisis de rasgos significativos en la obsidiana de 10s Andes Centrales. Revista del Museo NacionaI43:281-326. Burger, Richard L., Frank Asaro, Guido Salas, and Fred Stross 1998 The Chivay obsidian Source and the Geological Origin of the Titicaca basin Type Obsidian Artifacts. Andean Past 5:203-223. Castanos Echazu, Arturo 1971 Estudio petrografico comparativo de las areniscas de Pumapunku. In Procedencia de las areniscas uti/izadas en el templo precolombino de Pumapunku (Tiwanaku), by Carlos Ponce Sangines, Arturo Castanos Echazu, Waldo Avila Salinas, and Fernando Urquidi Barrau, pp. 207-218. La Paz: Academia Nacional de Ciencias de Bolivia, Publicaci6n 22. Cordero Miranda, Gregorio 1977 Descubrimiento de una estela litica en Chiripa. In Arqueologia en Bolivia y Peru. La Paz: Jornadas Peruano-Bolivianos de Estudio Cientffico del Altiplano Boliviano y del Sur del Peru 2:229232. Glascock, Michael D. and Martin Giesso 1994 New Perspectives on Obsidian Procurement in the Titicaca Basin, Bolivia. Paper presented at the 29th International Symposium on Archaeometry, May 9-14, Ankara, Turkey. Proceedings in press, Ay Melek Ozer, editor. Graffam, Gray C. 1990 Raised Fields Without Bureaucracy: An Archaeological Examination of Intensive Wetland

315Cultivation in the Pampa Koani Zone, Lake Titicaca, Bolivia. Ph.D. dissertation, Department of Anthropology, University of Toronto. Ann Arbor: University Microfilms. Horn, Darwin D. 1984 Marsh Resource Utilization and the Ethnoarchaeology of the Uru-Murato of Highland Bolivia. Ph.D. dissertation, Department of Anthropology, Washington University, St. Louis. Ann Arbor: University Microfilms. Kent, Jonathan D. 1982 The Domestication And Exploitation of the South American Camelids: Methods of Analysis and Their Application to Circum-Lacustrine Archaeological Sites in Bolivi'a and Peru. Ph.D. dissertation, Department of Anthropology, Washington University, St. Louis. Ann Arbor: University Microfilms. Kidder II, Alfred 1956 . Digging in the Titicaca Basin. University Museum Bulletin (University of Pennsylvania) 20(3):16-29. Layman, Frederick C. and Karen L. Mohr 1965 Petrographic Analysis of Pottery from Chiripa, Bolivia. Proceedings of the Pennsylvania Academy of Sciences 39:220-225. Manzanilla, Linda 1992 Akapana. Una piramide en el centro del mundo. Mexico D.F.: Instituto de Investigaciones Antropologicas, Universidad, Nacional Aut6noma de Mexico. Marquez Ostria, Jorge, S. Kussmaul, and P. K. Hoermann 1975 Estudio petrog,.afico del Cerro Querimita y su comparaci6n con artefactos /fticos. La Paz: Centro de Investigaciones Arqueologicas, Publicacion 7, Instituto Nacional de Arqueologia. Mogrov~QTeITazas,G~ardo 1970 Estudio geol6gico petrografico. In Acerca de la procedenda del material Utico de los monumentos de Tiwanaku, by Carlos Ponce Sangines and Gerardo Mogrovejo TeITazas,pp. 189-258. La Paz: Academia Nacional de Ciencias de Bolivia, Publicaci6n 21. Mohr Chavez, Karen L. 1966 An Analysis of the Pottery from Chiripa, Bolivia: a Problem of Archaeological Classification and Inference.. Unpublished Master of Arts thesis, Department of Anthropology, University of Pennsylvania, Philadelphia. 1988 The Significance of Chiripa in Lake Titicaca Basin Developments. Expedition 30(3): 17-26. Newell, Nonnan D. 1949 Geology of the Lake Titicaca Region, Peru and Bolivia. Geological Society of America Memoir 36. New York: Geological Society of America. Ponce Sangines, Carlos 1970a Examen arqueologico. In Acerca de la procedencia del material /ftico de los monumentos de Tiwanaku, by Carlos Ponce Sangines and Gerardo Mogrovejo TeITazas, pp. 11-188. La Paz: Academia Nacional de Ciencias de Bolivia, Publicacion 21.

Browman: Formative Period Chiripa


1970b Las culturas Wankarani y Chiripa y su relacion con Tiwanaku. La Paz: Academia Nacional de Ciencias de Bolivia, Publicacion 25. 1971 Examen arqueologico de las ruinas precolombinas de Pumapunku. In Procedencia de las areniscas utilizadas en el templo precolombino de Pumapunku (Tiwanaku), by Carlos Ponce Sangines, Arturo Castanos Echazu, Waldo Avila Salinas, and Fernando Urquidi Barrau, pp. 13-205. La Paz: Academia Naciona1 de Ciencias de Bolivia, Publicacion 22. zas 1970 Iwawe y Copacabana. In Acerca de la procedencia del material /ftico de los monumentos de Tiwanaku, by Carlos Ponce Sangines and Gerardo Mogrovejo TeITaZas,pp. 261-274. La Paz: Academia Nacional de Ciencias de Bolivia, Publicacion 21. Ralph, Elizabeth K. 1959 University of Pennsylvania Radiocarbon Dates III. Radiocarbon 1:45-58. Ruppert, H. 1982 Zur Verbreitung und Herkunft von TOrkis und Sodalith in Praekolumbischen Kulturen der Kordilleren. Baessler Archiv, Beitrage zur VolkerkundeN.F.30:69-124. 1983 Geochemische Untersuchungen an TOrkis und Sodalith aus Lagerstlitten und Praekolumbischen Kulturen der Kordiller~n. Berliner Beitrage zur Archaeometrie 8:101-210. Sawyer, Alan R. 1981 House Structures at Chiripa, Bolivia. Paper presented at the 21st Annual Meeting of the Institute of Andean Studies. Berkeley, California. Stuiver, Minze and Paula J. Reimer 1986 A Computer Program for Radiocarbon Age Calibration. Radiocarbon 28(2B): 1022-1030. Urquidi Barrau, Fernando 1971 Geoquimica de las artmiscas de Pumapunku. In Procedencia de las areniscas utilizadas en el templo precolombino de Pumapunku (Tiwanaku), by Carlos Ponce Sangines, Arturo Castanos Echazu, Waldo Avila Salinas, and Fernando Urquidi Barrau, pp. 229-240. La Paz: Academia Naciona} de Ciencias de Bolivia, Publicacion 22.
Ponce Sangines, Carlos and Gerardo Mogrovejo TeITa-

--/

CHIRIPA and TIWANAKU STONE SOURCES o .10 20 30 . Km. Scale

LAGO

KEY:

. .-.

edit

_
m

ANDESITES

LIMESTONES
SANDSTONES

SETTLEMENTS

Figure 1. Location of Chiripa and the major stone sources near Chiripa, around the "Little Lake" portion of Lake Titicaca.

317-

Browman: Formative Period Chiripa

TENNIS COURT
.... III III a: .... II) III Q

MAIN HOUSE

;;

SOCCER FIELD

DATUM STAKE OLD EXCAVATED AREAS

c::::!J MODERN ADOBE WALLS

'-14 E-W GRID LINES A-J N-S GRID LINES CONTOUR '_III INTERVALS O.50m. RECENT EXCAVATIONS

SCALE
O'

2 '4

..

!
II

Figure 2. Contour map of the Chiripa mound site, showing locations of the 1974-75 excavation units I, II, and III.

ANDEAN PAST 5 (1998)

-318

r--~--\ \ \
\
~\G8

'UI'

i ~ \ ~

. . .

\~

i 1 i

~ .

\cP \ .""'\

~
\
i

, II,
I I
+~oo 1, I I II I 1

,II,
I

I ,D
II 0 II I I t
+ F7

~
.

1+IE8 II

\.

II

G7

, _n D
-:!:.~
.

~
:

;'0 ~ ~

ii

II 11.11 I. ~ ,

'

D
.

m q

1r------il:::)
l+ ID8 \1 ;jI'J ,fJ

0
l
II

O b7 ):J
I

HC8

II I I

i
I

r
I

! I

,r
! I

+ E7

J L + 07

r
L

-',
1+ f.!
"
I I
II

I I I I

II

1L l+f!
II I

I I I I

II

,'
I I

J.

II II

O& lf
,
\ \

,' \
+C

Cordero
Stela

<21
I
I
~

+IQ!>

~
+ G5

1I
.'
.

I I

I I
.
~

I I

~.Up..,
B.M.#18

1111 II
J L
I I I I I I
+ F5

~II ___~
+ E5

:i II O
+ 05

~
oo

t...c.! -j I I I

'\)
+ C5

I I I I

~'=J-~--ll

l .1 r I
.'
.

l I :-~ I
I I

~t""f)
.

(j)

'1 I

G l'--.

t:l:
: :

I I

I I

'-l

II"~~_J J".!___~C?~

~+.-G~-'7-- :f{.Q

!J
F3
3

~Q
I
L
G3

C4--, O O -~4J ~-'-O'rn-rt rb


I I II
D4

~~~--

I I I

,:
o
2

I I I I

4 ;
'+
4 5

+1

I I 1 I

E3'

- ..+03

I I I I

i
I I

-,+C3

Scale

I
+C5 Control Points

Temple Wall Continuation c:::Jao Remaining Building Stone

Cut Boundaries

Figure 3. Plat map of the excavation of the Tiwanaku phase temple, showing location of re-' maining wall and of abandoned robbed asWars.

I.H 1.0 I

.-

~ ~ . .

~ c

! ~.
~
[
Figure 4. Idealized reconstruction of the Mamani phase houses, mound facing wall and temple at.Chiripa. Based on a composite of information from Bennett, Kidder and our excavations. Some retaining walls omitted for clarity.
r"). ::s-o .... ~ '6' t:!

~ ~ ~ ~
00
(JI

>

-3
---....
\C \C 00 "-'

Figure 5. Idealized reconstruction of the Tiwanaku III phase temple at Chiripa, showing vertical ashlar construction technique.

I W N o

VJ N

.. NORTH

Ito LakeTltlcaca 350 meters I

SCALE: -= o 1

Sm.
Eatlrnatod 0'11I1081 wall Holght

~3837 m.
MQ.OEItIt SUItF1!CE

'I

,C'

C'

Chlrlpa Temple Wall

i i i

Chlrlpa Facing Wall

~ ~ '. ~ ~ ~ .... ~. ~ ~ ~ -. C ~ 9 ::;. ~. ~

.J

I i

G'13

G.12

G.11

G.10

G9

G.8

G.7

Figure 6. Composite 30 m. cross-section, showing relationship of the Mamani phase (Classic Chiripa) facing wall and temple wall to the subsequent Tiwanaku III phase constructions. Note relative position of the two temple walls and facing walls. Simplified section based on profiles from Bennett, Kidder, and our excavations.

ANDEAN PAST 5 (1998)

-322

.
Scale 0 200

.
300 km

BENI LA PAZ

CHUQUISACA

ANTOFAGASTA

Figure 7. Approximate locations of the major sources for the long distance procurement at Chiripa.

323-

Browman: Formative Period Chiripa

Table 1. Chiripa radiocarbon and thermoluminescence determinations.


Phose CondorillA (1350-1250 be) Lab Number Beta-31291 GX-4057 RL-496 WU-TU9A WU-TU9B P-129 RL-495 P-145 RL-494 RL-493 WU-TU9C WU-TU9D RL-491 RL-492 GX-3596 WU-TU9E GX-3595 WU-TU9F P-126 P-115 GX-4059 . Uncalibrated 1530 b.c. 1260 b.c. 1240 b.c. 900 B.C. 1380 B.C. 1290 b.c. 1220 b.c: 1020 b.c. 900 b.c. 870 b.c. 510 B.C. 1050 B.C. 950 b.c. 860 b.c. 435 b.c. 100 B.C. 775 b.c. 640 B.C. 600 b.c. 518 b.c. 510 b.c. 470 B.C. 350 b.c. 346 b.c. 341 b.c. .331 b.c. 285 b.c. :i: 180 :i: 170 :i: 120 :i: 430 :i: 500 :i: 130 :i: 120 :!: 120 :i: 110 :i: 110 :i: 370 :i: 450 :i: 140 :i: 180 :i: 165 :i: 310 :i: 100 :i: 390 :i: 116 :i: 133 :i: 165 :i: 365 :i: 155 :i: 114 :i: 115 :i: 113 :i: 240 Calibrated 1820 B.C. 1500 B.C. 1500 B.C. 900 B.C. 1380 B.C. 1545 B.C. 1430 B.C. 1180 B.C. 1040 B.C. 1010 B.C. 510 B.C. 1050 B.C. 1095 B.C. 1015 B.C. 575 B.C. 100 B.C. :i: 220 :!: 200 :i: 130 :i: 430 :i: 590 :i: 145 :i: 120 :i: 140 :i: 130 :i: 120 :i: 370 :i: 450 :i: 175 :i: 205 :i: 215 :i: 310

CondorilIB (1250-850 be)

Llusco/II (850-650 be)

Mamani/IIIA (650-350 be)

"

wu- TU9I
GX-4058 P-125 P-l24 P-142 1-8314 Mamani/IIIB (350-50 be) P-116 P-143B WU-TU9G P-141 Beta-31290 P-I44 P-143A WU-TU9H P-117 P-118 WU-TU9J

900 B.C. :i: 90 640 B.C. :i: 390 670 B.C. :i: 150 625 B.C. :i: 145 590 B.C. :i: 180 470 B.C. :i: 365 360 B.C. :i: 180 355 B.C. :i: 165 355 B.C. :i: 165 345 B.C. :i: 165 260 B.C. :i: 280 470 B.C. :!: 90 435 B.C. :i: 95 330 B.C. :i: 340 300 B.C. :i: 130 265 B.C. :i: 65 275 B.C. :i: 115 265 B.C. :!: 115 40 B.C. :i: 300 A.D. 35 :i: 115 A.D. 181 :i: 31 A.D. 160 :i: 270

427 b.c. :i: 110 368 b.c. :i: 113 330 B.C. :i: 340 325 b.c. :i: 116 290 b.c. :i: 90 243 b.c. :i: 111 227 b.c. :i: 112 40 B.C. :i: 300 a.d. 13 :i: 104 a.d. 22 :i: 105 A.D. 160 :i: 270

* All radiocarbon determinations (b.c., a.d), except P-143A, on charcoal. Calibrated determinations (B.C., A.D.) based on Stuiver and Reimer J 986, 20 year average statistic, method B. TL determinations (WU-TLxxx) are calibrated in sidereal years. Assignment to phases based on stratigraphic and cultural associations.

ANDEAN PAST 5 (1998)

-324

Table 2. Trace element composition of the unknown obsidian, second millennium B.C. Chiripa. (Sample in direct association with assay RL-493, see Table 1. Data from analyses by Larry D. Haskin conducted in 1986. Sample irradiated at the Missouri University Research Reactor, Columbia.)
As Ba Br CaO (%) Ce Co Cs Eu FeO (%) Hf La Lu NaiO (%) Nd Rb Sb Sc Sm Ta Tb Th U W Yb Zr 11.1 :f: 440:f: 1.8 :f: <1.0 68.9 :f: 0.14 :t 15.0 :f: 0.26 :f: 1.19 :t 7.7 :f: 33.7 0.673 :t 4.50 :t 29 :t 180 :f: . 1.23:t 2.73 :f: 6.41 :t 1.00 :t 1.13 :t 17.5 :t 7.5 :f: 3.0 :t 4.44 :f: 270 :t 0.4 20 0.2 1.0 0.03 0.3 0.02 0.03 0.3 0.015 0.05 4 8 0.05. 0.05 0.11 0.05 0.06 0.4 0.3 0.6 0.07 60

TEXTILES FROM THE LOWER OSMORE VALLEY, SOUTHERNPERU: A CULTURAL INTERPRETATION


Ran Boytner University of California at Los Angeles

Introduction
During the tenninal stages of the Tiwanaku state, the peripheral regions were the first to feel the impact of the state's collapse. The middle Osmore Valley, around the modem city of Moquegua in Southern Peru, was one of these places (Figure 1). The site of Omo, which had served as the Tiwanaku center in the middle valley for several centuries, was abandoned during the tenth century A.D. at the beginning of the Tumilaca phase (Bermann et al. 1989:282; Goldstein 1989). Around this time, some of the local population moved into the coastal segment of the Osmore Valley, which had not previously.been inhabited by Tiwanaku-related people (Owen 1991a:l). . These new, cOIIlplex societies in the coastal Osmore Valley were based on irrigation agriculture and constructed a long canal to incre~e agricultural production. They were also camelid herders (Owen 1991b). Perhaps the most striking aspect of this intrusive society is that it was clearly divided into two distinct, .contemporaneous groups, archaeologically identified as the 110Tumilaca/Cabuza culture and the Chiribaya culture.' Both cultures left extensive archaeological remains, including habitation sites, cemeteries, canals, and field systems, which can often be unambiguously assigned to one culture or the other. This dual occupation of the coastal Osmore Valley lasted from around A.D. 950 to A.D. 1250, while the Chiribaya culture persisted to perhaps as late as A.D. 1375 (Owen 1992). The lower Osmore Valley has been the focus of various projects. Some are completed, while others are still not published. This research analyzes the textiles from one
1 For extensive discussion of the differences between the two cultures see Owen (1993).

such project, the Proyecto Colonias Costera& de Tiwanaku (PCCT) (Owen 1991a, 1991b, 1993). Specifically, it analyzes the textiles from burials at the site of EI Algodonal, which is located in the lower third of the coastal valley (Figure 1) and was excavated by PCCT in 1989-1990. The focus of the research is to compare the textiles from 110Tumilaca/ Cabuza burials and Chiribaya burials, which can be clearly distinguished by the distinct style of ceramic vessels associated with each. Additional specimens from the sites of Chiribaya Alta and Yaral (Figure 1) were examined in order to increase the size of the Chiribaya sample and to evaluate the site level variability in Chiribaya material. The samples from these two sites were selected randomly from burials with associated ceramics in the AIgorobal phase (early Chiribaya) style (Jessup 1991). Only a small fraction of the textiles from these sites were analyzed, however, and future analyses of the complete textile collections from these sites will certainly increase ourunders~ding of the coastal valley peoples and cultures. . The initial problem that this research addresses is the definition of the 110Tumilaca/ Cabuza and Chiribaya textile styles. Though the two cultures are clearly differentiated by their ceramics (Jessup 1991; Owen 1991a; Goldstein 1989), this is the first attempt to correlate these ceramic differences with the textile products of these two cultures. However, the scope of this research is broader. Based on the definition of each culture's textile repertoire, the textiles of each group are analyzed independently to address specific problems. The first is the problem of the origins of the two cultures. The second problem is to assess the relationships of both cultures with the Tiwanaku culture in tenns of the processes that occurred during the collapse of the Tiwanaku state, and the overall influence of

ANDEAN PAST 5 (1998):325-356.

ANDEAN PAST 5 (1998) this collapse on the remote middle and coastal valley. Third, the social and political structure both between and within the two coastal valley cultures will be addressed. Differences of quality, decoration, and motifs in textiles were recorded for each culture, and were analyzed in order to suggest individual status differences, as well as cultural identities. Finally, the apparent lack of gender differences in the textile record of both cultures will be discussed. The majority of the textiles analyzed, 395 fragments, came from the site of EI Algodonal (Tables 1 and 2). All of the textiles recovered from the cemetery of EI Algodonal are considered to be 110Tumilaca/Cabuza textiles. Every diagnostic ceramic vessel and sherd from this cemetery pertains to the 110Tumilaca/Cabuza tradition; not. one single Chiribaya sherd was found there (Owen 1991b); The cultural affiliation of this material is quite secure. Textiles from burials in the EI Algodonal habitation area were .assigned to one or the other culture according to intact ceramic ves~els found in situ in the same burials. Many of the textiles came from mummy bundles excavated from intact burials. A large number of textiles from EI Algodonal, however, were isolated pieces in various conditions of preservation collected from disturbed burials or severely looted surface areas at this site. The intact mummy bundles were unwrapped by Juana Lazo (Universidad Cat6lica "Santa Maria", Arequipa), Niki Clark (University of Chicago), and their associates prior to this project. The present research deals only with the textiles themselves, not the ways in which they were used to prepare mummy bundles.2 Other textiles came from the site ofChiribaya Alta (54 pieces) and Yaral (12 pieces) (Tables 1 and 2). The cultural affiliations of these textiles were based on the intact ceramic vessels found in the same burials as the textiles. All of these ceramics were examined
2 Research addressing the issue of mummy bundle construction will be carried out by Clark and Lazo.

-326

and identified by Owen and myself, using Jessup's (1991) seriation ofChiribaya ceramics. Methodology Each piece of textile was laid flat on a plastic supporting screen and mechanically cleaned with a soft brush. Each was photo-, graphed and then recorded on forms adopted from an earlier version that was used by Niki Clark for her research on the textiles from the middle Osmore site of Estuquifia (Clark ]988). Many pieces were illustrated with colored draWings. These drawings are not to scale, but a scale was included in the photos. Data recording emphasized types of patterns and decorations, using both drawings and written descriptions. Measurements of width and length of the complete article were taken when present or reconstructible, as were yam thickness and color, spin and ply directions (where evident), yam count, weaving and decorative techniques, and character of decoration and form (tunic or shirt, bag, etc.). After being recorded, each textile was wrapped in acid free paper and stored in a box containing all the textiles from the same provenience. To reduce curatorial problems, some extra large textiles were stored separately in a special large box. The data were coded and entered into dBase 111+ directly from the original recording forms, and data on the age and sex of associated human remains were added. Ages and sexes were provided by Shelly Burgess (also see Burgess n.d.). The computer database includes variables selected for their relevance to specific research questions. These variables and research questions include: 1. Presence or absence of mending was recorded as one possible measure of raw material availability, and as a possible indicator of economic and social status differences within and between different sites and cultures. 2. Textile area and density were measured as possible indicators of status, assuming that larger as well as denser pieces of cloth may indicate higher status (Roach and Eicher 1965; Murra 1989; Weiner and Schneider 1989).

327One density measure used is "crossing count", calculated as the number of warp/weft crossings per square centimeter. In practice this figure is the product of warps per centimeter times wefts per centimeter. Taken together, crossing and yarn thickness give a general measure of the density of the cloth. In the majority of cases, the measurement of the total original area was impossible to achieve because not enough of the textile was preserved to allow the extrapolation of the original size of the item. 3. Direction of spinning and plying (where evident) was recorded in order to detect po~sible cultural differences. It quickly became clear that no spinning differences between the
.

Boytner: Osmore Textiles 7. For decorations that contained geometric or naturalistic motifs, as opposed to simple color stripes, individual types of designs and motifs were defined and recorded (Tables 12 and 13). 8. This category was primarily used for the 110Tumilaca/Cabuza textiles, because the Chiribaya textiles from EI Algodonal were usually decorated only with color stripes. In the Chiribaya sample from Chiribaya Alta, on the other hand, many of the motifs were much more elaborate. Because the sample was small, no typology was established for these motifs, which were coded simply as "elaborated forms". A Comparison between 110Tumilacal Cabuza and Chiribaya Textiles The initial research problem was to identify differences betw~en the 110Tumilaca/ Cabuza and Chiribaya textiles. This analysis emphasizes .variables that might have been intended to indicate cultural differences, that is, traits that are easily visible on the textiles' surface. Shape, main weaving technique, finishing, and decoration are the primary variables considered. These components are the visible parts that are the end results of the intention and labor invested in the creation of each textile by the weavers of each specific culture. Most of the analysis focuses on tunics and bags because other forms are represented by very few examples. Tunics were the most prevalent type of textile among those that were identifiable (Figures 2, 5, 6, and 12; Table 1). The shape of tunics is consistent among the 110Tumilaca/Cabuza and Chiribaya tunics from EI Algodonal, Yaral, and Chiribaya Alta. The tunics are all sleeveless, with a slit neck, and range from square to trapezoidal in comparable frequencies in both cultures (Figures 2 and 3). Tunic size varies from 54 by 57 cm to as large as 120 by 107 cm. No specific tunic size could be associated with each culture. All but one tunic were single web products, folded and sewn along the warps, leaving holes for the arms. A neck slit was left unwoven and when the structure of the tunic was trapezoidal, warps were added in

two culturescouldbe observed.

4. Type offweaving technique was recorded for the main weaving surfac~ (as distinguished from weaving used for finishing or decoration). In rare cases where a variety of weaving techniques were used on one single textile, each individual technique was recorded. These data were collected in order to indicate possible cultural differences. 5. Finishing techniques were recorded by types and the specific location on the piece. Finishing techniques included technical and decorative aspects of salvages and seams. Locations on the piece included seams on the sides of tunics and bags, bottom edges of tunics and top edges of bags, edges of necks and arm openings, and other edges of other types of textiles such as panuelos (handkerchiefs), mantas (blankets), etc. These variables were examined not only for cultural differences, but also for differences by sex, age, and site (Tables 7, 8, 9, and 10) within and between each culture. 6. Wherever decoration was present, its appearance was recorded. The techniques used, the location on the piece, and the design pattern were all noted. A series of 17 categories (Tables 12 and 13) was developed to code these data, and certain categories were further analyzed as described below.

ANDEAN PAST 5 (1998) the shoulder area. Warps run length-wise on all tunics investigated. Bags were the second most abundant type of textiles identified (Figures 3 and 9, Table 1). Many individuals had a bag full of coca leaves buried with them (Owen 1991b), and some had two such bags. Bags were rectangular or trapezoidal in shape and had similar average sizes in both cultures. The same weaving technique utilized for tunics was used in bags. The pouch portions of bags are single web products; warps run vertically and are sewn on the sides leaving one end open as the mouth of the bag. When bags are trapezoidal, warps have been added to the lower section, making the bottom wider then the top (Figure 9). No cultural or gender differences in bag size or shape could be observed in the sample. It seems that the shape and size of bags were individual choices. Unfortupately, due to logistical problems, bag samples were available only from Algodonal and Chiribaya Alta. There is, however,. no reason to believe that the bags from Yaral would differ significantly. The most common technique used for the main weaving of the textiles is a plain weave 1/1 warp-faced technique (Table 6). In this technique, the warp yarns of the textile are the ones that are seen on the surface (VanStan 1958a; D'Harcourt 1962; Albers 1963; Collingwood 1987). The warp-face technique predominates in both the 110Tumilaca/Cabuza and the Chiribaya textiles. Some of the textiles were woven with a combination of weaving techniques including the warp-faced and simple plain weave techniques, among others. These textiles have low crossing counts (13-15 crossings per square centimeter, c.p.s.c.) and show a disorganized mixture of weaving techniques. These textiles apparently represent pieces made for purposes that did not require standard quality cloth, or by people who did not have access to material and/or labor resources available to other members of their society.

-328 Only seven of the textiles were woven completely in techniques other than warp face. These pieces are too few to indicate any intravalley cultural patterns, and they might equally represent exotic objects or unique individual preferences. Based on both yarn diameter and crossing' count (density), bags are generally woven finer than tunics, a characteristic found in both cultures. 110Tumilaca/Cabuza bags have wefts 0.8-1.0 mm in diameter, and warps from 0.6 to 1.0 mm in diameter, while Chiribaya bags have wefts between 0.8 and 2.0 mm in diameter, and warps 1.0 mm in diameter (only two Chiribaya bags were analyzed from El Algodonal). Ilo-Tumilaca/Cabuza tunics at EI Algodonal have about the same weft thickness as bags (0.96-1 mm.), but have 20% smaller warps (0.66-0.74 mm.), with an accompanying reduction in yam crossil}g(108-156 c.p.s.c. in bags, 91-101 c.p.s.c. in tunics) (Tables 3-4). The same phenomenon is evident among Chiribaya tunics from El Algodonal. Although only a small sample is available, there is little significant difference between the two cultures (Tables 3-4). Finishing techniques were sometimes used for decorative purposes in both the 110Tumilaca/Cabuza and Chiribaya traditions. Many tunics have elaborated finishing at the neck and the arm hole. These places were evidently the most stressed parts of the garment and tended to be the first to break down. For this reason, tunics were often reinforced at these points, and often this reinforcement also served as decoration. Even in many simple tunics, some extra weaving or embroidery was done at these points in more sophisticated ways than the main weaving itself. Less fragile parts of the tunic, however, such as the bottom or the sides, were decorated only if other parts of the tunics were decorated. Finishing was used occasionally for decoration at the sides of the tunics. In tunics of this type, the side seam is sewn together below the arm hole with one or more colored yams, and often with unusually fine yams. The bottom edges of tunics vary little and were generally. only

329reinforced for functional reasons. However, in a few cases the bottom edges of tunics were not only reinforced but were also embroidered to further increase this part's durability. The sides and the top part of bags were almost al~ ways done with decorated finishing, which formed an integral component of the overall decoration of these objects (for range of finishing techniques and frequency of use see Tables 7-10). Cultural differences in finishing types are not clear. Ilo-Tumilaca/Cabuza and Chiribaya people from the different sites used all the various finishing techniques, and weavers of each culture applied each technique to some location on their textiles (Tables 7-10). There were preferences for using specific techniques on specific locations on textiles in each culture at each site (Tables 7-10). I attribute these differences, how~ver, to the small sample of Chiribaya textiles from El Algodonal, and.the even smaller sample of Ilo~ Tumilaca/Cabuza textiles at Chiribay.aAlta and Yaral (Tables 1 and 2). A clearer idea of patterns in decora~ tive finishing techniques will be available only when the full analysis of the Chiribaya Alta, Chiribaya Baja, and Yaral textiles is completed. The important point is that no specific finishing type or technique can be exclusively identified with one or the other culture. Cultural differences in the textiles are evident only in the most obvious and intentional of their features: their decoration. It is difficult to identify differences in any of the other categories examined. However, cultural distinctions are very clear in the types, locations, and techniques of decoration on the textiles. 110Tumilaca/Cabuza tunics are typically decorated with an embroidery technique that creates geometric designs in two different locations on the tunic. The most visible and common is a narrow horizontal stripe directly under the neck (Figure 4, Table 11). This stripe is composed of five to seven adjacent loop-stitch embroidered lines usually no more than 10 centimeters long. The yams compos-

Boytner: Osmore Textiles ing each line vary in colo.-3to create geometric designs. The most dominant color is red, a color that during the Late Horizon was commonly used among the Inca to indicate high status (VanStan 1958b:53-54; Murra 1962:719; Uhle 1903). The lines also contain blue, green, and yellow. The geometric designs are always presented in a "double feature" format, with each side decorated with modified "M" motifs, solid triangles, elaborated S shapes, or combinations of these (Figure 4). Usually this "double feature" format was laid out in an almost, but not quite, symmetrical pattern. Minor but discernable differences make each side distinct from the other, creating a "pseudo-symmetrical" design. The designs consist of both vertical reflection' and 1800 rotation or pmll and p1l2 respectively according to Ascher (1991:160). Adding the color scheme, a design both vertically reflects itself in differe:gtcolors, or pm'II, of the main body, and also has vertical rotation, or p'll2 of the added triangles (ibid.: 164-165). The other major location for decoration on Ilo-Tumilaca/Cabuza tunics was done on the sides, starting under the arm hole and ending at the bottom, covering the seam that joins the front and back panels of the tunic. This stripe usually comprises 5 to 11 loop-stitch embroidered lines done in the same range of colors as the ones used for the under-the-neck stripe. These side stripes are typically 20 to 100 centimeters long, depending on the length of the tunic. The geometric designs include ones similar to those used under the neck, plus additional motifs found only on side stripes (Figure 11). In this case, because the stripe is much longer, the design pattern is allowed to repeat. Unlike the two sides of the neck stripe, the side stripes on a given tunic appear to be identical to each other, or vertical reflection pmll (Ascher 1991:160). This may be because the stripes are long, narrow, and placed widely apart from each other, so that even if they were different, the observer might not notice the asymmetry. All the tunics that

3 For analysisof dyestuff from this textile collection, see Boytner and Wallert 1993; Wallert and Boytner 1996.

ANDEAN PAST 5 (1998) have side decorations also have neck stripes, but some tunics have neck stripes only. Another way that the Ilo-Tumilaca/Cabuza people decorated their tunics was to place series of divided stripes along the two vertical exterior sides of the tunic (Figure 5). Each of these divided stripes is comprised of a wide stripe with a narrow line of a contrasting color running down the center. A wide series of these divided stripes, in different color combinations, is placed on the two sides of the tunic. This decoration usually occupies about 2/7 of each side of the tunic, leaving the center area plain. The stripes are not done in the embroidery technique, but rather by weaving dyed warp yarns to create the lines as part of the primary structure of the fabric. At El Algodonal, this kind of decoration was found only in Ilo-Tumilaca/Cabuza contexts.. However, this decoration was also found in Chiribaya contexts at Chiribaya Alta and I suspect that it was associated with high status and prestige privileges. . Chiribaya tunics were decorated differently from Ilo-Tumilaca/Cabuza ones. Based on associated ceramics, four Chiribaya tunics were recovered for El Algodonal (Table 1). One of these tunics is undecorated and plain. The other three, however, are decorated in a distinct and similar style, unlike the one seen on the Ilo-Tumilaca/Cabuza textiles. The limited number of tunics from Chiribaya Alta investigated here (20 tunics) are similar to the Chiribaya tunics found at El Algodonal, which suggests that the. El Algodonal sample is a reasonable representation of Chiribaya material. The only decorative technique commonly used on Chiribaya tunics at El Algodonal was done with colored warp yarns that form stripes in the primary structure of the cloth. This decoration results in long colored stripes on the sides of the tunic, starting at the shoulder and ending at the bottom edge. On each side of the tunic, there is a narrow line in one color, usually 1-2 centimeters wide, and a wide line of a different color, usually 3-7 centimeters in width (Figure 6). On a few tunics, this pair of stripes is delimited on each side by a narrow

-330
line (2-4 millimeters) composed of two to three colored warp yarns. The symmetry of the design is completely different from those on the Ilo-Tumilaca/Chiribaya. It is done in glide reflect, or pial (Ascher 1991:1960), which presents a completely different concept of spatial organization of the design. Bags were more elaborately decorated than tunics. Unlike the tunics, there is little variation in bag colors; they are either brown or red. Brown bags were always made of what seems to be natural brown wool with a few narrow vertical colored bands. Red bags are usually red overall with the same type of vertical colored bands. Typically, each bag is decorated with three vertical, wide bands containing various designs (Figure 9). The designs are done with supplementary warps in red, blue, green, white, brown, and yellow. The motifs used on the stripes are geometric and repeat along the vertical axis (Figure 8).' The stripes are symmetrical and each stripe is similar in its decoration to the others. Also, the back and front of the bag are identical (although differences in the front and back decoration of bags have been noted in many bags scattered on the looted surface of Chiribaya Alta). The differences between Ilo-Tumilaca/ Cabuza and Chiribaya bags at El Algodonal are slight. All bags from Ilo-Tumilaca/Cabuza contexts have linear patterning (Figure 8 left), and only one bag from a Chiribaya context had different, triangular patterning (Figure 8 right). Both cultures used the same supplementary warp technique to create the band motifs. Because only one example of the triangular pattern was found, it is unclear if it reflects a real cultural difference. Many such bags can be seen on the looted surface of Chiribaya Alta; only future research will be able to assess these differences. The differences in decoration on IloTumilaca/Cabuza and Chiribaya tunics might be attributed to one of two possibilities. The first possibility is that there was a chronological difference between the two cultures. In that case, the distinct decorations would be

331because the two cultures never met. However, recent radiocarbon dates from sites in the valley indicate a substantial length of time during which 110Tumilaca/Cabuza and Chiribaya people coexisted in the valley (Owen 1992). Moreover, 110-Tumilaca/Cabuza and Chiribaya burials were found in the same habitation area at El Algodonal (Owen 1991b), in the same cemeteries in Yaral with no stratigraphic differences between them (Lozada 1991; Lozada et ai. 1991), and probably also in the same cemeteries in Chiribaya Alta, although these.data are still preliminary. Also, there are 5 textile pieces from Chiribaya Alta, 3 from looted surface contexts and 2 from tomb excavations, all of which have 110Tumilaca/ Cabuza and Chiribaya style decorations on the same piece. This mixing of decorative techniques might be .expected among contemporary cultures. The technical similarity in textiles, including weaving technique, yarn thickness, yarn density, spin, and size and shape of tunics, bags, and other forms also suggests that the two cultures coexisted, at least for a limited time, in the'lower valley. Another possible explanation for the differences in the tunics of the two cultures is that they were used to mark ethnic differences. As the largest single piece of cloth worn by these people, and possibly the one mostly commonly worn, tunics were well suited to use as cultural identity markers. Ethnohistoric material suggests that clothes were important tools for marking cultural identity, status, and roles in Andean societies. Textiles were so important that a naked enemy was considered to be powerless and not a threat (Murra 1962; Murra 1989:280; Clark 1988:1). Moreover, "no political, military, social, or religious event was completed without textile volunteered or bestowed, burned, exchanged, or sacrificed" (Murra 1989:293). Although the 110Tumilaca/Cabuza and Chiribaya tunics share many similar traits, it is easy to distinguish between them because the differences are found in obvious, visible traits that were easily manipulated to serve explicit message functions. Wobst (1977:333) suggested the importance of size and location of symbols used to transmit cultural messages.

Boytner: Osmore Textiles The large size of the Chiribaya stripe motifs and the location of the 110Tumilaca/Cabuza neck stripe on the high central part of the tunic makes them both easy for an observer to distinguish, and fit well with Wobst's observations concerning stylistic elements that are used to communicate ethnic affiliations: These two cultures must have interacted continually, and because no evidence of warfare was found in the valley, they evidently interacted in a peaceful manner (Owen 1992). Nevertheless, the two groups probably competed for .control over resources, mainly water, which were severely limited in the narrow valley. Hodder (1979:450) suggests that in times of economic stress, cultural differences between groups are often highlighted. These differences are manifested as a stronger emphasis on characteristic material culture symbols, making the boundaries between cultures particularly distinctive. At the same time, one possible response to difficult times is to increase cooperation between the .groups in order to enhance the total efficiency of resource use from a given area. While the level of cooperation is increasing, each participant will expand his or her presentation of, and emphasis on, symbols of cultural identity. He or she will try to exaggerate these symbols in order to make them as distinct as possible from those of the other culture. The situation in the Osmore Valley seems to have worked in a similar way. The two cultures coexisted in the narrow valley, where resources were limited and the environment was not similar to the areas from which they had come (probably the middle Osmore Valley). Coping with limited means for survival, cooperation formed between the two cultures, probably with each developing its own technicalor economical specialization. Parallel to, and because of, this situation, distinct cultural identities were highly emphasized, and so the use of individual cultural markers was important. The size of the designs that comprise the textile differentiation, especially the 110Tumilaca/Cabuza markers, suggests another

ANDEAN PAST 5 (1998) important aspect of the two cultures' interaction. An observer could identify the cultural affiliation of an individual only from a distance of a few meters or less. The implication is that in general there was no need for early reconnaissance when individuals from the two cultures approached one another. Individuals could approach one another without threatening or being threatened. Wobst (1977:322) has suggested that ethnic markers visible only at such short range imply interaction, rather than aggressive conflict, between the groups. In this. sense, the textile data support the conclusion that the relationship between the 110Tumilaca/Cabuza and Chiribaya cultures in the valley was one of cooperation and peaceful coexistence.

-332 were the same in both cultures, which may indicate that they have a common origin. Camelids and camelid wool were clearly of great importance to both' the 110Tumilaca/Cabuza and Chiribaya people of the coastal Osmore Valley. Masses of raw' camelid wool were found in some burials, sometimes on the bottom surface of the tomb, and sometimes inside the mummy bundle. Many of the burials contained camelid bones, most often the cranium, mandible, and/or the foot bones (Owen 1991b; Lozada 1991; Lozada et al. 1991). Although all the textiles examined from burial contexts were made of wool, excavations in the habitation areas of El Algodonal and other coastal Osmore sites frequently encountered bolls, seeds, yarns, and textile fragments made of cotton in midden contexts, so cotton was clearly available and used by the people of El Algodonal (Owen 1992: personal communication). Their exclusive use of wool textiles in burial contexts is unusual among Andean coastal cultures, where cotton yarns were often incorporated with wool yarns for ceremonial, as well as utilitarian textiles (Murra 1989:153; Rowe 1986:153). The fact that the vast majority of yarns used were single yarns S-spun is difficult to explain. S-spun yarns typical of the North Coast tradition are usually cotton warps (Rowe 1984:89). Moreover, this type of yarn construction is different from Tiwanaku textiles and other Late Intermediate textiles found in the region, which are typically Z-spun 2Splied (Conklin 1983; Lothrop and Mahler 1957; Oakland 1986). Is it possible that the 110Tumilaca/Cabuza and Chiribaya cultures both adopted North Coast yarn preparation technology? The answer for this question is still far from clear and this investigation suggests no explanation for the phenomenon. This unusual emphasis on wool, and the similarities between the textiles of the two cultures, suggest that both 110Tumilaca/ Cabuza and the Chiribaya traditions derived not from the coast, but rather from the highlands, where camelids traditionally played a more central role in the economy and ideology. The similarities in textiles suggest a close relationship between the two cultures,

Evidence for Highland Origins

Andean textiles typically embody great ideological power, and in order to preserve this power, textile manufacturing methods and decoration designs tend to be conservative (Conklin 1983:20;' Frame 1986:52; Murra 1962:713,721). Rene-e, textiles can be efficient tools to assist in determining geographical origins. Comparing textiles of suspected immigrants with textiles from areas where they are thought to have resided for long periods of time may indicate cultural relationships or place of origins. As the 110Tumilaca/Cabuza and Chiribaya people were cooperating in the valley, the need for clear and immediate distinction was evident. Tunic decorations were highly emphasized as the primary medium for exhibiting cultural identity. Once the goal of distinctiveness was achieved, further distinctions were unnecessary, and the other textile traits remained the same. Clear status differences within each culture probably existed, but cannot be observed in the investigation of the textiles from El Algodonal. The brief analysis of the textiles from Chiribaya Alta clearly indicates intra-cultural social differences (see below), but only a complete analysis of this site's textile assemblage will identify the textiles characteristic of each social class. Bag decoration techniques and forms, main weaving techniques, and other textile characteristics

333perhaps even a common origin. It is clear, however, that at the time when the two cultures are first archaeologically recognizable in the valley they were already distinct from each other. Based on the textiles alone, it is impossible to select a precise point of origin in the highlands, but the textiles are certainly consistent with other archaeological data that suggest that at least the 110Tumilaca/Cabuza culture derives directly from the terminal Tiwanaku tradition of the middle Osmore, near Moquegua. 110TumiiacalCabuza, Chiribaya, and the Tiwanaku State Ideology Our understanding of the nature of relationships between the Chiribaya and Tiwanaku cultures is still vague. A distinctive tunic that casts some light on the ideological relationships of the 110TumilacafCabuza people and the Tiwanaku state was recovered from a looted burial at El'Algodonal. This tunic is unique in its decoration, and is the only piece examined that has depictions of an anthropomorphic figure. The figure is done with supplementary warps on stripes vertically decorating the plain weave, 1/1, warp-faced tunic. The image of the figure alternates with an image. of a square within a ~quare done in the same technique. Between two figurine/square stripes, there is another color stripe similar to the one used for decorated bags done with supplementary warps without any apparent design (Figure 13). This figure is different from any other 110Tumilaca/Cabuza or Chiribaya motif encountered. Although this figure lacks the usual staves and the lines radiating from its head, its face is depicted in a frontal view and not in profile. Its general appearance and temporal context suggest that it represents a simplification of a Tiwanaku deity figure found on earlier Tiwanaku textiles from the region (see Conklin 1983, especially T7 on p.15). The weaving technique used to depict the deity on this tunic is similar to the supplementary yam technique used in decorated bands of 110Tumilaca/Cabuza and Chiribaya bags. In addition, pa,rallelto the stripes bearing the depiction of this deity are other stripes

Boytner: Osmore Textiles with decorated bands identical both in technique and color to those used on bags. This tunic, together with the general similarities in ceramic motifs, suggests that the people of the coastal Osmore Valley knew the Tiwanaku ideology and were familiar with the former Tiwanaku center in the middle Osmore valley, at the site of Omo. Although the presence of the Tiwanaku culture in the middle valley is clear (Goldstein 1989), the relationship between the ideology of Tiwanaku and those of the 110Tumilaca/Cabuza and Chiribaya cultures is still unknown. Clarifying these relationships will increase our understanding not only of political and social processes in the valley's history, but also of the processes and consequences of the collapse of the Tiwanaku state. There are interesting contrasts between Tiwanaku textiles and coastal Osmore textiles in their treatments of symmetry and perspective. In the 110Tumilaca/Cabuza tunic decorations, as well as on the Chiribaya ones, there is always a slight distortion of symmetry. Dividing the tunic on a vertical axis from the neck down, the two sides are never symmetrical mirror images. On the contrary, they are distinctively different from side to side. In 110Tumilaca/Cabuza decoration, the horizontal decorative stripe below the neck slit is always slightly asymmetrical along its central vertical axis (Figure 4). In Chiribaya decoration, the sequence and width of color stripes on the tunics is always different on each side. Each side has the opposite combination of color and stripe width of the other (Figure 6). This phenomenon might not be immediately obvious to the observer, but the overall impression of the decoration is not balanced. This ~ame idea was executed with more so,. phistication in tunics from Chiribaya Alta, where changes of color sequence and width did not occur from side to side, but rather from front to back, with the change in stripe patterns occurring at the top of the shoulder (Figure 12). This type of change required a more sophisticated technical mastery of weaving, because the change at the shoulder was done along a single weft yam. Even if we put the tunic on its side and we look for the "Horizon phenomenon" as one possible way of weaving,

ANDEAN PAST 5 (1998) and thus observing,4 the tunic (Conklin 1986:125), the symmetry of the design is clearly broken. Neither the characteristic of symmetry nor that of the "Horizon effect", both typical of Tiwanaku textiles (Bird and Skinner 1974; Conklin 1970, 1983, 1986), are evident in either the 110TwnilacaJCabuza or the Chiribaya textiles. Both of the coastal Osmore groups replaced the most notable Tiwanaku organizational principles with the distinctly different concept of "pseudosymmetry". A few Chiribaya tunics found at Chiribaya Alta embody yet another variant of the "pseudo-symmetry" organizational principle, or the glide reflect pial (Ascher 1991:160). These tunics have a horizontal band of zigzag decoration under the neck done in continuous zigzag embroidery (Figure 7). This decoration probably originated from the functional need to reinforce this part of the tunic, as described earlier. . It is almost the same size as the typical 110Tumilaca/Cabuza loop-stitch embroidery stripe, and shows a comparable "pseudosymmetrical" organization. The stripe is effectively cut along its vertical center line, and the sequence of colors inverted, so that the top of one side is depicted on the bottom of the other side, and vice versa. Yet, this band is not similar in its shape or size to the 110Twnilaca/Cabuza band, and can be clearly distinguished from it. It had no boundaries, no complicated geometry, and simply utilized the "pseudo-symmetry" rule in its most simplistic way. The motifs' selected for each decoration were not random choices of geometrical de4 Conklin (1986) suggested that the correct way to look at South Andean Middle Horizon textiles should be sideways because this is the way they were woven. The concept of looking into the horizon, where the wider lines represent areas closer to the observer, and thinner lines represents areas far from the observer is useful here. When a textile was constructed to make a garment, however, instead of presenting the lines in horizontal mode, the garment was constructed to present the lines in vertical mode, making it difficult for the untrained eye to notice the view-to-the-horizon concept depicted on it.

-334

signs (Weiner and Schneider 1989:1). Both cultures selected motifs widely used in the Andean region at the time (tenth and eleventh centuries A.D.), including S shapes, triangles, and squares (see Figures 4, 8, 10, and 11) (Frame 1986:55; Rowe 1986:156-157). These motifs were probably part of a pan-Andean ideology common to many cultures in the region at that time. But why did the 110-' Tumilaca/Cabuza and Chiribaya people choose. not to adopt Tiwanaku insignia after the collapse of the Tiwanaku state? Were not the symbols historically recognized as illustrating power and prestige, representing widespread and powerful ideology? Why did the people of the valley simply. adapt the symbols of the collapsed entity to their own use, utilizing an already-known set of symbols to indicate social status and economic power? Goldstein (1989) found evidence that suggestS deliberate destruction of many of the structures and burials at the Tiwanaku center at Omo. I suspect, with Bermann et al. (1989: 162), that the absence of the standing deity and Tiwanaku condor was part of the same process. Both the people in the middle valley and those in the coastal Osmore Valley rejected the Tiwanaku ideology as such. It is possible that the overall structure of the ideology persisted, because many decorative motifs typical of the Tiwanaku repertoire remained in the Ilo-Tumilaca/Cabuza and Chiribaya textiles. The symbols such as the standing god and the eagle that were associated directly with Tiwanaku dominance, however, simply vanished from Ilo-Tumilaca/Cabuza and Chiribaya textiles and ceramics, and the traditional Tiwanaku forms of symmetry were deliberately broken. It is currently not completely clear how the populace of the middle Osmore Valley related to the Tiwanaku state around A.D. 950. Their apparent rejection of the explicitly Tiwanakurelated aspects of the ideology, however, may indicate that there was some conflict between them and the representatives of the Tiwanaku state in the valley. Evidence from Omo (Goldstein 1989) suggests that the Tiwanaku temple at the site was deliberately destroyed by the local 110Tumilaca/Cabuza people.

335Rejection of Tiwanaku dominance was reflected, then, not only by demolishing power centers, but also by abolishing the symbols of their power. The Tiwanaku icons served not as a source for power among the valley's people, but as reminders of being subject to powers they did not welcome. It seems that the act of rejection was so powerful that it was remembered by the valley people immigrating to lower elevations, who never readopted Tiwanaku design modes. Intra-Valley Social Stratification Inter-site differences in the textile assemblages strongly suggest hierarchical status relationships between the sites, and, indirectly, among individuals. For example, there are dramatic differences in the labor and material expended on bags from El Algodonal as compared with those from Chiribaya Alta. The yarns used for bags from Chiribaya Alta. are 40-60% thinner than those from El Algodonal (0.8-2 mm. at El .Algodonal versus 0.39-0.7 nun at Chiribaya Alta), while the density of the textiles was higher at Chiribaya Alta (108156 c.p.s.c. at El Algodonal versus 170-194 c.p.s.c. at Chiribaya Alta) (Tables 3-4). Considering that the bags from both sites are similar in size, it appears that people buried at Chiribaya Aha had bags that required significantly more time to weave. Comparing the tunics from El Algodonal to those from Yaral, it is evident that the tunics from Yaral are coarser and less dense (Tables 3-4). While the shapes of the tunics from Yaral are similar to those from El Algodonal, their crudeness suggests that the people of Yaral could not afford as much weaving time and investment in raw materials. It seems that the Yaral people were poorer than the people of El Algodonal. As was the case with bags, the tunics from Chiribaya Alta are finer and denser than those from El Algodonal and Yaral. Although there is not much difference in weft thicknesses (0.75-1 mm at El Algodonal, 0.87-2 mm at Yaral versus 0.6-1.98 mm at Chiribaya Alta), Chiribaya Alta tunics have much thinner warps (0.6-0.74 mm at El Algodonal, 0.69-1.7

Boytner: Osmore Textiles at Yaral versus 0.3-0.49 mm at Chiribaya Alta), and much higher yarn densities (15-101 c.p.s.c. at Yaral and EI Algodonal versus 112167 c.p.s.c. at Chiribaya Alta). Because of their high density, tunics from Chiribaya Alta required substantially longer periods of time to weave. Unidentified fragments of cloth from El Algodonal have about the same yarn thickness and density as the tunics, suggesting that the general quality differences can be extended to much of the El Algodonal textile assemblage. The textiles from Chiribaya Alta were also more finely finished than those from 1 Algodonal and Yaral, and greater amounts of work were invested in finishing. The use of more than one color, particularly what seem to be dyed colors, was much more common at Chiribaya Alta than at El Algodonal or Yaral (Tables 7-10). Although the finishing of bags from Chiribaya Alta is more elaborate than that of bags from El Algodonal and Yaral, the differences are only slight. Bags of both cultures were evidently important objects in burials. They

contained coca for the dead, and they were


always highly decorated.. Because bags are relatively small, they required comparatively little wool and labor to make. Even poor people could evidently afford to have relatively elaborate bags (sometimes more than one) to accompany them in death. In some cases, bags were mended in various places. Despite being tattered, they were very elaborate in their weaving and decoration. These bags were probably used and repaired by their bearers for a long time because they could not afford to acquire new ones. At 1 Algodonal, as well as at Yaral, only a small fraction of the textiles are decorated, and those that are have only minimal amounts of decoration. Some of the decorated textiles were used for a long time, and were mended before the person was buried. This was not the case at Chiribaya Alta. In burials of both cultures at Chiribaya Alta, the decorative techniques were the same as at El Algodonal and Yaral, but the density was higher (Tables 5 and 11). Ilo-Tumilaca/Cabuza tunics from

'

ANDEAN PAST 5 (1998) looted surface contexts at Chiribaya Alta have side stripes up to 2S loop-stitch rows wide, although the neck decoration is about the same as at El Algodonal. The designs in these stripes are more elaborate, and involve complex geometrical forms (Figure 10). Although the range of colors is the same and red is still the dominant color, the greater area of decoration and the use of more colors in any given stripe allowed the tunics to create a stronger impression. The relative elaboration of decoration at Chiribaya Alta is even greater in Chiribaya textiles. Although the same colors are used, many Chiribaya Alta tunics have the additional complication of color sequences that change at the top of the shoulder, and many have elaborate zoomorphic and abstract motifs,most commonly a lizard figure, but also including serpents, felines, and birds. These motifs are executed in the same supplementary yarn technique that is used on bags. This technique must have increased the cost of .the tunic, because it requires not only more colored yarns, but also a completely woven surface below the decoration, and hence larger amounts of wool for the same size tunic. Such sophisticated decoration also implies access to more skilled labor. Colored yarns were evidently valuable, because in 90% of both bags and tunics, only the warp yarns are colored. These articles look completely colored, even though only some of their yarns were colored. This was an efficient way to minimize costs without sacrificing appearance. Unfortunately, it is difficult to estimate the time required to make these textiles. As shown by Franquemont in his ethnographic research (1984:322-323), weaving time is highly variable, depending on the loom and spinning technique, the age of the weaver, the time of year when the weaving was done, and the quality of wool. Nevertheless, Franquemont's work suggests that these textiles would have taken hundreds of hours to produce. Whatever the exact time was, the time spent to make the Chiribaya Alta textiles would have been greater than that for equiva-

-336 lent size El Algodonal and Yaral textiles, because the decorated areas are larger and the decorative techniques more complex. Finally, there is a greater variety of textile types at Chiribaya Alta than at El Algodonal. Hats, "handkerchiefs" (panuelos), belts, bagbelts ifaja-bolsas), and other types that are. rare at El Algodonal are common at Chiribaya Alta. All in all, the textile evidence seems to reflect the location of Chiribaya Alta above the valley floor, on top of the steep slopes, far from water resources, and surrounded by fortification walls. This apparently was a central site and elite cemetery for the valley. The clear site-level status differences in the textiles suggest that both cultures were socially stratified. The site of Chiribaya Alta was used by the elite of both cultures, and it is possible that the two cultures shared not only the same site hierarchy and geographical environment, but also the same social structure. The textiles alone do not reveal the specific nature of the social hierarchy, nor the exact relationships between the two cultures. Further research integrating the textile record with other elements of the material culture may clarify this issue. Gender Differences The tunics of both cultures come in rectangular and trapezoidal varieties. Clark (1988:14) suggested that rectangular tunics were worn by men and trapezoidal tunics were worn by women at the Late Intermediate period site of Estuquifia in the upper Osmore drainage. At El Algodonal, however, there was no evidence of differentiation in tunic shape by gender at any of the investigated cultures. This may be due to the poor quality of preservation and the small number of complete tunics with individuals of known sex from El Algodonal. Both the 110Tumilaca/ Cabuza and the Chiribaya sample suffer from these limitations. There appear to be no significant differences between bags buried with males and females of the 110Tumilaca/Cabuza culture at El Algodonal. No differences in yarn thick-

337ness of wefts or warps were observed (Tables 3-4). Bags buried with males were slightly denser than those buried with females (Table 5), but the 10% difference in density is probably not significant given the large standard deviation of the sample. No gender preferences were detected in finishing types in either culture. Nor are there any evident gender differences in decorative motifs at EI Algodonal (Tables 12-13). The sample of Chiribaya decorative motifs is too small to assess gender differences, but at least among the 110Tumilaca/Cabuza people, there were evidently no specific textile motifs that were strongly associated with either gender. Overall, the textiles do not suggest any gender preferences in weaving techniques, finishing, or even decoration. It. is, however, important to note that the number of individuals of known gender from El Algodonal is small, and it is possible that such preferences simply cannot be detected in:this sample. The apparent absence of gender differences is particularly surprisinR because Jessup (1991) reported that in his excavations in Chiribaya cemeteries, he found different ceramics in burials of males and females. Societies in which males and females share the same status are not unknown in the Andes, but they are not common. Only future investigation of textiles from other coastal Osmore Valley sites would resolve this problem and will either support or contradict the above observation. Conclusions The textiles analyzed here comprised only a small segment of the cultural universe of the 110Tumilaca/Cabuza and Chiribaya people. The conclusions offered serve best to bolster other lines of evid~nce, and to suggest hypotheses for future research. One such conclusion is that textiles point to a common highland origin for both 110Tumilaca/Cabuza and Chiribaya cultures. The absence of Tiwanaku deities in textile decoration is significant, and combined with the evidence of intentional destruction of the Tiwanaku center at Omo, can be interpreted as a rejection of Tiwanaku ideology. The great similarity of the

Boytner: Osmore Textiles textiles from the two cultures strongly suggests cooperation and coexistence in the coastal Osmore Valley. Nevertheless, the inherent differences in the symmetrical organization of designs provide clear evidence of the distinctiveness of each cultural identity. Ascher (1991) claims that although there is no necessary functional explanation for the rela., tionships between the spatial design organization and the cultural interpretation of the world, this relationship does exist. Moreover, it reflects distinctive and particular cultural sets of ideas and beliefs. Thus, the differences in the organization of design in the two cultures reflect the differences between them, and provide an indication that each is a distinct and individual cultural system. Two important issues remain unanswered. First, what is the true nature of the relationship between the 110-Tumilaca/Cabuza and Chiribaya cultures? How is it that these two cultures engaged in such close relations, while so strongly maintaining ahd emphasizing their individual identities? Second, there is the question of how and why the 110Tumilaca/ Cabuza culture disappeared from the valley. It is possible that the competition within the symbiotic relationship of the two cultures finally ended the 110Tumilaca/Cabuza people's ability to control resource~, forcing them outside the valley. The absence of signs of warfare and the lack of evidence of diffusion or blending suggests that the 110Tumilaca/ Cabuza people probably left the valley peacefully, rather than integrating into the Chiribaya society. This is, of course, only one possibility, and should be addressed using the full range of available archaeological data. Although the main purpose of this research was to define differences between the two coastal valley cultures, it also helped define other questions. It addresses issues regarding 110-Tumilaca/Cabuza and Chiribaya ideological and political structures. Furthermore, it examines a process of change in these structures that occurred by the end of the Middle Horizon. By looking at its margins, we may better understand the collapse of the Tiwanaku state and the events that followed the destruction of central authority. Once we understand

ANDEAN PAST 5 (1998) these processes in the lower Osmore Valley, we may apply our model to other regions of the Tiwanaku state and examine the results. Eventually, we may be able to achieve a better understanding of the general ideological and political processes of the Andean region. Acknowledgements I wish especially to recognize Bruce Owen's assistance in conducting my research and in writing this report. There was no limit to his willingness to share data and experience with me at any time and place. I do, however, bear sole responsibility for the interpretation of these textiles. Thanks go to Timothy Earle and Christopher Donnan for their constructive criticism and helpful advice in the various stages of this research. Thanks are also due to Jane Buikstra for granting access to material from Chiribaya Alta and Yaral, as well as to Shelly Burgess, Marla Cecilia Lozada, Liliana Ulloa Torres and to the rest of the staff at the Azapa Museum of the University of Tarapaca at Arica, Chile. Niki Clark, Cecilia Quequezana, and Manuel Patcheco deserve thanks for their enthusiasm i~ sharing data, and for their interest and dedication to the understanding of Andean prehistory. I would especially like to acknowledge Patricia Anawalt and Barbara Sloan of the Center for the Study of Regional Dress at the Fowler Museum (UCLA), and Katherine Spilker of the Los Angeles County Museum. Southern Peru Copper Company (SPCC) provided valuable support for this research. Lastly, I would like to offer my deepest appreciation to Karen Marshik who dedicated long, patient hours to correcting and re-correcting many drafts of this paper. I could never thank her enough. References Cited
Albers, Anni 1963 On Weaving. Middletown: Wesleyan University Press. Ascher, Marcia 1991 Ethnomathematics: A Multicultural View of Mathematical Ideas. Pacific Grove, California: Brooks/Cole Publishing Co. Bermann, M., P. Goldstein, C. Stanish, and L.M. Watanabe 1989 The Collapse of the Tiwanaku State: A View from the Osmore Drainage. In Ecology, Settlement and History in the Osmore Valley, Peru, edited by

-338
D.S. Rice, C. Stanish, and P.R. Scarr, pp. 269-286. BAR International Series 545ii. Bird, Junius B. and M.D. Skinner 1974 The Technical Features of a Middle Horizon Tapestry Shirt from Peru. Textile Museum Journal 4(1):5-11. Boytner, Ran and Arie Wallert 1993 Dyes from the Tumilaca and Chiribaya Cultures, South Coast of Peru: What Can We Learn? Paper presented at the 33rd Annual Meeting of the' Institute of Andean Studies, Berkeley, California. Burgess, Shelly D. n.d. Human Remains: Preservation and Demography. In Mortuary Investigations at Chiribaya Alta: A Late Intermediate Period Site Near 110, Peru, edited by Sloan Williams and Jane E. Buikstra (iq press). Clark, Niki R. 1988 Archaeological Textiles in Their SocioCultural Context: Late Prehistoric Period, Estuquifta Site, Far South Peru. Paper presented at the 16th Midwestern Conference on Andean and Amazonian Archaeology and Ethnohistory. Ann Arbor, Michigan. Collingwood, Peter 1987 Textile and Weaving Structures. London: B.T. Batsford Ltd. . Conklin, William J. 1970 Peruvian Textile Fragment from the Beginning of the Middle Horizon. Textile Museum Journal (I): 15-24. 1983 Pucara and Tiahuana.so Tapestry: Time and Style in Sierra Tradition. Nawpa Pacha (21):1-44. 1986 The Mythic Geometry of the Ancient Southern Sierra. In The Junius Bird Conference on Andean Textiles, edited by Ann P. Rowe, pp. 123-135. Washington D.C.: The Textile Museum. D'Harcourt, Raoul 1962 Textiles of Ancient Peru and their Technique. Seattle: University of Washington Press. Frame, Mary 1986 The Visual Images in Fabric Structures in Ancient Peruvian Textiles. In The Junius Bird Conference on Andean Textiles, edited by Ann P. Rowe, pp. 47-80. Washington D.C.: The Textile Museum. Franquemont, Edward M. 1984 Cloth Production Rates in Chinchero, Peru. In The Junius Bird Conference on Andean Textiles, edited by Ann P. Rowe, pp. 309-329. Washington D.C.: The Textile Museum. Goldstein, Paul S. 1989 Omo, A Tiwanaku Provincial Center in Moquegua, Peru. Ph.D. dissertation, Department of Anthropology, University of Chicago. Hodder, Ian 1979 Economic and Social Stress and Material Culture Patterning. American Antiquity 44(3):446454. Jessup, David A.
,

1991 General Trends in the Development of the ChiribayaCulture,South CoastalPeru. Paperpre-

339sented at the 56th Annual Meeting of the Society for American Archaeology, New Orleans. Lothrop, S.K. and Joy Mahler 1957 A Chancay-Style Grave at Zapallan, Peru: An Analysis of its Textiles, Pottery, and other Furnishings. Papers of the Peabody Museum of Archaeology and Ethnology 50( 1). Lozada, Maria C. 1991 Mortuary Excavations at La Yaral, Southern Peru. Unpublished Manuscript. Lozada, Maria C., E. TOlTes,and Jane E. Buikstra 1991 New Chiribaya Funerary Patterns at La Yaral, Southern Peru. Paper presented at the 90th Annual Meeting of the American Anthropological Association, Chicago, Illinois. Murra, John V. 1989 Cloth and Its Function in the Inca State. In Cloth and Human Experience, edited by A.B. Weiner and J. Schneider, pp. 275-302. Washington: Smithsonian Institution Press. Oakland, Amy 1986 Tiahuanaco Tapestry Tunics and Mantles from San Pedro de Atacama, Chile. In The Junius Bird Conference on Andean Textiles, edited by Ann P. Rowe, pp. 101-121. Washington D.C.: The Textile Museum. Owen, Bruce 1991a Colonization and Complexity in the Twilight of Tiwanaku: The Coastal Osmore Valley, Peru. Paper presented at the 56th Annual Meeting of the Society for American -Archaeology,New Orleans. 1991b Proyecto Colonias Costeras de Tiwanaku, 1989-1990. Field report. 1992 Coastal Colonies and the Collapse of Tiwanaku: The Coastal Osmore Valley, Peru. Paper presented at the 57th Annual Meeting of the Society for American Archaeology, Pittsburgh. 1993 A Model of Multiethnicity: State Collapse, Competition, and Social Complexity from Tiwanaku to Chiribaya in the Osmore Valley,.Peru. Unpublished Ph.D. Dissertation, Department of Anthropology, UCLA. Roach, M.E. and lB. Eicher (eds.) 1965 Dress, Adornment, and the Social Order. New York: John Wiley and Sons, Inc. Rowe, Ann P. 1984 Costumes and Feather Work of the Lords of Chimor. Washington D.C.: The Textile Museum. 1986 Textiles from the Nasca Valley at the Time of the Fall of the Huari Empire. In The Junius Bird Conference on Andean Textiles, edited by Ann P. Rowe, pp. 151-182. Washington D.C.: The Textile Museum. Uhle, Max 1903 Pachacamac. Department of Archaeology of the University of Pennsylvania, Philadelphia. VanStan, Ina 1958a Problems in Pre-Colombian Textile Classification. Tallahassee: Florida State University.
American Anthropologist (62):710-728.

Roytner: Osmore Textiles


1958b A Peruvian Tasseled Fabric. Notes in Anthropology 3. The Florida State University, Department of Anthropology. Wallert, Arie and Ran Boytner 1996 Dyes from the Tumilaca and Chiribaya Cultures, South Coast of Peru. Journal of Archaeological Science 23:853-861. Weiner, Annette B. and Jane Schneider 1989 Introduction. In Cloth and Human Experience, edited by A.B. Weiner and J. Schneider, PI>'. 1-29. Washington, D.C.: Smithsonian Institution Press. Wobst, Martin H. 1977 Stylistic Behavior and Information Exchange. In For the Director: Research Essays in Honor of James B. Griffin, edited by C.E. Cleland, pp. 317342. University of Michigan, Museum of Anthropology, Anthropological Papers 61.

1962 Cloth and Its Function in the Inca State.

ANDEAN PASTS (1998)

-340

1"';:':1Padftc

_ Figure 1. The Osmore valley (from Owen 1991a).

ocean

Arable floodplain Osmore river

341-

Boytner: Osmore Textiles

Figure 2. Variation of shapes among tuni~s.

neckslit top ~

top
.1

,
band (stri :)

under-the-neck stripe

-- -...

+-

side

side~

I
bottom

I
bottom

,\

Figure 3. Key to specific area locations and terminology used in the text: left, tunic; right, bag.

ANDEAN PAST 5 (1998)

-342

mainobject

triangles

Figure 4. Variations of under-the-neck Ilo-Tumilaca/Cabuza (I-TIC) decoration.

(Left) Figure 5. 110Tumilaca/Cabuza divided line decoration on tunic (Right) Figure 6. Chiribaya tunic decoration.

343-

Boytner: Osmore Textiles

Figure 7. Under-the-neck zigzag decoration.

TIC version, right is the Figure 8. Variations of stripe motifs in decorations of bags; left is the IChiribaya version.

ANDEAN PAST 5 (1998)

-344

Figure 9. Variati0I?-s of stripe width in bags.

Figure 10. Sample variant of elaborated side decoration of tunics from Chiribaya Alta.

345-

Boytner: Osmore Textiles

Figure 11. Variations.of! -TIC side decoration on tunics.

..

Figure 12. Changes in stripe width on the shoulder in Chiribaya tunic from Chiribaya Aha.

ANDEAN PAST 5 (1998)

-346

111111

III

II III III

III III III III III III III II III

11./

III III

III III III III II III

II\ IIIIII
III III III

II\ II)III
III 1\/ll'

[]

IJ

Figure 13. The tunic with the front face deity found on the looted surface at El Algodonal.

w
EL ALGODONAL CHIRIBA YA ALTA

~ j.J

YARAL I-TIC
M F N/A TO CHIRIBAYA M F N/A TO

I-TIC
SEX M BAG TUNIC
UNKNOWN

CHIRIBAYA M 2 4 2 3 2 3 3 1 1 F N/A TO 2 4 4 3 3 6 3 M 1

I-TIC
F N/A TO M 8 2

CHIRIBAYA F. 3 NIA TO 9 18 20 20 1

F 2 18 19 4 9 4 2

N/A TO 7 41 90 7 27 4 3 1 15 76 125 13 46 12 S 13

6 17 16 2 10 4

-I

2-PLY YARN RE-PLmDYARN ROPE


HANDKERCHIEF (PANUELO)

4' 1 4

7 1 4 1 2 2

OTIIERS

12

EL ALGOOONAL EXCAVATION LOOTED SURFACE

CHIRIBAY A AL TA

YARAL 24

99 218

54

~ ~ :'!

i
~
(Top) Table 1. Textile types found on each site. (Bottom)Table 2. Collectiontypes.

~
:::to

~ c.)

EL ALGODONAL I-TIC SEX BAG N SIZE TUNIC N . SIZE UNKNOWN N SIZE M 6 0.97 14 0.96 12 1.01 F 1 1.0 15 0.97 13 1.11 N/A 3 0.8 34 . 1.0 62 097 M 1 0.8 2 0.7S CHIRIBAYA F N/A 1 2.0 1 .1.0 3 0.6 M

CIURIBAYAALTA I-TIC F N/A M 2 0.7 8 1.98 CHIRIBAYA F N/A 8 0.39 18 0.72 1 2.0 M I-TIC F

YARAL CHIRIBAYA N/A M F N/A

i
(fl

~
too3 (II

-. 2 1;5 7 0.87

\0 \0

EL ALGODONAL

CIURIBAYAALTA

YARAL I-TIC M F NIA M


CHIRIBAYA

I-TIC SEX BAG N SIZE TUNIC N SIZE


UNKNOWN N SIZE

CHIRIBAYA

I-TIC M F N/A M 2

CHIRIBAYA

M 6 0.93 .14 0.72


12 0.73

F 1 1.0 15 0.66
13 0.97

N/A 3 0.6 34 0.74 62


0.71

M 2 1.0 2 0.6

N/A 1 1.0

N/A 8 0.25 18 0.49

N/A

0.55 1 0.8 3 0.3 8 0.41

1 1.7

2 1.0

7 0.69

(Top) Table 3. Weft thickness (average thickness in millimeters). (Bottom) Table 4. Warp thickness (average thickness in millimeters).

~ 00

EL ALGOOONAL I-TIC SEX BAG N SIZE TUNIC ,., SIZE


UNKNOWN CIDRIBAYA

CIURIBAYA ALTA . I.T/C M F NlA M 3 170 I 112 1 187 3 167.3 5 127


c:IURIBAYA

01:>\0 I

YARAL I-TIC M F NIA M


CIURIBAYA

M 5 156.2 16 91.9 IS 86.6

F 2. 108 17 101.2 16 89.5

N/A 7 144 34 97.9 86 98.8

M 2 112.5 4 83.2 2 71.5

N/A 2 82.5

NIA 7 194 17 164

N/A

I 15

I IS

7 99.1

N SIZE

EL ALOOOONAL

CHIRIBAYA ALTA

YARAL I.T/C I
CHIRIBAYA

I.T/C
WARP FACE

CHIRIBAYA

I-TIC 3 1 6

CHIRIBAYA

105 23 12 1
1 4 1

9 3

35 I

7 4

WARPFACE+
REINFORCMANT WARP MIXED WI WARP FACE

MIXEDW/OUT WARP FACE


BASKET (212) SEMI BASKET (211) INTERCOURSING

~ :"!
5?

~ ~

(fop) Table 5. Yam count (weft x warp per centimeter). (Bottom)Table 6. Main weavingtechniques.

~
:::to

~ ~

EL ALGODONAL I-TIC SEX M NO WEFT FINISH


NO WARP FINISH

CHIRIBAYA I-TIC M F NIA TO M

ALTA CHIRIBAYA F N/A TO M


'

YARAL

CHIRIBAYA TO 1 16 1 1 1 M F N/A .TO

I-TIC
F N/A TO

CHIRIBAYA M F N/A TO

i
~
4 1 3
\0 \0 00

N/A 1

~ rn

7 1

12

17

3 1 2

CA

REGULAR
EMBROIDERY

1 2 1 1 1 1 1 S 2 8 2 2 3 2 2 4 1 2 3 1 1 1 1 S 6 1
"

'-"

DENSE
EMBROIDERY LOOP STITCH: ONE COLOR LOOP STITCH: MULTI COLOR

BOUBLELOOP STITCH:COLOR
IN PATTERN

BOUBLELOOP STITCH:COLOR
IN TRAINGLE LOOP STITCH STRIPE PART OF

FINlSH
EMBROIDED

LINES IN SLlVEINECK

Table 7. Sleeve finishing techniques.

I W U\ o

w
I

VI ..... EL ALGODONAL CHIRIBA YA ALTA

YARAL I-TIC
TO 4 M F N/A TO CHIRIBAYA M F /A TO 22

I-TIC
SEX M NO WARPFINISH 3 REGULAR EMBROIDERY DENSE EMBROIDERY LOOP STITCH. ONE COLOR LOOP STITCH.
MULTI COLOR

CHIRIBAYA TO 10 M 2 F N/A TO 2 M

I-TIC
F 1 N/A TO l. M 2

I-TIC
F N/A 2

F 3

N/A 4

1 6

2 8

3 8 2

6 22 2 4 1 1

1 1

1 1

DOUBLELOOP
STITCH-COLOR IN PATTERN

DOUBLELOOP
STITCH-COLOR

IN TRIANGLES
LOOP STITCH

STRIPEPART OF FINISH
EMBROIDED

2
,

LINES IN SLEEVEINECK

~ ~ :"!
~ ~ ~

Table 8. Neck fmishingtechniques.

~ ::t -: ~

EL ALGODONAL

CHIRIBA YA AL TA

YARAL I-TIC
TO 4 25 1 1 1 3 M F N/A TO CHIRIBAYA M F N/A TO 1 7 1 10

I-TIC
SEX M
NO WARP FINISH

CHIRIBAYA
F 1 6 1 N/A TO 1 6 1 1 M

I-TIC
F N/A TO M

I-TIC
F N/A 4

i
~

F 3 19 3

N/A 34 41 1 2

TO 41 74 4 4

4 14

REGULAR
EMBROIDERY

8 1

17

--

f:/.! IooiJ

(.A
..... \0 \0 00'

DENSE
EMBROIDERY LOOP STITCH-

ONE COLOR LOOP SmCHMULTI COLOR OOUBLE LOOP

smCH-COLOR
IN PATTERN OOUBLE LOOP

SmCH-COLOR
IN TRIANGLES

LooPSmCH STRIPEPARTOF FINISH

10

Table 9. Side fmishingtechniques.

, w u. t-.)

w
I

UI W EL ALGODONAL I-TIC
SEX NO WEFT FINISH WEFT REINFORCMENT WEFT REINFORCMENT +LOOP S'ITfCH WEFT REINFORCMENT +EMBROIDERY DENSE EMBROIDERY LOOP STITCHONE COLOR LOO SMCHMULTI COLOR DOUBLE LOOP STITCH-COLOR IN PATTERN 4 4 1 1 1 . 1 18 27 46 91 6 1 1 8 2 '2 8 1 19 28 1 1 3 M F N/A TO M

CHIRIBAYA I-TIC
TO M F N/A TO M

ALTA I-TIC
F N/A TO M

YARAL I-TIC
F N/A TO

CHIRIBAYA
F N/A

CHIRIBAYA
M F N/A 1 2 TO 1 S

IS

19

~ :'!
& ~ ~

Table 10. Shirt bottom and bag top fmishingtechniques.

~
:::to

~ ~

EL ALGODONAL I-TIC
SEX WEAVED COLOR STRIPES M 2 F 3 N/A 22 TO 27 M 5

CHIRIBAYA I-TIC
TO 5 M 1 F N/A TO 1 M 7

ALTA CHIRIBAYA
F N/A 8 TO 15 M

YARAL I-TIC
F N/A TO

CHIRIBAYA
F N/A

CHIRIBAYA
M 1 F N/A 4 TO 5

DIVIDEDLINE IN
SEQUENCE EMBROIDERY UNDER.THE-NECK ONLY EMBROIDERY. SIDES+NECK EMBROIDERYNARROW NECK+WIDE SIDES BAG STRIPESEQUAL WIDTH BAG WlTII WIDER CENTRAL STRIPE BROCADE

I'

18

23

i ~ ...(I) too3 <.18

\0 \0 00

13

Table 11. Decoration techniques and location on cloth.

w
NECK I-TIC.
SEX M F 2 N/A TO 1 4 1 1 M CHIRIBAYA F N/A TO M SIDE

VI VI I

STRIPE
CHIRIBAYA

I-TIC
F N/A TO M

I-TIC
M F N/A TO

CHIRIBAYA M F N/A TO

N/A TO

roE A /\n..
TYPEBAyA. TYPE C
ED TYPE E

2 1

'

HH

e.ra
..

"

1 1 1

1 4

1 4 5

TYPE F ....m TYPEG TYPE A+B TYPE E+F


ELABORATED

tit

1.1 1 1 1 2 3

13 ..

r
1 4 1 2 2 4 1 4 2 .1 3 1

1 2

FORMS ZIG.ZAG ZIO ZAG WI


CHANGING COLORS

REINFORCEDLINES UNDERNECK

b:I

ii
:"!

& ~ ~ Table 12. Embroidered motif types by sex ~d culture at EI Algodonal.

~ :::t ~
~

NECK I-TIC
SEX M nuN COLORLINE ON THE SIDE WIDESTRIPES-NO
CHANGE WIDE STRIPECHANGE BETWEEN SIDES DIVIDED LINE PATTERN SEQUENCE 4 1 18

/"

SIDE CHIRIBAYA I-TIC


M F 2 2 1 N/A TO 15 7 M

STRIPE CHIRIBAYA
F N/A TO M

I-TIC
F N/A TO

CHIRIBA YA M F N/A TO

i
(I)

N/A TO

N/A TO

F
10
S 5
./

.tit

-23 1 1

\0 \0 00

Table 13. Woven motif types by sex and culture at El Algodonal.

I ~ VI 0\

CORBEL VAULTED SODSTRUCTURES IN THE CONTEXT OF LAKE TITICACA BASIN SETTLEMENT PATTERNS Sergio J. Chavez Instituteof AndeanStudies,Berkeley CentralMichiganUniversity
Introduction This article is an introduction to the study of a specific kind of corbel vaulted structure known as putuku or phullu uta, and a variation of this called taklla putuku.\ The corbel vaulted design and the use of sod (ch 'ampa) as building material employed in its construction provide an excellent alternative to the use of timber for beams in this environment where appropriate trees are absent, and allows an efficient response to the persisten~ and severe climatic conditions of this region. The present-day distribution of these little-known structures in the northern Lake Titicaca Basin of Peru includes the Province of Huancane in the Department of Puno, for example along the Taraco-Huancane road where this investigation was conducted (Figure 1). Two settlements were selected within this area, the town of Taraco, capital of the District of Taraco, and the community of Yanaoqo, permitting comparison between the kinds of putuku found in an urban and in a rural setting, respectively. Both sites lie within the same altiplano environment and are about 16 km apart. Since 1968, while conducting archaeological reconnaissance and excavations in the area with Karen Mohr Chavez, the author has investigated putuku, their materials, and the present-day settlement patterns of which they are an integral part. In July, 1985, and ,again in August, 1988, more detailed investigations were conducted and interviews were made, including with builders, such as the one observed in the process of constructing a putuku. Aside from presenting a description of the different architectural attributes, tools and building materials employed in construction, and the different uses given to these one-room structures, interpretations are proposed aimed at understanding: (a) the relationship between environmental conditions (for example, lack of trees, topography, low soil absorption, and rainy sea.son inundation) and the presence of corbel vaulted putuku using local resources; and (b) how differences in building materials (stone, tapia, adobe, and ch 'ampa), in form, in function, and frequencies of putuku and other structures provide architectural indicators of status, whether social, economic, and/or political, in a nucleated urban settlement (Taraco) and in a dispersed rural one (Yanaoqo). Finally, comparisons are made with a different kind of corbel vaulted sod structure in the southern end of th~ basin, and the historic and

1. Although our infonnants in the region say the word putuku is Quechua, no such word is found in relevant dictionaries and may Itot be a Runasimi (Quechua) expression. Mesa and Gisbert (1966:492-493) and Gisbert (1988: 124-125 and figures on pp. 132, 140) report similar corbel vaulted structures also known as putuku, belonging to rural Chipaya in Oruro, Bolivia. These one-room' structures, however, are circular in plan rather than rectangular as found in the northern portion of the Titicaca Basin. The closest architectural tenn similar to the word putuku is phutu, reported to be an Aymara word meaning "niche" (Gasparini and Margolies 1980:133, 343), or phuthu meaning "a hole" (Apaza Suca et al. 1984:166). Bertonio's 1612 Aymara vocabulary glosses putu as "edificio de bobeda" ("vaulted building") (Bertonio 1956 [1612]:120). However, Guaman Poma de Ayala in his Second and Third Ages of Indians describes and illustrates a vaulted structure called pucu//o, which he defmes as "a house" ("cacitas que parece homo"); and elsewhere also as funerary constructions, and burials for Indians of the Colla Suyos (Guaman Poma 1936 [1615]:53, 54, 56,59,69, 188,259,289-291,296, and 298). On the other hand, the second tenn mentioned by our infonnants that also refers to the corbel vaulted structures is phu//u uta. This composite tenn is defmed by Apaza Suca et a/. (1984:237) as Aymara meaning: Phu//u. Manta pequefia para la espalda (prenda de mujeres). Uta. Casa. Utachana. 1. Construir una casa. 2. Techar una casa. Utachayaifa. Hacer construir 0 techar casa. Repajar una casa.
Utachkatana. Construir una casa junto a otra.

ANDEAN PAST 5 (1998):357-408.

ANDEAN PAST 5 (1998) prehistoric background is given, along with Ii discussion of archaeological implications for the region. Tapid

-358

The Use of Stone, Tapia, Adobe, and


.

Ch'ampa as Basic Local Building Materials Stone Stone in Taraco is not only extensively employed for house foundations where it is required, but also is utilized to cover patios, as door portals or jambs, for the curbing of some streets, and in the large steps and platform in front of the church. Stone is accessible both because of the economic ability of the residents of Taraco to obtain and transport it, and because of its relative abundance at or near Taraco itself. For example, worked and unworked slabs and blocks of different shapes and sizes belonging to prehistoric occupations have been found at or near the surface in Taraco and are reused today.2 Furthermore, th~re are also stonecutters (picapedreros) in Taraco (Martinez 1962:62; Ghersi and Arquinio 1966:38). . In contrast, the use of stone in Yanaoqo is limited tQthe very few people who can afford to have stone foundations in their gabled oneroomed adobe structures. Rarely, unworked stones are also used as layers at the base of corrals built of sod blocks called ch 'ampa (Figures 19-20). We also observed a household cluster near Taraco with three putuku enclosed within four tapia walls, that had three courses of stone at their base (Figure 2). Stone is rare and expensive here. The only major quarry nearby is a hill called Cupisco just north of Yanaoqo, where a kind of pinkish banded quartzite is commercially exploited, making the material beyond the economic reacl).of most Yanaoqo residents.

Tapia, as our informant at Taraco indicated, is an economical alternative to adobe, requiring less labor and time. The elaboration of tapia begins with the preparation of earth using shovel and pick, to which water is added, to form mud. This first step requires a few days of soaking which is carried out near the site where the tapia will be used. With a shovel, the soaked earth is then mixed to achieve a homogeneous mass of mud. Subsequently, this mud fills and is tamped into a long rectangular wooden mold having the thickness of the wall. The mold is already situated directly on the wall under construction. As the first tapia block is drying, the mold is taken apart and rebuilt to continue with the horizontal layer. The same operation is repeated, with each new layer resting directly on top of the previous. one without mortar. We observed one case in Taraco in which medium-sized stones were unevenly distributed between each horizontal row of tapia, and only at the inner and outer surfaces of the wall (Figure 23).4 The use of tapia blocks in Taraco is extensive in enclosure walls (cercos) that delimit individual property lots aI?-dstreets (Figures 22-23), and in the lower portions of some putuku (Figures 24, 28). In Yanaoqo, especially in places along the road to Huancane, tapia is also used in walls enclosing household clusters (Figure 2). These walls, however, appear to be built more frequently of adobe and ch 'ampa blocks (Figures 9, 20), and as far as we can confirm, tapia is never used in Yanaoqo to build putuku or gabled structures.

2. See, for example, Kidder (1943:17). Archaeological research carried out by us in Taraco indicates a long sequence of occupations going back to at least 600 B.C., and many of the carved slabs,.stelae, and statues relate to the Yaya-Mama and Pucara styles (Kidder 1943:16-18, plates III-V; S. Chavez and K. Chavez 1970:32-36; K. Chavez 1977:9, 263-264, 1026, 1064, 1091).

3. Tapia is a prepared mix of mud, straw, and water rammed into molds to make walls.. 4. In Cajamarca (north highland Peru), Karen L. Mohr Chavez noted in July, 1985 the frequent occurrence of tapia structures with rows of medium-sized stones between each layer as illustrated by Muelle (1978: figures 5-6); Muelle's figure 6 clearly shows that stones are positioned at the outer wall surfaces only.

359Adobe The initial steps in the preparation of adobe are similar to those described above for tapia. As the mud is ready, however, straw is added by dispersing it onto the surface and walking over it to mix the two. The kind of straw used is a native grass called chilliwa, which is cut into segments of about 20 cm long. Subsequently, this mixture of mud and straw is put into a single, four-sided rectangular wooden mold (adobera), and the upper surface is evened out by hand. As this process is repeated for each adobe, the wooden mold is constantly dipped in water to prevent the next adobe from sticking to it. Additional time, space, and periodical turning over of the adobes are required to dry them and ready them for use. During the construction of permanent structures, each horizontal course of adobe has to be leveled and aligned, and, mud mortar must be applied between each course. Finally, walls are plastered with mud, a process which is periodically repeated over .the years. Adobe is considered'to be the most expensive building material, compared to tapia or ch 'ampa, because it requires more time, space, and labor in the manufacturing j>roc,ess: Most of the buildings in Taraco, including the church of San Taraco, are made of adobe (Figure 22); in very few cases adobe is also used in wall enclosures. Furthermore, there are some putuku which are either built entirely of adobe' or in combination with tapia (Figure 26). In contrast, all of' the gabled one-room structures in Yanaoqo are built entirely of adobe (Figures 8, 10), and none of the putuku we visited were of adobe. Instead, the putuku is made entirely of sod blocks or ch 'ampa. In addition, adobe is also employed in walls and enclosures, although always between the uppermost and lowermost course of ch 'ampa (Figure 9). Ch'ampa Ch 'ampa refers to a rectangular block of sod characteristically containing sod of a specific kind of low-growing native grass (called

Chavez: Corbel Vaulted Sod Structures

kimillu or dimillu), with long and profusely distributed roots, or sometimes also having grama and layu grasses, the latter a variety of high-altitude ichu. Terrains where this type of grass grows are highly valued because they have to be flat (located on a pampa), undisturbed, and must possess a homogeneous and lush growth on rich dark farming soil (hence harvesting ch 'ampa is in potential conflict with agriculture). Such an area exists south of Yanaoqo and was the source for the ch' ampa the master-builder was using in the new putuku (see information on Yanaoqo, below). Informants in the community of Collana (north of Taraco) mention a test which is sometimes carried out to determine the quality of ch 'ampa. This test involves soaking a sample in water for about four days, followed by drying it for four additional days, at which time the ch 'ampa block will crumble when its quality is not appropriate or suitable for construction. This low quality ch 'ampa is associated with sandy soils, .little top soil, and/or contains mostly grama grass. The labor needed to extract the sod blocks is still obtained through the traditional mink'a system used on communal lands (1<.Chavez 1987:179-180). Once a specific location is chosen, the maestro marks on the ground the length and width of each block of ch' ampa to be removed. He uses string and two measuring sticks cut from the straight stems of the domestic kiwfla plant; one stick is for the length and the other is for the width of the blocks. The tool used to cut out the blocks is the traditional foot-plow (chaki-taklla), that is inserted obliquely into the ground by hand, rather than by foot (Figure 4). Blocks are cut in parallel adjacent rows. As each block is removed, the underside is evened out with the foot-plow, and the block is turned over (Figures 5-6). The moisture then begins to evaporate and the block quickly dries in the characteristic high solar radiation of the region. The end result is a series of heavy rectangular blocks having a consistent size and trapezoidal cross-section (Figures 7-8). The dimensions of ch 'ampa blocks we observed during construction in Yanaoqo were about 53 by 42 by

ANDEAN PAST 5 (1998) 13 cm, and two larger blocks, to be used as a lintel were 70 by 41 by 13 cm. The importance and effective use of ch 'ampa have not been properly documented, and to some extent ch' ampa has been ignored in the literature where more attention has been given to adobe and stone as traditional building materials. Our observations in Yanaoqo (and to some extent in other areas of the altipiano), however, indicate that ch 'ampa is efficiently and extensively used to solve many structural problems that include the following: (1) The construction of enclosures of sizes varying from 1 m to 1.50 m high (Figure 19), and longer walls marking land boundaries. (2) . Ch 'ampa may be used in combination with adobe for added protection -against rain and ground moisture. A large wall 2 m high enclosing the school at Yanaoqo was built by first laying down four sucGessive layers of ch 'ampa on the ground, then adding eight layers of adobe and mortar, and finally placing two layers of ch'ampa on top (Figures 9, 20). The placement of ch ;ampa layers above and below adobe ones protects the entire structure from the annual rain and floods. Ch 'ampa is more resistant to such persistent conditions, and hence is better than adobe alone. We also observed that, unlike adobe structures, ch 'ampa ones do not always require plastering of the exterior. Furthermore, exposure to rain and floods actually helps in some way to settle firmly and to seal the layers, and sometimes encourages the green regrowth of naturally occurring grass in the sod at the first course of ch 'ampa near the ground (Figure 3).5 (3) Ch 'ampa is used as an economic alternative in the building of the corbel vaulted putuku (Figures 12-18), as well as one-room thatched and gabled structures and smaller ones for keeping dogs and chickens. Further5. Describing the sod houses in Nebraska, Welsch (1973:71) indicates that the heavy walls settled six to eight inches in the first year or two, while at the window and door where the load was much less, settling was only two to three inches. This example provides an independent confirmation of the settling of sod-built structures.

-360 more, when our informant in Yanaoqo was questioned about the differences between adobe and ch 'ampa structures, he indicated that although the use of adobe allows a larger interior space, a ch 'ampa-made putuku is better because it will last longer, resist severe flooding better, and provide a far more com-. fortable interior environment during extreme variations of temperature. 6 (4) Ch 'ampa is used to elevate terrain prior to the construction of rooms, to level the floors within structures, to create causeways between household clusters in temporarily inundated areas so that communication is maintained (Figure 3), and to build large causeways in flooded areas as in prehistoric times (Julien 1989:53-54, figure 11). Water canals can be banked with (5) ch 'ampa, which are also used to form the fill of convex surfaced raised field planting platforms (Erickson 1989:14, figures 6, 9).
_

There are no permanent structures of any kind in Taraco in which ch 'ampa blocks are used. In the community of Yanaoqo, however, the use of ch 'ampa is widespread. The following chart summarizes, in order of preference and frequency, the use of these basic local building materials (see also Table 1): TARACO
YANAOQO

Most frequent to leastfrequent

1. Adobe 2. Tapia 3. Stone

1. 2. 3. 4.

Ch'ampa Adobe Tapia Stone

6. Welsch (1973:71) also states, based on interviews with former sod house dwellers in the United States Great Plains, that unlike frame houses, these structures were well suited to severe weather and wind conditions in the prairies, as well as resistant to fire, rot, and damage by rodents and insects. Gabled sod houses, or the earlier dugout dwellings (see for example, Dick 1979: figures facing pages 112 and 116), were also built without mortar, with the grass side facing down, and were plastered on the interior with a mixture of clay and fme sand. Each sod block was about 90 by 60 by 8-10 em.

361Ch 'ampa appears to be a word in Runasimi as well as Aymara (Apaza Suca et al. 1984:41). The dictionary published in 1608 by Gon~ales Holguin (1952:93, 469) defines and describes the use of ch 'ampa as follows:
Chhamppa. Sod with roots Chhampa chhampa huay/la. Sod-cutting place. Chhamppani, 0 chhamppahuan pirccani. Make a wall or enclosure of sod. Chhamppani 0 chhamppaycuni chhampahuan harccani. Cover a canal with sod [blocks]. Cesped. Turf, and champani, Make enclosure walls~

Chavez: Corbel Vaulted Sod Structures

Yanaoqo and its Rural Characteristics8 Yanaoqo is one of many communities in the area whose primary activities include agriculture, followed by trade and some craft production. Yanaoqo is on a pampa between two main rivers, the Huancane on the east and the Ramis on the southwest, that eventually join and flow into Lake Titicaca (Figure 1). The pampa, having an elevation of 3810 m, is bordered rather abruptly on the north by Yanaoqo Hill with a maximum elevation of 4145 m. Lagoons and rivers almost entirely surroUnd the pampa, creating peninsulas, marshes, and temporarily inundateci places. On the other hand, these water sources provide the inhabitants with resources including fish, totora, and lake algae called llachu, as well as with routes of communication. Furthermore, this area possesses a complementary distribution of resources and utilization: the eastern half is mainly cultivated with potatoes,. barley, and kiwna, and the non-cultivated portions of this half are characteristically covered with native grasses. The western half is an inundated marsh zone adjoining small lakes where cattle feed on llachu. People on wooden and totora rafts were observed cutting totora and llachu
7. Chhamppa. Cesped de tierra con rayzes. Chhampa chhampa huayl/a. Lugar para sacar cespedes. Chhamppani, 0 chhamppahuan pirccani. Hazer pared de cespedes, 0 cerca. Chhamppani 0 chhamppaycuni chhampahuan harccani. Tapar acequia con cesped. Cesped. Champa, y champani, hazer cercas. 8. Due to its close geographical proximity to Taraco, the environmental conditions as well as agricultural production described for Taraco, also apply here.

using a kinina, a pole about 4 m long with a metal blade attached to one end. The totora has multiple uses including roof thatching, fuel, and food for people (the last 25 cm near the root constitutes the edible portion called chullu). Cutting totora by leaving about 25 cm of the stem encourages regrowth in about one month. Depletion of this resource is also avoided by transplanting it into new OI sparsely occupied areas. Llachu and a thinner . variety called chinki are exclusively used as cattle feed. It should be noted that many informants admitted not having seen either llamas or alpacas. Most of the domestic animals are the European-introduced cattle, sheep, pig, donkey, and chicken. Native domesticates observed include guinea pig, muscovy duck, and dog. An important climatological and hydrological factor in this region is the periodic rise in lake and Ramis River levels, which causes inundation of extensive residential and agricultural areas (see also section on Taraco below). Monheim (1963:'19, 101) indicates that according to local Indian informants, major lake inundations occur every 15-20 years, and that when he visited this area in March of 1954, Lake Arapa had temporarily joined Lake Titicaca. In 1986 severe inundation, especially in the Province of Huancane, extended up to 15 km inland and caused 30% loss of the potato, kiwna,' and barley production and covered roads so that boats were the only mode of transportation(La Region 1986: 2, 8, and 9). The rising of lake levels had also displaced the floating islands of the Uru people and covered the totora reed plants (Aquize Jaen 1986). Additional related environmental factors here include the low gradient of the terrain, slow water permeability, and the formation of flat river margins. Furthermore, sporadic droughts also occur. For example, it has been estimated that during the severe drought of 1943, the lake level dropped some 6.30 m. The settlement pattern here, as well as in similar rural areas, is dispersed, consisting of widely separated household clusters (Figure 21). Sometimes household clusters are adjacent where close relatives are involved. Aside

ANDEAN PAST 5 (1998) from lake shores or foothills, the boundaries of these settlements are difficult to determine visually (Figure 21). There are no streets, plazas, or public buildings except for one school. Each household cluster occupies slightly elevated terrain, and sometimes these elevations extend like causeways between households to connect them in the rainy season. Each household cluster shelters a. family unit and is enclosed by four walls about 1.50 m high. There are usually three separate and independent one-roomed structures facing an open courtyard. The backs of the structures may interrupt and form part of the enclosure wall. Household clusters may contain only putuku (Figures 15, 16), putuku and one story (very rarely two story) gabled structures (Figures 10, 18), or only gabled structures, depending on the socio-economic condition of the occupants. Of the two kinds of structures, gabled ones require more time and labor to construct and necessitate the use of expensive building materials brought from distant places. They are rectangular in plan and require stone foundations that, when possible, sometimes extend about 40 cm above ground to avoid water damage (Figure 10). The vertical walls are made of adobe bricks with mud mortar. Additional characteristics include: padlocked carpentry-made wooden doors, glass windows, wooden lintels and roof beams, and plastered exterior and interior. It w~ observed that the gabled structures may be divided into two groups according to the quality of additional building materials: (a) those which are constructed mostly of local or nearby resources: these structures are built of adobe, or even eh 'ampa as observed especially in the district of Plateria, south of Chucuito, have thatched roofs of either iehu grass or totora secured with woven rope of iehu grass, are mud plastered inside and outside (Figure 18), and sometimes have a piece of glass set directly into an opening made in the adobe without a frame. Although the thatched roof repels water and provides a comfortable interior environment, it requires periodical re-

-362 newal. For example, Arquifiiba (1978:601) reports that the iehu roof lasts one year, while a roof which is sewn and secured with rope also woven of iehu can last two to three years. (b) Gabled structures that use more expensive and distant materials: these structures have corrugated iron ("tin") roofs and glass windows set into welded metal frames; only these' structures may have two stories and may be painted white or light green, with blue window frames and doors (Figures 10, 18). Furthermore, we were informed that these "tin" roofed houses are not well adapted to the climate of the region, because they tend to absorb the sun's heat during the day and are cold at night. The roofs produce noise from rain and hail, and can also rust, requiring eventual replacement. Within this basically subsistence economy, however, the price of metal roofs is very high, and hence their presence is associated with, or reflects, the family's high social and economic status within the community.9 We observed that some of these gabled rooms were used to store bicycles, machine-made clothing, eh 'arki (dried meat) hung from a rope line, grains in pottery vessels, and varieties of freeze-dried oqa, ulluku, and potatoes kept in cylindrical containers called seje that are made from totora mats. The structure known as putuku is designed for multiple uses (see the section below, A Household Cluster in Use). It requires only readily available local resources. This unique architectural solution to the lack of timber in the altiplano has a reputation among the natives of the region as producing a more stable structure well adapted to the severe climatic conditions. The putuku, made entirely of sod called eh 'ampa, consists of two superimposed shapes (Figure 17). The lower half has a rectangular plan and four inwardly slanted walls. Each is hence trapezoidal in elevation. The upper half is a conical corbeled vault. The small trape9. Hector Martinez (1962:78) also observed in 19571958 that metal roofs were just being introduced into these rural areas, and that their use indicated prestige, economic capacity, and acculturation.

363zoidal door is elevated above the floor and is enclosed by an additional entryway that extends to ground level. This entryway is added to the exterior of the door and may also be trapezoidal. The inner door is not secured, other than by a piece of sheet metal nailed down to a wooden frame, or by a piece of cloth hanging as a curtain, for protection against the winds at night, or to keep out domestic animals. However, elsewhere in this region we have observed the use of doors made of wood and cow hide. Interior ventilation is achieved by the following architectural attributes: two small smoke vents opposite one another at the apex, small vents through the upper conical vaulted portion of the structure, and small vents in the trapezoidal walls usually near the floor. On the lower portion of the structure there are also one or two niches or recesses carved into the wall and some wooden pegs for hanging things. The interior walls are mud-plastered on the rectangular lower portion only. Without exception in the corbel vaulted structures we obse~ed at Yanaoqo, there was a consistent arrangement of the bed, oven, and grinding stone areas. As one enters the putuku, the bed is to the right of the door, the earthen or pottery oven (q 'oncha) to the left of the door, and the rocker mill (maran, or flat base, and tunau, or hand-held rocker) on the left side of the wall opposite the door. Construction Technique of Corbel Vaulted Structures in a Rural Setting The information in this section derives from an intensive interview conducted with Mariano Quispe Bautista, one of two master builders of putuku in the community of Yanaoqo. We were fortunate to observe him and three of his assistants constructing a putuku on July 17, 1985, allowing us to document and ask pertinent questions about the building process (Figures 10, 12-14). The number of ch 'ampa blocks used in the putuku we observed being built was about 600. According to Mariano, a putuku may have as many as 1000 ch 'ampa, indicating that such a structure could be almost double the

Chavez: Corbel Vaulted Sod Structures size of the one described here. 10 Removal of the approximately 600 blocks was accomplished in three days with four people working. The transportation of these blocks to the site where the putuku was to be built required three trips by truck. We were informed that according to the distance involved, donkeys can also be used, although each would carry only two blocks at a time. Both males and females can carry blocks by themselves or by using a wheelbarrow. Faustino Huanca Cusi, owner of a household cluster situated west of the school and on the road to Huancane, wanted to add a putuku to the cluster (Figure 10). Through the mink'a system he requested the service of Mariano who, as a master builder, W8$ to receive 15,000 soles (the dollar at that time was about 12,100 soles) for constructing the new putuku. In addition, Faustino and two younger workmen (Jesus Mamani Huanca and Jorge Caira Bautista) helped Mari~o during the entire building process which lasted one day. The tools used include the foot-plow (chaki-taklla in Runasimi, and wiri in Aymara), an adze (rawk'ana in Runasimi, and lijwana in Aymara), a sharp-edged shovel, a small pick, and rope. Using these tools, the men begin the building process by compacting, filling holes, and evening out the surface of the ground, an area that is already elevated. The locations of the comers for the rectangular plan are marked by placing a stake at each of the four comers connected by a string at about 90 degree angles, and the lengths of the sides (3.53 by 3.1 by 3.43 by 3.18 m) are measured by outspread arms. The first layer rests directly on the surface of the ground without foundation. The rectangular blocks of ch 'ampa are laid face down without mortar in this and each successive layer so that the surface of the next block on top rests directly on the underside of the one below. During the entire construction, blocks are laid down so that the vertical joints between blocks of one course occur over the centers of the blocks of the course just below,
10. Such a large putuku composed of some 39 layers of ch'ampa can be seen in Bela6nde Terry (1961).

ANDEAN PAST 5 (1998) known as isodomic construction (aparejo isodomo, Mesa 1978:13). Where necessary, this technique is achieved by further reducing the size of blocks and accommodating them to fit using the small pick and shovel. The base of the doorway or entrance begins after the second row of ch 'ampa. Doorways are placed to avoid the prevailing winds, and, in this case, the door was oriented to the northeast. The door is trapezoidal and measured 56 cm wide at'the bottom and 42 cm at the top, and was 1.22 m high. At the conclusion of raising the walls flanking the space for the doorway (Figure 14), two small wooden supports about 4 to 5 cm in diameter and made of the native qewiia tree, were set across the top of the doorway. The only large ch'ampa measuring 70 by 41 by 13 cm, was then placed as a lintel over them. Gradually each block and layer in this lower portion were placed slightly towards the interior so that the four -walls, of 15 layers when complete, have an inclination of about 8 degrees and were about 1.72 m in height. During the construction of this lower portion, several temporary long wooden boards were used to brace each wall at different locations on the interior, one end of each board on the floor aild the other about half way and higher up the wall. In preparation for the upper conical vault a small qewiia wood branch or comer brace about 4 cm in diameter and 40 cm long was permanently embedded in the last layer diagonally across each of the four comers. About 20 cm of the length of the horizontally positioned sticks was exposed on the inside at each comer. These comer braces, which presumably served to reinforce the corner joints, have also been reported on several stone-adobe Inca structures of Cusco (Moorehead 1978:73, 84-85, 91, figures 6, 9, 13, and 27). Finally, a small rectangular vent (10 by 11 cm) was left open within the ninth layer at about 55 cm to the right of the door jamb, as viewed from the exterior (Figure 14).11 Once the last layer of the lower portion has been concluded, the upper corbel vaulted
II. Although the vents are called ventanitas (small windows), they provide a very small field of vision and little light, due to the wall thickness. A more appropriate term might be air vents, and they function as additional smoke vents.

-364

structure begins. The first course of ch 'ampa rests almost entirely on the flat horizontal surface of the rectangular lower portion that measures about 3 by 2.72 m. This first layer of the vault follows a circular to slightly oval plan, leaving only four small triangles of the lower walls exposed on the exterior at the four comers. The blocks of ch 'ampa used are also placed face-down without mortar and require ' additional reshaping of each rectangular block into trapezoidal forms (Figure 11). Using the foot-plow but without the aid of his foot, the master builder strikes each rectangular block vertically to remove two comers from the same longer edge, achieving a roughly trapezoidal form. No measuring device is used. Subsequently, the master builder climbs up the structure and receives each block which has been tied to a rope (Figure 12). As the person above pulls up the rope, the workmen below lift the block in a synchronized manner to avoid rubbing or damaging it against the existing walls. The blocks, which are constantly evened out at the edges with the small pick and shovel to achieve the desired fit, are also placed with staggered joints. The conical vault is gradually achieved as each successive circular course is reduced in diameter by torbeling. The width and length of blocks in each subsequent layer are also reduced by removing additional portions at the four sides and rounding the faces of the blocks that will form the curved exterior and interior surfaces of the vault. This process gradually makes the wall thinner in the upper portions. As the corbel vault is being erected, three small rectangular vents are placed in the following locations: one 15 by 9 cm within the second course directly opposite the door, and two vents (10 by 11 cm) within the third and ninth layers, respectively, in a line directly above the door (Figure 14). Furthermore, in the nineteenth row, close to the apex, two smoke vents are made opposite one another. The last block which closes the structure is a circular ch 'ampa. The last and final step made by the master builder is to shave and even out the exterior of the entire corbeled portion, using a shovel

365(Figure 13). He then leaves and assigns the remaining tasks to the three workmen who will complete them in one additional day. At this point the entire structure is composed of 35 layers of ch 'ampa (15 layers on the lower rectangular portion, and 20 layers on the upper conical vault portion including the last circular ch 'ampa at the apex), and has a maximum interior height of3.80 m. When the main structure has been finished, the three workmen remove the long temporary wooden supports from inside and clean the debris accumulated there during construction. Subsequently, the interior floor as well as the trapezoidal walls are scraped and evened out in preparation for plastering with mud. The entire floor is then covered with a single layer of tightly fitted ch 'ampa also laid face down, and an additional layer is added on the right side (as one enters the room) occupying about one third of the interior space to form a platform for use as- a bed (p 'atjati). Furthermore, as we were informed and were able to observe in some other putuku, a niche (t'ojo) is carved out on the interior above the bed. One final architectural attribute, an enclosed entryway, is attached to the trapezoidal door on ~heexterior (Figures 17-18). It is also built with ch 'ampa and its lintel similarly rests on two qewna wood supports. The mud plastering is then applied to the floor, bed, and interior walls of the lower rectangular portion only. As the walls are being plastered, some small wooden pegs are inserted into the walls to use for hanging things. Additional Observations Comments and Informant's

Chavez: Corbel Vaulted Sod Structures documenting the settlements nearby, we saw a household cluster with both a putuku having a convex-walled vault formed by a total of 20 courses of ch 'ampa, and one next to it with a straight-walled vault formed by 22 rows (Figure 16). Likewise, depending on individual preference, the total number of vents (without counting the two smoke vents at the apex which are always present) in most cases is three or four, but can be as many as 10 as ob-, served in Mariano's own putuku. The last single circular ch' ampa at the apex may alternatively include two circular ch 'ampa (Figure 17). Furthermore, Mariano indicated that other builders do not include the permanent wooden "braces" at the four wall comers prior to building the conical vault. We were also informed that some people plaster the entire exterior with fine/sifted mud mixed with cut ch'illiwa ichu grass. Although we did not observe many cases of exterior plastering in Yanaoqo, informants indicate that this procedure helps maintain the structure and can be repeated every 5-15 years to make the structure last longer. Regarding the total number of layers or courses of ch 'ampa employed in the construction of putuku' and the proportions used in each of the two portions, we have a sample indicating the following distribution:12
Lower Totaratio upper: Dortion lower portion 32 1.1:1 15 32 12 1.7:1 12 33 1.8:1 33 13 1.5:1 35 1.3:1 15 36 1.6:1 14 15 39 1.6:1 15 41 1.7:1 12-15 32-41 1.1:1-1.8:1 1.5:1

This section summarizes our additional observations and comments made by Mariano Quispe Bautista and other informants in communities between Yanaoqo and Taraco, concerning variations in technology, design, location, duration, and ceremony associated with a new putuku. With respect to the shape of corbeled vaults, Mariano indicated that the conical vault may also be slightly convex-walled, depending on individual preference. While

1 2 3 4 5 6 7 8 Range Averae:e

17 20 21 20 20 22 24 26 17-26

21.25 I 13.88 I 35.13 I

In one case we observed a putuku having 35 layers in the upper portion and 16 in the lower, totaling 51 layers. The upper portion always
12. For ease of reference, we consider even the last single circular block of ch'ampa at the apex as a layer. This sample was not selected probabilistically.

ANDEAN PAST 5 (1998) has more courses than the lower, and has a broader range of variability in numbers of courses than the lower. A variation of the typical. putuku is the taklla putuku (Figure 29), that is built for larger families. Although it has the same inclined walls of the lower portion and same relatively small trapezoidal doorway (about 1.10 by 0.60 by 0.45 m in one case), it is more oblong in its rectangular plan. The upper corbeled portion forms a sub-rectangular or ovoidal shape approximating or imitating a gabled structure with four slopes. Furthermore, the portion at the apex and just under the last row of ch'ampa, requires the support of long straight beams (usually from the recently introduced eucalyptus tree) across the length of the "roof." In addition, beams are placed at several points along the width about halfway down the "roof' on the corbeled portion. This kind of structure is relatively rare in the region and is associated with well-to-do campesinos. A walled household cluster composed entirely of three taklla putuku was documented in the. community of Pataskachi west of Yanaoqo. These structures face an open patio and are surrounded or flanked by a corral. Within the household enclosure is a smaller putuku for chickens and dogs as well as a small enclosure to keep piles of harvested kinua. The largest taklla putuku measured 7 by 3.10 m in plan, had 40 cm wall thicknesses and was 3.50 m in total height. It was built of 18 rows of ch 'ampa in the lower rectangular portion (to a height of 2 m) and 17 layers in the corbeled portion (to a height of 1.60 m). In addition, this structure was large enough to have a partition wall that separated a bedroom, measuring 3.30 by 2.25 m, from a kitchen, each with its own doorway to the outside. The number of beams in this case was three along the length, and eight along the width. One of the other taklla putuku was used for storage and the other as a bedroom. This example shows that a well-to-do family has putuku built for special functions rather than having a single multiple function putuku as in Yanaoqo (see section below).

-366 Small-sized putuku are also built with ch 'ampa and plastered on the exterior. They are located within household clusters, and are specifically used to keep pigs, chickens, or sheep and cow dung for cooking fuel. Inquiring as to why putuku are not found outside this region, we were told that their ab-

sence was due to the lack of good ch'ampa,,


and that such ch 'ampa-producing fields are destroyed by farming. On the other hand, the appropriate location to build a putuku must meet the following criteria: it must be a warmer area (zona abrigada), on solid high ground to avoid inundation, and be surrounded by, or near, good soils and pasturage. Likewise, the best time to construct a putuku is within the months of September and October when the weather is mild. The addition of a new putuku is directly related to increased family size, and the formation of a new household cluster occurs when offspring marry and move elsewhere. Furthermore, some older informants indicated that before the influence of Adventists and Evangelists, people used to avoid "wrong" places inhabited by ancient spirits or gentiles. At the conclusion of building a putuku a ceremony or pago is performed. Although we have not witnessed these, many informants in the region indicate two kinds of associated pagos which can be carried out by the builder himself, or a yatiri (indigenous priest): one involves splashing sheep or llama blood onto walls, and the other is burying a small pot containing subsistence items in the middle of the putuku. These pagos are performed to aviod the collapse of the structure which could "eat" the people in it, and to ensure a long life for the putuku. The putuku, sometimes present with other kinds of structures (see Household Cluster in Use), are grouped together, forming household clusters as protection against the prevailing winds. The four most severe climatic conditions present in this region include rain, water, wind, and cold. According to informants, the corbel vaulted structure allows a more efficient response to these conditions in terms of building material and form. For the former,

367ch 'ampa blocks provide thick lower walls of about 50 cm that are slightly thinner at the upper portion, as well as high resistance to water filtration at the base. In form, the well-shaved exterior walls (sometimes plastered) and the approximately 8 degrees inclination of the lower rectangular portion protect the structure and allow fast rain water run-off. A single relatively small, trapezoidal door that is oriented towards the sun or against the prevailing winds, helps maintain interior heat. Likewise, the additional enclosed entryway framing the doorway, as well as the wooden or cowhide door, serve to protect the interior from rain, dust, and cold. When a household cluster is abandoned, as we observed in some cases in Yanaoqo, the doors are closed with ch 'ampa blocks (Figure 15). In old and deserted structures the first portions to collapse are the layers at the apex and the entryway (Figure 15). Asked how long a putuku would last, Mariario indicated that a putuku will last much longer than a structure built of adobe &11dcorrugated iron, and speculated that it could survive from 80 to 100 years. In this regard, Martinez (1962:76) reports that many of the still habitable ones had already been in. existence for perhaps three generations. Furthermore, La Barre (1948:95) reports the presence of many ch 'ampa-built chullpa (prehistoric burial towers) south of Lake Titicaca. Unfortunately he does not mention the shape, size, or content of these tombs that could help determine their relative age. We know these burial structures to have been built at least in Inca times. Conservatively, La Barre's ch 'ampa chullpas may be 400 years old, but may be later or even earlier.13 Several informants indicated that the origin of the putuku is very old, and referred to the Inca, Qolla, and Lupaca peoples as their ancestors who might have originated the putuku. Finally, we also learned that there is a basic division of labor by sex in the process of building putuku. Only men can build putuku
13. Welsch (1973:71) documents sod houses still in use in the north central states of the United States dating to the 1800s and early 1900s (see also Dick 1979:passim, and figures facing pages I 12 and I 16).

Chavez: Corbel Vaulted Sod Structures because "women cannot lift ch 'ampa." However, women can help to make and prepare ch 'ampa in the fields. Similar structures, but entirely circular in plan, have been reported from Bolivia, pertaining to the Chipaya ethnic group in the De.. partment of Oruro (e.g., Mesa and Gisbert 1966), and also from the district of Desaguadero in the province of Chucuito in Peru (see Comparisons and Distribution). We therefore asked our informant and builder in Yanaoqo whether a putuku having a circular plan could be built. His response was simple and definite, "it cannot be done." A Household Cluster in Use This household cluster belongs to Mariano - Quispe Bautista who is married to Paula Caira Larico, and they have a son (then 22) and a daughter (25) who now reside in Arequipa. Their main sUDsistenceactivities include the cultivation of potatoes, kiwna, and barley, as well as the herding of a small flock of sheep. In addition, Mariano fishes along the nearby lake shores, using a kind of fishnet he manufactures called a karuna, and through the mink'a system he cuts ch'ampa and is a well-knownputuku builder in the community. Mariano's household is situated just south of the Taraco-Huancane road, adjacent to two other household clusters on the north and east. A wall enclosure delimits a rectangular area oriented north-south, with its entrance in the southeast comer. The north and south walls of the enclosure are made of eight layers of adobe combined with one layer of ch 'ampa at the top to a maximum height of 1.40 m (Figures 19-20). However, on the east (adjacent to his brother's property) and south, the walls are lower and of uneven height, and are entirely made of ch 'ampa. In the southwest comer there is a gabled room made out of adobe and mortar that is roofed with totora brought from the nearby shores (Figure 18). Two of its sides form part of the enclosure wall. This structure has a carpentry-made padlocked door and a single sheet of glass set directly in the adobe window

ANDEAN PAST 5 (1998) opening. It is used to store agricultural products and as a temporary bedroom. To the north there is an old putuku that had been built by Mariano's father (Figure 18). Mariano was in the process of taking it apart in order to rebuild it, reusing the old ch 'ampa blocks. Near the northwest comer there are two small, mud plastered structures made of ch 'ampa used as chicken coops, and in the northwest comer there is a rectangular corral (kancha) for sheep. The two walls.that abut against the enclosure walls to form the corral are made using two layers of irregular stones at the bottom and 6-7 layers of ch 'ampa, to a height of about 1.20 m (Figures 19-20). Cow dung for use as cooking fuel was being dried on top of the corral wall. In the northeast comer, an introduced cypress tree grows inside a ch 'ampa enclosure which protects it from frost. Finally, on the east wall, near the entrance, there is a putuku with its back wall forming part of the enclosure'wall (Figure I.7). In the middle of its north wall there is a small square exterior extension about 70 by 70 by 30 cm. The putuku has a-simple door, oriented to the west, that is made out of a piece of metal nailed onto a wooden frame. Nine small vents are distributed around the conical vault,14in addition to two smoke vents at the apex and one small vent to the left of the doorway. The use and distribution of interior living space for sleeping, cooking, and eating in the putuku here is the same as in others we observed. In fact the arrangement was remarkably uniform. In the comer left of the door, formed by the north and west walls, there is a fired clay stove (q 'oncha) with three openings. The fuel used is cattle and sheep dung (toqra wanu) and occasionally eucalyptus firewood brought from the nearby places called Kirichaya, Qaqocha and Qaparaya. The small extension built in the middle of the north wall is used to store the pottery vessels for the kitchen. The comer formed by the north and
14. We observed here that the exteriors of most of the vents located at the conical vault were blackened from smoke. Consequently, these air vents also function as additional smoke vents.

-368 east walls is designated for the rocker mill brought from a hill called Timoray. Inserted into the lower rectangular portion of the east wall are wooden pegs, and a niche (t' ojo) is carved into the middle of the south wall. Finally, the platform used as a bed is located along the entire south wall, measuring about 1 m wide and 18 cm above the ground. We were informed that four people could sleep on ' this bed. The sheep wool blankets used in the bed had colorful bands on black or brown backgrounds, and were all woven by Mariano. Additional kitchen implements found in the space between the stove and grinding stones include pottery vessels made near Putina such as the q 'ocho (bowl), tachu, manka or hayku (cooking pots) and /lata (shallow bowl). In addition, there was a glazed pottery bowl from Pucara called p 'uku, a plastic bowl, and spoon called wishlla made from the qewiia wood in Putina, silverware, aluminum vessels, and tin buckets. Outside and near the putuku we observed one large square stone vessel as well as other round stone vessels that are used as wash basins and as containers to feed pigs (Figure 17). These stone vessels are made and distributed by the inhabitants of the Island of Amantanf in Lake Titicaca, who, during the month of August, arrive in rafts at the bridge over the Ramis River and barter these vessels for food. Mariano exchanged these vessels for barley and potatoes. In recent years, bartering (rather than buying) has been emphasized in many regions of the south highlands. Taraco and its Urban Characteristics The town of Taraco is the capital of the district of Taraco, in the province of Huancane, department of Puno (FigUre 1). The urban population in the province is 12,209 and the rural is 96,904; while the population in the district of Taraco is 14,970 (Peru: Censos Nacionales 1981). The topography of the district is relatively flat (between about 3800-3850 m elevation) with some isolated hills of low elevation, and one more prominent chain (with Cerro Imarucos at 4085 m) to the southwest of Taraco itself. The terrain is slightly inclined

369-

Chavez: Corbel Vaulted Sod Structures The town of Taraco, having a population of 535 inhabitants (according to the census of 1981), is composed of a predominantly mestizo population speaking Spanish, Runasimi, and some Aymara. The people possess a higher socio-economic status in relation to those in rural areas (e.g., Martinez 1962: 95104 and chart on page 26). The clustered set-, tlement pattern here (Figures 21-22) stands out in contrast to the dispersed rural settlements such as Yanaoqo. According to the classification of urban settlements proposed by John H. Rowe (1963:3-4), Taraco is a small synchoritic city. There are streets forming blocks and a main square (Plaza de Armas) in the center ornamented with trees and prehispanic stone sculptures. There are public buildings such as the church, schools, municipal building, police station, medical post, technical school (Centro de Formaci6n Profesional de Taraco), post office, stores, and recently, a public telephone service. Taraco has electricity as well as water (from a water tower) and sewage, and a weekly market on Thursdays (Ghersi and Arquinio 1966: passim). Some Taraco residents own land nearby, but have others work it. Building materials used in Taraco include plastered adobe and mortar, tapia, brick, and cement. Ch 'ampa blocks are absent. Roofs are mainly corrugated iron, but some are thatched with ichu grass, ~d tile occurs only on a very few houses and on the church of San Taraco. The fronts of the houses in Taraco are at the edge of the street and form part of the fapia walls that enclose the entire lot. Some of these tapia-made enclosures, especially those towards the river, are large enough to be used for cultivation. The houses include one- and two-story gabled structures (a dos aguas), and some with the roof inclined in only one direction (a una agua) towards the inside of the house or patio, giving the false impression of having a flat roof (or azofea) when seen from the street (Figure 22). Behind the houses and within the enclosure walls are some strawthatched, one-room rectangular structures, and sometimes an ornamental tree is grown such . as the small native qo/li and the introduced cypress. Finally, inconspicuously situated in the enclosures well behind some houses, espe-

with a gradient which usually ranges from 0.5% to 1%, and rarely from 2% to 5% (Ghersi and Arquinio 1966:26). There are basically two seasons: a cold and frosty period from May to October when there are strong winds (especially from July to September) and low temperatures that may fall to 0 degrees Celsius; and the period of abundant rain from November to April. At any given time during

the year one can feel a high contrast in tem- .


perature between the sun and shade. The annual precipitation ranges from 425 mm to 1121 mm, and the annual average temperatures vary as follows:
Maximum: 14C in January, and 21.1C in October Minimum: 3.SoC in January, and 4.7oC in June Median: 4C in June, and SOCin November/April (Ghersi and Arquinio 1966:4).

One of the major problems in this region is flooding of lakes and rivers that occurs during some years, creating disastrous consequences for local populations as huge agricultural and residential areas become inundated (e.g., Cuentas Gamarra 1971:60). For example, the Rami:; River, which flows from west to east and is at its river bed only 1.76 m above the level of Lake Titicaca, increases in depth from 2 m to 4 or 5 m in the rainy season; in some years it even overflows (e.g., Gilson 1939:9). Furthermore, this area is only approximately 18 m above the level of Lake Titicaca, and the soils are characterized by slow permeability and are resistant to water erosion (Martinez 1962:4-5, Ghersi and Arquinio 1966: 26-27). In addition to the Ramis River, there are also a number of small and temporary rivers formed in the rainy season called mishi, that maintain pastoral lands and facilitate cultivation. Cattle, sheep, pigs, and small quantities of horses and donkeys are raised. The absence of camelids in this region is noticeable; for example, Ghersi and Arquinio (1966:34) report only one or two kept as ornamental animals. The most important cultivated plants include potato, kiwiia, and barley; other plants grown in smaller qualities include kaiiiwa, peas, broad beans, ulluku, oqa, tarwi, and a small variety of com called maiz conflte grown in sheltered areas.

ANDEAN PAST 5 (1998) cially those near the river, are corbel vaulted structures of a different kind from those found in rural areas (Figure 22). Unlike the ones found in the nearby rural communities such as Yanaoqo, this kind of putuku is built entirely of adobe and mortar or combines both with tapia, mayor may not have vents, and possesses a rectangular door jamb. Another major difference is the design of the upper corbel vaulted portion; four surfaces incline towards a straight summit, the opposing broad sides of which have a trapezoidal silhouette while the narrow sides form isosceles triangles, similar to the ch 'ampa-built taklla putuku (Figure 26). Furthermore, based on direct observations and information provided by our informant, these one-room structures are only used to store harvested products, keep pigs and sheep, and rarely for cooking. Construction Technique of Corbel Vaulted Structures in an Urban Setting Our main informant here was Francisco Palomino Olivera, a long time resident of Ta~ raco who, around 1974, built a large putuku in addition to the small one already present on his property near the bank of the Ramis River. This site is called Aldonates Patamayu, an area of higher ground due to the presence of intensive. prehispanic occupation there, and was the location we excavated in 1973. Two of these structures face each other within a rectangular tapia wall enclosure about 14 by 9 m and 1 to 1.50 m in height (Figure 23). The large putuku occupies the middle of one wall, and the small one is ilear a comer and the entrance. In turn, the rectangular wall enclosure forms part of and is located in the middle of the central wall of a larger, three-walled tapia enclosure which limits a block size area. This large enclosure has walls bordered by three streets and is open toward the river bank on the north. The first step in the construction is the preparation of a flat rectangular plan about 3.67 by 2.90 m. The four superimposed layers of tapia are built, always avoiding the junctures of previous layers (isodomic construc-

-370 tion).ls At this point, the basal portion is about 2 m high. Each of the four walls has an inclination of about 5 degrees and forms a trapezoidal surface (Figure 24). The rectangular door jamb which is oriented towards the river to avoid the winds, measures 1.25 m in height and 60 cm in width; it is about 20 cm above the floor, with two wooden beams to support the tapia lintel. The second, upper, portion rests directly on the last tapia layer, and is built entirely of 19 courses of adobe and mortar. The first two layers of adobe follow the same inclination and rectangular shape of the previous tapia layer of the lower portion, after which the subsequent horizontal courses of adobe are gradually corbelled. The comers are rounded by placing adobes transversely across them, and adobes are placed polygonally in each course to form a more ovoid plan (Figure 25). Furthermore, the last three out of the 19 courses of adobe have a different arrangement: courses 17 and 18, havip.g eight and six adobes respectively, no longer form polygons; instead they form rectangles. The last course at the summit consists of three parallel adobes; their widths close the length of the rectangular opening. The stepped exterior is then smoothed by scraping with a small pick and shovel, but the interior of the structure remains stepped, with the adobes left protruding onethird to one-fourth of their width. As the main structure is concluded to a maximum height of 4.65 m, an entryway is built at the door with tapia, and the entire exterior is plastered with mud for a thickness of 5 em. The interior of the structure is also mud-plastered, but only on the lower tapia wall portion, and there are no vents of any kind. The only additional element on the inside is a wooden pole about 6 em in diameter set into and spanning the north and south walls at a height of about 2 m, to be used for hanging things. Finally, a layer of mud is also placed onto the interior floor. On the exterior, at the bases of the front and back, soil is shov15. The construction of this putuku was not observed by us. Consequently, this infonnation is based only on the memory of the builder at the time we interviewed him.

371eled to a height of approximately 40 cm, creating an inclined plane. It should be noted that, unlike other such structures we have seen in Taraco, this one has a more ovoidal or subrectangular vault in which the four comers and summit were purposefully rounded, but still give the impression of having a four sided "roof' when seen from a distance. The building of this structure required the labor of four people for five days (three days for the tapia portion and two days for the rest, not including the time used in making adobes). This putuku was built to store products Francisco cultivates, and when we visited in July of 1985 all of his broad bean harvest was housed inside. Before this structure was built, the small putuku had also been used for the same purpose. Abandoned Corbel Vaulted Structures On the east, near Francisco Palomino's property, there is a tapia-walled area with two abandoned putuku now used as public toilets (Figure 26). We measured and documented these structures to illustrate additional variations in. shape, .size, building material, and function, as well as the initial process of decay following their abandonment. One putuku is very large (maximum height 4.30 m) with the door facing south, and the other, adjacent one is medium-sized (maximum remaining height 3.60 m) with a door facing west. The lower portion of the large putuku is built of three layers of tapia to a height just below the door lintel, and the upper portion consists of 18 layers of adobe with mortar (Figure 28). The rectangular plan measures 5.30 by 3.50 m. The four walls are 1.40 m in height and are inclined approximately 5 degrees. The upper portion of the structure rests directly on top of the tapia walls which form a rectangle measuring 5 by 3.30 m. The corbel vault has four well-defined sides; two narrow ones forming isosceles triangles, and two wide sides forming trapezoids; constituting what may be called a four-sided corbeled vault with a 2.30 m-Iong ridge at the top. Unlike the polygonal, ovoid disposition of layers on the putuku described above, the entire corbel vault

Chavez: Corbel Vaulted Sod Structures here is achieved by layers forming rectangles in which the four comers are maintained, similar to what Mesa (1978:21) calls b6veda por tajadas (like that in Figure 27). The rectangular door jamb measures 1.26. m high and 80 cm wide, and is about 25 cm above the ground. A wooden beam supports the lintel which is formed by the first layer of adobe. There was no evidence for a covered entryway, although this might have collapsed; and the area in front of the door was elevated almost up to the door sill. Furthermore, there were indications that at least the lower tapia walls had once been plastered on the exterior, and the upper adobe portion was now exposed and beginning to erode. -The interior was also mud-plastered up to the top of the tapia walls. At about 1 m above the plaster, and to a height of about the upstretched hand, three poles spaced 90 cm apart were inserted into and across the north and south sides. Three small rectangular vents (13 by 13 cm) are present only in the upper adobe portion. One is in the fourth row of the south wall and near the southwest comer; and the other two are in the west wall, one at about the center of the third row, ~d one in the eighth row near the northwest comer. In addition, the interior of the north and west tapia walls has two niches which were directly carved out when the walls were dried. The niche on the north wall (47 by 25 by 10 cm deep) is 1.19 m from the floor and 1.17 m from the west wall. It is triangular in cross section. The one on the west wall (22 by 23 by 15 cm deep) is 1.21 m from the floor and 80 cm from the northwest comer. It is rougWy rectangular in cross section. Lastly, most of the interior walls and vault portions were blackened from smoke, indicating that this structure had been used at least for cooking, and that the small vents might have functioned as smoke vents. The second putuku has basically the same shape as the first (Figure 26). Other characteristics, however, such as its smaller size, exclusive use of adobe and mortar, absence of vents and niches, and lack of smoke blackening on the interior, indicate that this structure had a different function.

ANDEAN PAST 5 (1998)


.

-372 name given to this area is Umabamba, and it is near a hill called Ninantaya. In addition, we have also observed in the Cusco region, between Sicuani and Raqch'i, isolated corbel vaulted structures not forming household clusters, located near agricultural. fields, and used as temporary shelters, or for storage. Likewise, similar structures were noted midway between Santiago de Huata and Escoma

The rectangular plan measures 2.85 by

2.26 m, and the inclined walls of the lower portion include 10 layers of adobe to a height of 1.50 m up to and including the door lintel. The upper corbeled portion also has four welldefined sides and comers (Figure 27). It was built of 17 layers of adobe, with one additional last layer on top now missing; judging by the space, there had also probably been three adobes in this uppermost layer. The rectangular door, measuring 92 by 60 cm with a sill 20 cm above the floor, faces west. It has a wooden beam supporting the lintel, as well as impressions left by a door frame. Mud plastering was present only on the interior up to 1.74 m above the floor. Comparisons and Distribution And Settlements of Putuku

in Bolivia.

A search of the current literature dealing with the Peruvian side of the Lake Titicaca Basin indicates that this kind of structure, as well as the use of ch 'ampa,. has been poorly documented and to some extent ignored. Most of these references include very general descriptions and lack illustrations (e.g., Romero 1928:354; Kidder 1943:16, plate I, figure 4; Martinez 1962:76-77; Ghersi and Arquinio .1966:50,60; Sanchez Huanca 1970:24, 85; Cuentas Gamarra 1971:122-123; and Gutierrez et al. 1978:147, 1'49-151). Although these references do not permit close comparisons to be made with the putuku described here, nevertheless, they help confirm that at least for the last 60 years putuku distribution has been from Taraco (communities of Qollana, Patascachi, Ramis, and Requena) to near Huancane. Furthermore, Romero (1928:354) indicates another concentration of putuku to the northeast of Lake Titicaca at the PeruBolivia border (on the Peruvian side).16 The
16. Emilio Romero refers to these structures as "cuchi. putucos" (putuku for pigs) which are built of ch'ampa and adobe, resembling truncated cones having a diameter no larger than 2 m, and similar to the ones built by the Ostiakes of Siberia. It is not clear whether he is referring to houses or to the smaller putuku which are also built for keeping pigs. Such pig houses (known as k'uc'iputu) were reported by La Barre (1948:96) in Bolivia, near Lucunnata. He describes this small structure as a "cobblestone-and-mud domed house about four feet high".

Another area where the putuku structure has been reported is the southern end of the Lake Titicaca Basin to the region of Lake Poop6. These corbeled buildings, which are usually related to the Aymara, Uru, and Chipaya peoples, are also associated with gabled rectangular structures, as well as with circular ones having thatched domes. Additional characteristics of technique, form, and basic building materials stand out in contrast to the constructions found in the northern end of the basin. In comparing the shape of structures present at both the northern and southern ends of the basin, it is significant to note the absence of circular buildings in the former and preponderance of them in the latter, whether putuku or thatched structures. The following discussion of these circular structures found in Bolivia will reveal the similarities and differences they possess in comparison with the putuku in the northern end of the basin. Urban houses of the Chipaya in the Department of Oruro, Bolivia,17have a circular plan about 5 m in diameter, and the walls are made of sod blocks that corbel slightly toward the interior. The roof is a dome formed by firmly tied straw bundles (Posnansky 1937:
17. Mesa and Gisbert (1966:492-493) and Gisbert (1988: 124-125) refer to these one-room rural and urban structures as houses. However, no infonnation is given as to whether each structure is an individual house, or whether two or more of them form a household. This infonnation would be useful for comparison with the household clusters of Yanaoqo. The classification of houses into rural and urban is not clear either, and the building material referred to as chunks of mud directly cut from the ground, is most likely also
ch'ampa. . .

373figure 57; La Barre 1946: plate 114) made more solid by a mud paste, and is thatched with grass secured by a net rope over the dome. The door is 1.20 m high, the height of the walls 2.20 m, and the maximum height of the entire structure about 4 m (Gasparini and Margolies 1980:141-142, figures 125-126; Mesa and Gisbert 1966: 492-493). The . rural houses of the Chipaya, also known as putuku (see note 16), are one-room structures similar to the ones described above, but the corbeling of the walls extends upward to form a vault (Posnansky 1937: figure 104; Mesa and Gisbert 1966:493; Gasparini and Margolies 1980: figures 133-134). Both kinds of circular structures, the corbel vaulted and the thatch domed, may be found . together (Gasparini and Margolies 1980: figure 129; Posnansky 1937: figure 51). Furthermore, La Barre (1948:94-95) indicates that among the Urn of the Desaguadero River area, Ancoaqui, and in the Aymara village of Cafiaviri, circular' and rectangular gabled structures occur in equal numbers. From Wankarani, Ponce Sangines (1970: figure 9) shows a circular and a rectangular gabled structure~ both with thatched roofs, side by side. From the same period of time, Gisbert (1980: figure 16) reproduces a watercolor illustration of Poop6, in which four putuku having circular plans together with gabled structures are in front of a colonial church. Furthermore, Tschopik (1946:529, plate 110 top, from the desert of Carangas, and center, from the village of Punata) illustrates corbel vaulted sod structures of rectangular plan, that form clusters, each enclosed by a rectangular wall, similar to the putuku described in Yanaoqo. Regarding the orientation of doors, La Barre (1948:93) notes that 75% of the Aymara houses in rural areas have the doors facing east, and when these are located in villages the doors face toward the plaza. Likewise, La Barre (1946:578) indicates that most Uru houses have the entrance always facing east; Posnansky (1937: 115-116) notes on the Island of panza (Lake Poop6) an eastward door orientation for both prehispanic and modem

Chavez: Corbel Vaulted Sod Structures

houses; and Gisbert (1980:27, figures 12-15) describes a similar orientation for the prehistoric burial towers (chullpa), the church, and modem structures in the town of Anco-Cala, Department of Oruro. Although most authors do not explicitly indicate the orientation or shape of the door in different areas, it may be' noted that their photographs confirm at least a consistent orientation towards one direction. The door jambs appear to be rectangular, with the exception of what appear to be trapezoidal doors in circular thatch-roofed Chipaya houses (Gasparini and Margolies 1980: figure 126). The orientation and shape of doors may be seen in the rows of Chipaya one-room structures shown by Posnansky (1937: figures 5152; 1958: plate CVII.C.) and by Gasparini and Margolies (1980: figures 125, 129, 133-134). The basic building materials employed in the construction of structures having a circular plan and straw or totora.thatched roofs include sod blocks (Gasparini and Margolies 1980:141-142), or "clay-bound cut turf' (La Barre 1948:93) in combination with stone at the lower layers (Posnansky 1937: figure 36; Gisbert 1980: figure 15). Sod blocks are also employed in building corral walls (La Barre 1948:95-96). The corbel vaulted putuku is mainly built of sod blocks'. However, Forbes (1870:254-255 cited in La Barre 1948:95) observed round and oval stone houses with corbeled domes on the slopes of Illampu in the mid-nineteenth century. Another example, probably the remains of a corbel vaulted room, was built of very thin adobe (Ponce Sangines 1970: figure 10). In conclusion, due to the nature of the information given in dispersed bibliographical references, systematic comparisons between structures at both ends of the Lake Titicaca Basin remain relatively limited. It may be said, however, that the similarities lie chiefly in three aspects: (1) The use of sod blocks of ch 'ampa without mortar; (2) the technique of corbeling to form a conical vault; (3) similar door dimensions and total height of the structure; and (4) the association of putuku with gabled one-room structures. It may be pointed out that the thatched domes on frames of straw and mud represent an adaptation to the lack of

ANDEAN PAST 5 (1998) trees in this region (Mesa and Gisbert 1966:493),just as does the corbeling. On the other hand, outstanding differences include: (1) rows or isolated groups of putuku, quite different from the individual household clusters enclosed within rectangular walls in Yanaoqo; and (2) the absence of a putuku having a rectangular plan in the south, and the absence of a circular plan in the north. This distribution is a complementary one. An exception to these two characteristics, however, can be found in Bolivia as shown by Tschopik (1946: plate 110). Photographs from the Desert of Carangas provide an exception to the former, while photos of the village of Punata illustrate an exception. to the latter, in which there are dispersed household clusters and putuku with rectangular plans like those in Yanaoqo. Regarding the presence of rectangular as opposed to circular putuku, two alternatives may be proposed to explain 'this complem~ntary distribution. It may indicate or reflect two independent building traditions adapted to similar environmenial .conditions, or may indicate that people in the northern end of the basin had given up circular structures in favor of rectangular ones, as was the case in 1582 for circular houses in Jauja, central highland Peru (Vega 1965 [1582], Volume 1:171). A related case was reported in 1584 by Pedro Mercado Penaloza dealing with the province of Pacajes in Bolivia, where many caciques gave up their circular plan houses to adopt ones with rectangular plans as a result of the Inca conquest (cited in Gisbert 1988:149). Dispersed vs. Nucleated Settlements: Description and Interpretation This section provides a broader context for the dispersed nature of settlements as in the case of Yanaoqo, and nucleated ones such as Taraco. Here, the issue of settlements is directly or indirectly related to the patterns developed or instituted during the last four centuries, as well as to those preceding the Spanish domination. Consequently, this section deals with the modem distribution and characteristics of settlements, while the next

-374 section provides the historic and late prehispanic background. 18 With respect to the kind of settlement pattern found in the southern end of the Lake Titicaca Basin, information is meager. Loza Balsa (1971:73), however, very briefly notes
the dispersed nature of all the communities in

the Aymara territory. Tschopik (1946:528) also states: "True town life seems not to have been typical of the majority of the Aymara. Although no reliable figures are available, the bulk of the population appears to have lived in family groups scattered in the ayllus of each town. .. In some regions, the Aymara live in towns, but occupy houses near their fields during the agricultural season." These observations, then, coincide to some extent with my characterization of Yanaoqo and surrounding rural communities (Figure 1). Other writers, too, have reported a similar pattern in many rural areas of the north and western portions of the basin. For example, Ortiz Vergara (1965:19) states:
"And there are places in which the indigenous housing is notable for its dispersion, as is found on the Pampa de Have, Province of Chucuito, Department ofPuno, which is called "Little London", because of the great number of widely-dispersed houses, which at first glance looks like a forest 'of houses of immense extent; . . . as is the case in other parts of the department of Puno, such as the Pampa de Taraco, in the Province ofHuancane." 19

Similarly, Galdo Pagaza (1967:63), describing the area of the District of Capachica, notes that "In the 8 social divisions nuclear family
18. The reader should also be referred to a work by Teresa Gisbert (1988). Using a diachronic approach, she attempts to document and understand the problem of housing and settlement patterns in Bolivia.
.

19. "Y, hay lugares, en los cuales, la vivienda indigena tiene como distintivo el de su dispersi6n, tal como sucede en la Pampa de Have, provincia de Chucuito, departamento de Puno, a la cual se Ie llama 'Londres Chico', por el gran nfunero de viviendas dispersas, que en un primer golpe de vista, parece como un bosque de casas de inmensa extensi6n; . . . como en otros lugares del departamento de Puno, como la Pampa de Taraco, en la provincia de Huancane . . ."

375houses are dispersed and no villages are formed" ("En las 8 parcialidades las viviendas de las familias nucleares se encuentran disper.

Chavez: Corbel Vaulted Sod Structures Prehistoric Background), and our direct observations in. the south highlands, including the Lake Titicaca Basin, a hypothesis may be proposed here, indicating that there are environmental factors (such as climate, topography, and resources) in addition to social and political factors, that may generate or influence the' mass population to develop nucleated and/or dispersed settlements patterns as follows: (1) A dispersed settlement pattern composed of very scattered households is adapted to regions in the plateau of the Lake Titicaca Basin, where a topography of gentle gradient and relative environmental homogeneity is present over a large area. An environment like this would be more or less "evenly and extensively distributed," such that there are no "special" places that would encourage or require concentration or nucleation of settlements. Furthermore, maximization in the exploitation of resources .and/or use of land is better achieved by dispersing settlements and scattering individual household clusters. (2) A nucleated settlement pattern in which households are clustered forming densely populated towns and villages is adapted to regions like those in the Viicanota Valley in Cusco and the Qolqa Valley in Arequipa. Environments like these are.characterized by an irregular and steep topography composed of high slopes, alluvial fans, eroded terrain, tributaries leading to the main valleys, and canyons. Here the settlements are located mainly on rocky, infertile land, or half-way up a mountain slope, so that all possible land on the valley floor is preserved for cultivation and/or irrigation, while pasture lands are also available around the summit. The intensive use of cultivated lands includes terraces which can extend far up the alluvial fans and higher slopes. Johnson and Platt (1930:27, 31, figures 23-28) provide an example of this kind of topography and settlement located in the upper end of the main valley of the Qolqa River (Figure 32), and state that: "But for the alluvial fill of the glacial period the whole region would be pasture land only. It is a striking feature of the better-favored valleys that they are so completely occupied with fields that a larger population cannot find support except

sas, en ninguna de ellas forman poblado").


Finally, Emilio Romero (1928:174) writes in his monograph on Puno:
"The hacienda Indians live well apart in huts, separated one from another by thousands of meters. Those of the ayllus live in sheltered spots and in loose, rather than dense, hamlets, not forming villages or towns, so that they do not have urban life. The huts are nearby, separated by large fields. There is another sort of indigenous town that appears to be the remains of colonial reductions, that consists of collections of houses without urban life (known to the Indians as 'estancias,).,,2o
. The issue here, then, is whether a rural settlement is clustered, in which case each household unit is built side by side within a given area, or is dispersed. When a rural settlement in a plateau is dispersed, households are widely separated from each other, as in.the case of Yanaoqo, where its "boundaries" blend with the surrounding communities, making visual determination of the community difficult (Figure 1). However, recent observations (1992-93), particularly in the Copacabana Peninsula of Bolivia, indicate that each of the present-day dispersed Aymara communities is readily determined as they are bounded or limited by topographic features such as steep hills. We were also informed that the decision as to where to build a house is often made following a consultation with the local yatiri who determines or imposes a dispersed pattern (Eduardo Pareja Sifianis, personal communication).

Based on evaluation of historical references (see next section on Historic and Late
20. "Los indios de hacienda viven dispersos en chozas distantes, separadas por miles de metros unas de otras. Los de ayllu viven en parajes abrigados y en caserios agrupados, no juntos, no formando aldeas 0 pueblos, pues no existe entre ellos la vida urbana. Las chozas estan proximas, separadas por grandes chacras. Hay otra c1ase de pueblos indigenas que parecen restos de aquellas reducciones coloniales: agrupamientos de casitas sin vida urbana (llamados entre los indios 'estancias')."

ANDEAN PAST 5 (1998) by diminishing the per capita food supply or by part-time employment elsewhere." Furthermore, it may also be concluded that the putuku forming household clusters occur in the context of the former settlement pattern discussed above, and most especially appear to be associated with areas of poor soil drainage and/or near water edges, especially fluctuating ones. These ch 'ampa structures constitute the best inexpensive weather and water resistant alternative using readily available local materials, and the corbeling provides a solution to the lack of suitable timber for beams in the region. Ch 'ampa require a flat terrain and undisturbed ground where agricultural activities might otherwise conflict with their appropriate acquisition. Conversely, the putuku forming household clusters are absent in the lower valleys where conditions make them less suitable; better drained sloping terrain and large agricultural areas occur there, and trees are avai,lable for roofing gabled structures. One might predict, however, that where there is flat terrain, poor drainage, and little. conflict with agriculture, sod structures might occur. Historic and Late Prehistoric Background: Implications for the Archaeology of the Region Corbel Vaulted Structures and Ch 'ampa The earliest documentation of a putuku structure was made by Ephraim George Squier on his journey to the north end of Lake Titacaca, probably around Taraco. At the same time, he pointed out similarities in design with prehistoric burial towers known as chul/pa, and observed the characteristic remains left after their subsequent abandonment and collapse. Writing more than 100 years ago, Squier (1877:391-392) stated:
"The inhabitants here are all shepherds; and as what there is of solid ground is covered with a thin but tough turf, this is used exclusively in constructing their dwellings and the corrals, or pens for their flocks. Quaint and curious structures they are, looking like tall quadrilateral haystacks. In some of them, attempts had been made at something like architectural adornment; and these, as well as the chulpas,

-376
have a kind of cornice at the point where the roof begins to converge from the vertical walls or a tradition of style descending from the ancient builders of the tombs . .. A few had been deserted and had fallen down, forming mounds of more or less regularity and elevation, in which digging would certainly expose what we generally find in mounds of earth all over the world bones, fragments of pottery, some battered implements not worth removal, and traces of tire."

- a feature

suggested perhaps by the chulpas,

However, it should be noted that although he describes the putuku as a quadrilateral structure, the illustration he includes is one having a circular plan, and enclosed within a circular wall. Many chroniclers such as Cieza de Leon (1962 [1553]: Chapter 63), Bernabe Cobo, and Vasquez de Espinosa (1948 [1617]: Section 69, page 26; Section 1609, pp. 558-559), have described these burial towers, but none of them (including Be!1onio's Aymara vocabulary and Gon~ales Holguin's Quechua dictionary) use the word chul/pa in reference to these structures (see also reference to pucul/o in note 1). For our purposes, however, Cobo (1890-93 [1653]:236, vol. 4) describes the inner core of these burial towers as a corbeled vault, and referring to those present in Bolivia, he states:
"Inside there are spaces a little more than six feet deep, like a vault, which are closed with some wide, thin stones."21

On the other hand, Squier's surveys and descriptions of chullpa in this region are certainly the most extensive undertaken before the turn of the present century. Six years before his well-known work appeared in 1877, he wrote an article in which he seriated the chullpa and other related structures based on the form and technique employed (Squier 1871), constituting the first attempt at an evolutionary seriation of burial structures in Peruvian archaeology (S. Chavez 1979:319321). Of importance here is his observation and documentation of an interior corbeled
21. "Por dentro estan huecas poco mas de un estado, Ii manera de b6veda, la cual cierran unas piedras anchas y

delgadas."

377vault in many rectangular and circular chullpa from the Puno region (Figure 30). Subsequent studies have confirmed that all chullpa retain an interior corbeled vault (M. Tschopik 1946:12-16; Gasparini and Margolies 1980: 154). In addition to chullpa, there is also an Inca "palace" with corbel vaulted roofs first documented by Squier (1877:343-346, and figure facing p. 343) from the Island of Titicaca (the Island of the Sun). This "palace" known as Pilco Kayma is a two-story building, where a ground floor formed by several rectangular rooms, each with very large double and triple jamb niches and corbel vaulted slab roofs, supports a second floor with additional rectangular rooms. Gasparini and Margolies (1980:262) observe that Tiahuanaco and. Inca architectural decorative elements are integrated at Pilco Kayma. Most of the references and comparisons made between the prehispanic corbel vaulted structures in the Maya-mesoamerican and Andean regions, as well as comparisons with the putuku of the southern end of the Lake Titicaca Basin may be found in: Smith (1940:202-221), Trimborn (1973:185-190), and Gasparini and Margolies (1980: 142-159). No detailed literature existed for the northern basin, so that comparisons can now be extended to include the putuku described here for the north portion of the basin, which combines a rectangular wall with a circular corbeled vault.22
22. Gasparini and Margolies (1980: 142) conclude that while "the Maya corbeled vault almost always covers spaces arising rom rectangular plans; the Andean corbeled vault, on the other hand, covers an interior space based on a circular plan." Although this statement is certainly true in light of the evidence provided among the prehistoric chu//pa and the modem putuku of the southern end of the basin, our northern ethnographic examples of corbeled vaults adds the rectangular form and extends the possibility of its presence in prehispanic times. The "palace" of Pilco Kayma shows corbeling on rectangular walls in Inca times. Likewise, their assertion that ". . . the corbeled vaults of the Andean cultures (figure 129) [referring to Chipaya houses in Bolivia] are rough, without the least attempt to smooth the finish" (Gasparini and Margolies 1980:144), certainly does not describe the putuku of the north. Here, this description does not apply unless they

Chavez: Corbel Vaulted Sod Structures As we have seen, corbel vaulted structures were present in the burial towers of the region dating back to perhaps the Late Intermediate Period. In this respect, Hermann Trimborn (1973:189) proposes that the Aymara of the altiplano may have discovered the corbel vault and introduced it to the coast. Similar structures are also present in the de-, partments of Junin, Pasco, and HuAnuco and might represent an influence from Tiahuanaco-Huari times (Gasparini and Margolies 1980:144-145). Although references to freestanding (i.e. apart from chullpa interiors) corbel vaulted structures are not mentioned by chroniclers, such structures were probably present from at least the Late Intermediate Period in the Lake Titicaca Basin, and may have developed as a result of an architectural adaptation to an environment where appropriate trees are absent, while alternative and readily available building material, like ch 'ampa and stone, are present. A related, but much simpler form of corbel vaulted structure is commonly used today in the highlands in the construction of temporary baking ovens know as huathia, which may also represent an ancient technique. These ovens are usually built near the agricultural fields using the compacted clumps of earth there (k'urpa), and can be about 70 cm in diameter and height. The relatively simple procedure of arching a door, corbeling about
are in a state of disrepair, great care is given during construction to shave and smooth the exterior which can also be fmely plastered, as well as the rectangular portion of the interior. Corbel vaulted structures may also be compared to the more recent house structures found in southeastern Italy known as tru//o (singular) or tru//i (plural), and the use of sod as building material to the sod houses of the north-central United States (see notes 5, 6, and 13, above). The corbel vaulted tru//i (Allen 1969; Rudofsky 1964:figure 49) are built of slabs or slates of field limestone and without cement or mortar, may be plastered on the interior, and are reported to be welladapted to summer and winter conditions. The lower wall portion has a rectangular plan, rectangular door, and windows are rare. The household clusters include several one-room structures adjoining one another which are added onto the original unit as needed, one at a time. The extrem.ely thick walls (5-7 feet) are due to the abundance of stone in the region and the need to clean the fields for agriculture (Wilstach 1930:passim).

ANDEAN PAST 5 (1998) six layers of k 'urpa, heating to red hot the interior, then collapsing the upper portion to place such products as oca and potato, and subsequently compacting and covering the oven with loose soil, provides the root crops with a different and appealing flavor. Furthermore, corbel vaulted structures are also made today as temporary shelters using unworked slabs. We observed such structures in the Huaqoto basalt quarry near Cusco. This site is above the tree line at 4150 m., and is currently quarried by a small population of stone workers who reside in a nearby village. At this quarry, some masons carve stone blocks inside temporary corbeled shelters made of natural stone slabs and open on one side. Some of these have been deserted and, based on the surface pottery and a nearby large stone enclosure, the site may date to Inca times. In addition, it should also be noted that the corbel vaulted.technique was not only employed in covering three-dimensional structures, but also occurred in fa~ades forming doors, corbeled arches, and stepped niches (e.g., at the Island of Koati [Gasparini and Margolies 1980:264, figure 255]; and as doors of xectangular chullpa at Curahuara de Carangas, Oruro Department in Bolivia [Mesa and Gisbert 1966: plate 3]). In this respect, several examples of niches (some with two stories) having corbel vaulted roofs and doors, one about 2 m in height and depth, are found at Huata (Figure 31). This hilltop site is some 29.6 km west of the city of Cusco, and about three hours by foot climbing steep hills to the summit which overlooks the valley around the town of Huarocondo. There are several fortification walls, some remaining to a height of 7 m, rectangular and circular structures (including a chullpa), most built of fieldstone and mud, unlike Inca stone masonry. A short description of the site was published by Rowe (1944:53, plate VII, figure 6, and plate VIII, figures 1-3), and a more extensive reference can be found in Kendall (1976:72-77, plates 15-16, and plan 8). In conclusion, the form of a corbel vaulted structure has been documented on an

-378
Inca building from the Island of Titicaca and on different burial towers and niches which may just predate and/or be contemporary with Inca occupation. Regarding the use of ch 'ampa (aside from the ch 'ampa-built chullpa of unknown date reported by La Barre in Bolivia [1948:95]), no other such structures or remains of households have been reported or preserved archaeologically. Even in the ' Cusco region no Inca sod structures have been preserved (Niles 1987:215), but the use of ch 'ampa must certainly be ancient. Unlike the relatively good preservation of adobe (e.g., K. Chavez 1982:258, figure 10 from a 1000 B.C. context at the site of Marcavalle in Cusco; Moorehead 1979 on Inca structures in the Cusco region), and pirca or round stones set in mud (e.g., K. Chavez 1989:18, figure 1 from the architectural complex at Chiripa in Bolivia dating from 900 to 100 B.C.), ch 'ampa by its very nature may disintegrate and quickly blend with the surrounding soil, making its archaeological detection more difficult. .On the other hand, there is a relative ignorance in the ethnographic and related literatUre about the importance of ch 'ampa as a basic building material in favor of stone and mud (and related uses of mud as adobe or tapia, and in pirca). In fact, a volume entirely devoted to Andean technology (Ravines 1978) makes no references to ch'ampa. At any rate, ch 'ampa should not be underestimated as a readily available building material. The extensive and effective use of the material documented here may have also solved similar structural problems in prehispanic times. These problems include the production of domestic architecture for large populations, especially in regions where severe climatic conditions existed and suitable trees for construction were absent. Therefore, it is possible that the bulk of many prehispanic populations in the region may have lived in structures and settlements not unlike those described here for Yanaoqo. In this respect, even the early structures at the site of Chiripa in Bolivia (ca. 600 B.C.) have now been reinterpreted as part of a temple-storage complex of the Yaya-Mama Religious Tradition, rather

379-

Chavez: Corbel Vaulted Sod Structures rarily inundated areas. Other mounds may indicate raised platforms for structures. Settlement Patterns The bulk of prehispanic populations probably followed a dispersed pattern here, with a major nucleation around ceremoniaJ, centers beginning in the Early Intermediate Period such as at the site of Pucara. For example, Kidder (1943:18), based on archaeological reconnaissance carried out in Taraco and in Saman 2.5 km to the west, indicates that "both places may well have been part of a single scattered settlement with ceremonial centers at various points." Kidder (1956:153) also noted that: "with farmland less restricted than on the coast and with herds to care for, the Titicaca highlanders, at least, were probably widely dispersed at this period [pucara, Tiahuanaco]. I also believe that this continued until fairly late, when large villages and towns begin to appear [just pre-Inca/Inca]." Likewise, John H. Rowe (1946:228-229), describing the Inca towns, states that:
"Most Indian towns grew up without benefit of archite'ct, and the houses were consequently scattered in a haphazard manner along the paths already in use . .. Cobo remarks that such towns had no regular streets or squares, and the houses were quite widely separated. They were built on slopes and rocky places when possible so as not to occupy land that can be cultivated. .. The Inca initiated a policy of urbanization to relocate the Indian towns where they would have easier access to their fields and be further removed from their old forts... The ideal town seems to have been laid out in square blocks, each containing one, two, or four kancha-type enclosures, but this
was modified in practice to fit the topography . .. Inca towns were not much bigger than the

than as a nucleated village (K. Chavez 1989). Likewise, the large and complex Pucara site in Peru (ca. 100 B.C.) still lacks convincing evidence for a truly domestic occupation (S. Chavez 1992). Hence, domestic structures and settlements may still await detection as dispersed rather than nucleated settlements, with houses made of the more perishable ch 'ampa rather than of adobe and/or stone, and containing relatively modest household/cultural remains. The following characteristics may be used as possible archaeological indicators in the detection of ch 'ampa built structures, and possibly even of corbel vaulted structures. The list derives from the ethnographic cases and observations presented here: 1. Following Squier's observations (1877:391392), a collapsed putuku (or any ch 'ampa built house) . would form small mounds, each of about consistent size and shape, and perhaps even forming clusters. Based on our observations, the first blocks to collapse are those on the upper corbeled portion (Figure 15), and the inclination of all walls and the dome toward the interior would probably also encourage further collapse of blocks toward the interior of the room. 2. Pottery and other non-perishable remains may be found in front of and/or inside the structure(s). Sometimes a stone slab is permanently placed in front of the exterior door sill. 3. If the structure had a multi-purpose use, then it is possible that a consistent arrangement of interior features may also be present such as observed in Yanaoqo: as one enters the room, a stove in the comer left of the door, a rocker mill or grinding slab in the left corner opposite the door, and a platform used as a bed on the right side wall. In addition, some kind of pago or offering (like a small pot, for example) should remain buried in the center of the room. 4. Long and narrow mounds visible on the surface may indicate remains of previous ch 'ampa built walls or causeways in tempo-

unplanned towns of earlier periods, and the Inca seem not to have practiced urban concentration in the European sense."

A review of the early historical literature indicates that beginning early in the Colonial Period this pattern was changed. However, two territorial divisions were established. First was the corregimiento for rural Spanish administration (dispersed settlements were moved and nucleated). Subsequently, the capitania was established and set up to obtain labor for the mines. According to Julien

381of Pucarani). Likewise, for the northern portion of the basin, Ortiz Vergara (1965:18-19) remarks that in the great majority of cases Toledo's reforms failed to achieve its goals, and the settlements continued to be dispersed as can be seen today in the Pampa de Have and Taraco. Nevertheless, it should also be pointed out that in areas where Toledo's reforms were successfully implemented, the effects on the traditional settlements were drastic and massive. For example, within the province of Pacajes in the southern basin, Mercado Penaloza (cited in Gisbert 1988:122) reports the reduction of more than 73 ancient settlements into 12 pueblos. The nature of the two sites used in our study and the characteristic sett~ements they represent (the dispersed rural community of Yanaoqo and the urban/nucleated town of Taraco) may certainly also be related to the same forces that shaped the present nature of tQose described for the southern portion of the basin. Yanaoqo and its s':llT0undingrural communities may have maintained or returned to their original dispersed character after the corregimiento. Taraco, on the other hand, retains many characteristics of the Spanish-imposed plan, as proposed by Juan de Matienzo in 1567 (who also accompanied Toledo in his journey through the southern portions of the Viceroyalty): A plaza surrounded by calles, cuadras con solares, casa del padre (streets, blocks of houses, and a rectory) next to the church, and casa de concejo (town hall) (Matienzo 1910 (1567). Here, the social and natural environments in which these communities developed await the input from archaeology and history to document, explain, and incorporate them within a wider Andean context. Specifically, the determination of whether "a site" represents a dispersed or nucleated pattern is an important issue to be considered during archaeological reconnaissance and surface survey. The issue of rural and urban settlements, as well as the use of appropriate building materials and technology, continues to be part of an ongoing process. Despite several expen-

Chavez: Corbel Vaulted Sod Structures sive developmental programs over the last 50 years, instituted and/or supported by the Central Government to "integrate the campesinado into the national society" (Fonseca and Mayer 1988:185), this region continues to be one of the forgotten parts of the country. Substantial populations have migrated to the larger cities attracted by ideals of "progreso .y adelanto"("progress and getting ahead"). On the other hand, places like Taraco, having a population of higher social and economic status, maintain their position in relation to communities like Yanaoqo. In the former, brick and cement structures are already beginning to be built (including a Catholic complex for spiritual retreat, and a cement bridge over the Ramis River halted for lack of funds). In Yanaoqo, in contrast, materials like "tin" or corrugated iron roofs have become so expensive that there is a continued building of structures like the putuku in appropriate technology and materials. On the other hand, a university-trained professional builder knows more about building a brick and cement chalet, is always knowledgeable about modern designs, and is willing to apply foreign and "efficient" materials (such as asbestos-based roof tiles and water pipes in Puno; Teobaldo Yabar, personal communication), wJ;rilehe most likely ignores even the existence of putuku. Likewise, the campesino who migrates to the city is readily available to be trained as a construction worker to help, more often than not, to build a house which he cannot afford to own. In this respect, a statement made by Jorge Flores Ochoa (1988:258) regarding programs of technical cooperation to "improve Puno' s livestock," seems also appropriate here:
"In this manner and with national backing, one will have closed once again a new cycle of neocolonialism, technological dependency and economic domination, sustained by the imposition of foreign technologies, thanks to our myopia or inertia when faced by the Andean techniques, the efficacy of which we do not test experimentally, much less improve, because we do not even know of their existence."25

25. "De esta manera y con apoyo nacional, se habra vuelto a cerrar una vez mas un nuevo cicIo de neocolo-

ANDEAN PAST 5 (1998) Acknowledgements The data upon which this article is based were gathered by the author since 1968, while conducting archaeological reconnaissance and excavation with Karen L. Mohr Chavez in the northern and western portions of the Lake Titicaca Basili. In subsequent trips we were accompanied by Teobaldo Yabar Valencia, then a student of architecture at the University of Cusco. In July 1985, and in August 1988, I returned to the northern portion of the basin to gather more detailed information aimed at developing a research design for an extensive project dealing with this kind of native architecture, a project halted by the military situation in the region. Special thanks are due to Karen L. Mohr Chavez for her appropriate suggestions, encouragement, and editorial assistance. Appreciation also goes to Jeffrey A., Devantier for the time-consumi.ng task of tracing the map in detail. The trips to Peru were made possible in part by several research grants provided by Central Michigan University. . . A paper dealing with the same topic was presented during the 16th Annual Midwestern Conference on Andean and Amazonian Archaeology and Ethnohistory (February 27, 1988), held at the University of Michigan. Furthermore, a shorter Spanish version will appear in La tecnologia en el mundo andino, Volume 2, edited by Heather Lechtman and Ana Maria Soldi, Universidad Nacional Autonoma de Mexico. References Cited
Allen, Edward Sam 1969 Stone Shelters. Cambridge, Massachusetts: The MIT Press. Apaza Suca, N., K. Komarek, D. Llanque Chana, and V. Ochoa Villanueva 1984 Diccionario Aymara-Castellano. Arunakan Liwru Aymara-Kastillanu. Puno: Proyecto Experimental de Educaci6n Bilingue-Puno (Convenio Peru-Republica Federal de Alemania). nialism, dependencia tecnol6gica y dominaci6n econ6mica, sustentados en 1aimposici6n de tecnologias foraneas, gracias a nuestra miopia 0 inercia frente a las tecnicas andinas, cuya eficacia ni siquiera nos ponemos a evaluar en vias de experimentaci6n, mucho menos a mejorar, porque ignoramos hasta su misma existencia."

-382
Aquize Jaen, E1eodoro 1986 Las islas de 10sUros estAncondenadas a desaparecer. EI Comercio p. A-16. Lima. Arquifiiba, Prudencio 1978 Wasi Pirqay/Construcci6n de una Casa. In Tecnologia Andina, edited by Rogger Ravines, pp. 601-603. Fuentes e Investigaciones para 1a Historia del Peru 4. Lima: Instituto de Estudios Peruanos. Ballefteros, Thomas de 1752 Tomo Primero de las Ordenanzas del Peru. . . Recogidas, y coordinadas por el Lic. D. Thomas de Ballefteros. Con Licencia del Superior Govierno. Lima: Imprenta de Francisco Sobrino y Bados. Belai1ndeTerry, Fernando 1961 PirAmidesde los vivos; Campesinos olvidados . de puno viven en 'putucos' que asemejan tumbas fara6nicas. EI Arquitecto Peruano 285/287: 33 and two plates. Lima. Bertonio, Ludovico 1956 [1612] Vocabvlario de la Lengva Aymara (I y II parte). La Paz, Bolivia: Litografia Don Bosco. Chavez, Karen L. Mohr 1977 Marcavalle: The Ceramics from an Early Horizon Site in the Valley ofCusco, and Implications for South Highland Socio-Economic Interaction. Ph.D. dissertation in Anthropology, University of Pennsylvania, Philadelphia. 1982 [1980] The Archaeology of Marcavalle, an Early Horizon Site in the Valley of Cuzco, Peru, Part I. Baessler Archiv, N.F., Band 28, Heft 2: 203-329. Berlin. 1984-85 [1987] Traditional Pottery of Raqch'i, Cuzco, Peru: A Preliminary Study of its Production, Distribution, and Consumption. Nawpa Pacha 22-23: 161-210. 1988 [1989] The Significance of Chiripa in Lake Titicaca Basin Developments. Expedition 30(3):17-26. Special issue on Andean archaeology edited by Karen L. Mohr Chavez. Chavez, Sergio Jorge 1979 The Development of Methods and Theories in Peruvian Archaeology (1524-1900). M.A. thesis in Anthropology, Michigan State University, East Lansing. 1992 The Conventionalized Rules in Pucara Pottery Technology and Iconography: Implications for Socio-Political Developments in the Northern Lake Titicaca Basin (Vols. I-III). Ph.D. dissertation in Anthropology, Michigan State University, East Lansing. Chavez, Sergio Jorge and Karen L. Mohr Chavez 1970 Newly Discovered Monoliths from the Highlands ofPuno, Peru. Expedition 12(4):25-39. 1976 [1975] A Carved Stela from Taraco, Puno, Peru, and the Definition of an Early Style of Stone ~culpture from the Altiplano of Peru and Bolivia. Nawpa Pacha 13:45-83. Cieza de Leon, Pedro de 1962 [1553] La Cronica del Peru. Co1ecci6n Austra1507. Madrid: Espasa-Calpe S.A.

383Cobo, Bernabe 1890-93 [1653] Historia del Nuevo Mundo, edited by Marcos Jimenez de la Espada. Sevilla: Sociedad de Bibli6filos Andaluces, 4 volumes. Cuentas Gamarra, Leonidas 1971 Huancane; Enfoque: antropol6gico, social y econ6mico. Album de oro: monografla del Departamento de Puno, Volume 2, coordinated by Samuel Frisancho Pineda, pp. 49-124. Puno, Peru. Dick, Everett 1979 The Sod House Frontier 1854-1900; A Social History of the Northern Plainsfrom the Creation of Kansas & Nebraska to the Admission of the Dakotas. Lincoln: University of Nebraska Press. Diez de San Miguel, Garci 1964 [1567] Visita hecha a la Provincia de Chucuito por Garci Diez de San Miguel en el ano 1567. Documentos regionales para,la etnologfa y etnohistoria andinas, Volume I. Lima: Ediciones de la Casa de la Cultura del Peru.

Chavez: Corbel Vaulted Sod Structures


Gilson, H.C. 1939 The Percy Sladen Trust Expedition to Lake Titicaca in 1937. I. Description of the Expedition. Transactions of the Linnean Society of London, Series 3, 1:1-20. London. Gisbert, Teresa 1980 Jconografla y mitos indlgenas en el arte. La Paz: Gisbert y Cfa., S.A., Libreros Editores. 1988 Historia de la vivienda y los asentamientos. humanos en Bolivia. Publicacion 431, Instituto Panamericano de Geografla e Historia. Mexico, D.F., Mexico. Gon~aIes Holguin, Diego 1952 [1608] Vocabvlario de la Lengva General de todo el Perv Llamada Lengva Qquichua 0 del Inca. Lima: Instituto de Historia, Universidad Nacional Mayor de San Marcos. Gutierrez, R. et al. 1978 Arquitectura del Altiplano Peruano. Puno, Peru: Resistencia. Guaman Poma de Ayala, Felipe 1936 [1615] Nueva Coronicay Buen Gobierno(Codex Peruvien 1//ustre). Travaux et Memoires de l'Institut d'Ethnologie 23. Paris. Johnson, George R. and Raye R. Platt 1930 Peru from the Air. American Geographical Society, Publication 12. New York. Julien, Catherine J. 1983 Hatunqol/a: A View of Inca Rule from the Lake Titicaca Region. University of Califormia Publications in Anthropology 15. Berkeley. 1989 [1988] The Squier Causeway at Lake Umayo; Notes on Ancient Travel in the Northern Lake Titicaca Basin. Expedition 30(3):46-55. Special issue on Andean archaeology edited by Karen L. Mohr Chavez. Kendall, Ann 1976 Preliminary Report on Ceramic Data and the Pre-Inca Architectural Remains of the (Lower) Urubamba Valley, Cuzco. Baessler Archiv, N.F., Band 24:41-159. Berlin. Kidder, Alfred II 1943 Some Early Sites in the Northern Lake Titicaca Basin. Papers of the Peabody Museum of American Archaeology and Ethnology, Report I, Harvard University 27(1). Cambridge, Massachusetts. 1956 Settlement Patterns-Peru. In Prehistoric Settlement Patterns in the New World, edited by Gordon R. Willey, pp. 148-155. Viking Fund Publications in Anthropology 23. New York. Kubler, George 1946 The Quechua in the Colonial World. In Handbook of South American Indians, volume 2, edited by Julien H. Steward, pp. 331-410 and 8 plates. Bureau of American Ethnology, Bulletin 143. Washington, D.C. La Barre, Weston 1946 The Uru-Chipaya. In Handbook of South American Indians, volume 2, edited by Julian H. Steward, pp. 575-585 and 8 plates. Bureau of American Ethnology, Bulletin 143. Washington, D.C.

Dobyns,HenryF., andPaulL. Doughty .


1976 Peru; A Cultural History. New York: Oxford University Press. Erickson, Clark L. 1989 [1988] Raised Field Agriculture in the Lake TitiCaca Basin; Putting Ancient Agriculture Back to Work. Expedition 30(3):8-16. Special issue on Andean archaeology edited, by Karen L. Mohr Chavez. . Flores Ochoa, Jorge 1988 Distorsiones en el uso de ecosistema de la Puna y los programas de cooperaci6n tecnica." Llameros y Paqocheros; Pastores de Llamas y Alpacas, edited by Jorge Flores Ochoa, pp. 225-271. Cusco: Consejo Nacional de Ciencia y Tecnologfa, Editorial Universitaria UNSAAC. Fonseca, Cesar and Enrique Mayer 1988 Diferenciaci6n campesina en los Andes peruanos. Comunidad y produccion en la agricultura andina, edited by Enrique Mayer, pp. 165-196. Lima: Asociacion Peruana para el Fomento de las Ciencias Sociales (FOMCIENCIAS). Forbes, David 1870 On the Aymara Indians of Bolivia and Peru. Journal of the Ethnological Society 2:193-305. London. GaIdo Pagaza, Raul 1967 Area del Distrito de Capachica. In Tres areas interculturales: Chupaca, Uripa y Capachica, pp. 53-69. Ministerio de Trabajo y Comunidades, Instituto Indigenista Peruano, Unidad de Investigacion y Programacion, Serie Monografica 19. Lima. Gasparini, Graziano, and Luise Margolies 1980 Inca Architecture, translated by Patricia J. Lyon. Bloomington: Indiana University Press. Ghersi, Humberto, y Jose Arquinio 1966 Taraco. In Sociedad, cultura y economia en 10 areas andino-peruanas: Asillo, Chumbao. Chuyas-Huaychao, Huata y Quita, Julcamarca, Layo, Pirapi, Soras, Taraco, Yaucao, pp. 1-68. Ministerio de Trabajo y Comunidades, Instituto Indigenista Peruano, Serie Monografica 17. Lima.

ANDEAN PAST 5 (1998)


1948 The Aymara Indians of the Lake Titicaca Plateau, Bolivia. American Anthropologist 50(1, part 2). American Anthropological Association Memoir 68. La Region 1986 La tragedia. Quincenario ''La Region" 1(5), Febrero. Puno. Loza Balsa, Gregorio 1971 La vivienda Aymara. Pumapunku 3:68-73. Instituto de Cultura Aymara de la Honorable Municipalidad de La Paz. Martinez, Hector 1962 EI indigenay el mestizo de Taraco. Ministerio de Trabajo y Asuntos Indigenas, Plan Nacional de Integracion de la Poblaci6n Aborigen, Serie Monogratica 8. Lima. Matienzo, Juan de 1910 [1567] Gobierno del Peru, obra escrita en el sigloXVI por el Licenciado Don Juan de Matienzo oidor de la Real Audiencia de Charcas, edited by Don Jose Nicolas Matienzo. Buenos Aires: Universidad de Buenos Aires, Facultad de FilosofIa y Letras. Mesa, Jose de 1978 Glosario minimo de terminos d? arquitectura virreynal en el area andina. Cursos de Restauraci6n de Monumentos y Urbanismo, Centro de Investigaci6n y Restauraci6n de Bienes Monumentales. Cusco: Instituto Nacioqal de Cultura-Filial Cusco. Mesa, Jose de, and Teresa Gisbert 1966 Los Chipayas. Anuario de Estudios Americanos 23:479-506. 'Publicaciones de la Escuela de Estudios Hispano-Americanos de Sevilla 174. Sevilla. Monheim, Felix 1963 Contribucion a la climatologia e hidrologia de la cuenca del Lago Titicaca, translated by Carlos Pecka B. trom Beitriige zur Klimatologie und Hydrologie des Titicacabeckens. 1m Selbstverlag des Geographischen Instituts der Universit1it Heidelberg, 1956. Puno: Universidad Tecnica del Alti. piano. Moorehead, Elizabeth L. 1979 [1978] Highland Inca Architecture in Adobe. Nawpa Pacha 16:65-94ff. Muelle, Jorge C. 1978 Tecnologia del barro en el Peru precolombino. In Tecnologia Andina, edited by Rogger Ravines, pp.573-579. Lima: Instituto de Estudios Peruanos. Niles, Susan A. 1987 Ca/lachaca: Style and Status in an Inca Community. Iowa City: University ofIowa Press. Ortiz Vergara, Pedro 1965 Las sub-culturas peruanas. Ministerio de Trabajo y Asuntos Indigenas, Plan Nacional de Integraci6n de la Poblaci6n Aborigen, Serie Monografica 16. Lima. Peru: Censos Nacionales 1981 Censos nacionales de poblacion (Departamento de Puno). Lima: Instituto Nacional de Estadistica, Direcci6n General de Censos y Encuestas.

-384
Ponce Sangines, Carlos 1970 Las culturas Wankarani y Chiripa y su relacion con Tiwanaku. Academia Nacional de Ciencias de Bolivia, Publicaci6n 25. La Paz. Posnansky, Arthur 1937 Antropologia y sociologia de las razas interandinas y de las regiones aclyacentes. La Paz: Instituto "Tiahuanacu" de Antropologia, EtnografIa, y Prehistoria, Editorial "Renacimiento." 1958 Tihuanacu:the Cradle of AmericanMan. Ti3-4. La Paz: Ministerio de Educaci6n. Ravines, Rogger, editor 1978 Tecnologia andina. Lima: Instituto de Estudios Peruanos and Instituto de Investigation Tecnol6gica y Normas Tecnicas. Relaciones Geograficas de Indias, Peru 1881-97 Edited by Marcos Jimenez de la Espada. 4 volumes. Madrid: Ministerio de Fomento. Romero, Emilio 1928 Monografta del Departamento de Puno. Lima: Imprenta T. Aguirre. Rowe, John Howland 1944 An Introduction to the Archaeology ofCuzco. Papers of the Peabody Museum of American Archaeology and Ethnology, Harvard University, volume 27(2). Cambridge, Massachusetts. 1946 Inca Culture at the Time of .the Spanish Conquest. In Handbook of ~outh American Indians, volume 2, edited by J. H. Steward, pp. 183-330. Bureau of American Ethnology, Bulletin 143. Washington, D.C. _ 1963 Urban Settlements in Ancient Peru. Nawpa Pacha 1:1-27. Rudofsky, Bernard 1964 Architecture Without Architects: an Introduction to Non-Pedigreed Architecture. New York: Doubleday and Co., for The Museum of Modem Art. . Sanchez Huanca, Felipe 1970 Vision historia Huancane. Puno, Peru. Smith, A. Ledyard 1940 The Corbelled Arch in the New World. In The Maya and Their Neighbors, pp. 202-221, plate III. New York: Appleton-Century, Inc. Squier, Ephraim George 1871 [1870] The Primeval Monuments of Peru Compared with Those in Other Parts of the World. The American Natura/ist4:1-17. Peabody Academy of Science, Salem, Massachusetts. 1877 Peru, Incidents of Travel and Exploration in the Land of the Incas. New York: Harper and Brothers, Publishers. Trimborn, Hermann 1973 La falsa b6veda en las culturas costeras del antiguo Peru. Revista del Museo Nacional 38:185190. Lima. Tschopik, Marion H. 1946 Some Notes on the Archaeology of the Department of Puno, Peru. Papers of the Peabody Museum of American Archaeology and Ethnology, Harvard University 27(3). Cambridge, Massachusetts.

huanacu;la cuna del hombreamericano,volumes '

385Tschopik, Harry Jr. 1946 The Aymara. In Handbook o/South American Indians, edited by Julian H. Steward, volume 2, pp. 501-573 and 12 plates. Bureau of American Ethnology, Bulletin 143. Washington, D.C. Vazquez de Espinosa, Antonio 1948 [1617] Compendio y descripci6n de las Indias Occidentales, transcribed and with preliminary notes by Charles Upson Clark. Miscellanea Collection 108. Washington, D.C.: Smithsonian Institution. Vega, Andres de 1965 [1582] La descripci6n que se hiz6 en la provincia de Xauxa por la instrucci6n de S.M. que a la dicha provincia se invi6 de molde. In Relaciones geograficas de Indias - Peru, edited by Marcos Jimenez de la Espada. Biblioteca de Autores Espafioles. Madrid: Atlas.

Chavez: Corbel Vau/ted Sod Structures


Welsch, Roger L. 1973 Sod. In Shelter, edited by Kahn Lloyd. Bolinas, California: Shelter Publications. Wilstach, Paul 1930 The Stone Beehive Homes of the Italian Heel. The National Geographic Magazine 57(2):228260.

Table 1. Comparisons of Putuku Attributes in the Northern Lake Titicaca Basin


CHARACTERISTICS Rectangular plan without foundations Approx;aejl;rees of wall inclination Circular vault Rectanlar vault or equivalent Trapezoidal door jamb Rectangular door jamb Entfyway Door sill above ground Use of wooden beams in door lintel Smoke vents at the apex Vents in the upper and lower portions Niches In the interior Wooden supports at the four corners Wooden pole(s) for hanging thinjl;s Mud plaster1Ojl; ofthe extenor Mud plastering of interior lower walls Wooden pegs on interior lower walls Small addition for storajl;espace Use of champa in the construction Use of adobe and/or tapia in construction Putuku wlthm wall enclosures Used for cooking, eatinjl;,and sleepinjl; Used for stor1Ogharvested products Used for hous1Ogdomestic animals Time spent 10construction Labor spent in construction YANAOQO
X.

gu X

TARACO X 5u X X X X X 1 case 1 case X X X

xa
X X X X X X X X XD Xc X X I case X X X XO X 2 days 4 people

X X X X 5 days 4 people

a. This shape has been observed at Yanaoqo only in the tak//a putuku. b. The long poles supporting the rectangular vault of the tak//a putuku are also used to hang things. c. Exterior mud plastering occurs infrequently at Yanaoqo. d. A putuku or takl/a putuku built exclusively to store products occurs only among families who can afford separate putuku for cooking and sleeping.

~
t.!!J

~ ~
> 00
too3

.CJI

-. ......
\.0 \.0 00 --

KEY Provincial capital District capital Isolated house Road Cultivated & flooded zone Flooded zone Seasonal watercourse
HUANCANE

Taraco

. ~':~J

~
"",...-...

Seasonal lake Marsh Uncultivated area

~~
~ m1 .@ill
~

I
<...

!?

10 Kryls

... 10'

Figure 1. Map of the northern end of the Lake Titicaca Basin, based on Carta Nacionall:100,OOO, hoja 31x (Huancane), Instituto GeogrMico Militar, 1964. The arrow indicates the location of the putuku in construction at Yanaoqo. Rivers, lakes and lagoons are indicated in black; the cultivated terrains are in white; and the contour lines indicate elevations at 3850 and 4000 meters above sea level.

0'\

, w 00

387-

Chavez: Corbel Vaulted Sod Structures

Figure 2. View of a household cluster near Yanaoqo composed of three putuku built of ch 'ampa and enclosed within tapia walls. These dispersed household clusters typify the settlement pattern of rural areas.

ANDEAN PAST 5 (1998)

-388

Figure 3. A putuku near Yanaoqo in which guinea pigs are kept. Note the slightly elevated terrain with ch 'ampa surrounded by a temporary flood from one of the lagoons, and the regrowth of grass at the first layer of ch 'ampa.

389-

Chavez: Corbel Vaulted Sod Structures

Figures 4 (left) and 5 (right). Steps in the preparation of ch 'ampa blocks: 4. The chaki-taklla is inserted obliquely into the ground 5. The rectangular block which is trapezoidal in cross section is removed.

ANDEAN PAST 5 (1998)

-390

It ', ,.'.
,.,;, ~

Figure 6. Third step in the preparation of ch 'ampa blocks: The underside is evened out with the chaki-taklla.

391-

Chavez: Corbel Vaulted Sod Structures

Figure 7. View of an area in Yanaoqo after the ch'ampa blocks have been removed. Note the regrowth of grass in the lowermost layers.

Figure 8. Blocks of ch'ampa brought trom Yanaoqo Pampa and stacked near the site where a putuku will be built.

ANDEAN PAST 5 (1998)

-392

Figure 9. View of the wall surrounding the school at Yanaoqo. This 2 m high wall was built of adobe and mortar OIia .foundation of ch 'ampa layers set at ground level for protection against ground moisture, and was capped with ch 'ampa layers above to protect against rain.

Figure 10. View of a household cluster in Yanaoqo composed of a putuku, a small corral on the left, and a gabled structure. The gabled structure is built of adobe with stone foundations which extend about 40 cm above ground. The roof is built of corrugated iron sheets, the door is carpentry made, and the glass windows have metal frames. The stnicture being built is a putuku.

393-

Chavez: Corbel Vaulted Sod Structures

Figure 11. A ch 'ampa block in the process of being reshaped with the chaki-taklla to be used in the corbel vaulted portion of the putuku.

ANDEAN PAST 5 (1998)

-394

Figures 12 (left) and 13 (right). Two views showing the construction of the new putuku: 12. The last ch'ampa block is being placed at the apex. 13. The exterior of the corbel vault is being scraped and evened out with the shovel.

395-

Chavez: Corbel Vaulted Sod Structures

Figure 14. Close-up view of the front of the new putuku in Figures 12-13 Note the vents at the right and above the door, and one of the temporary wooden supports inside the door.

ANDEAN PAST 5 (1998)

-396

Figure 15. An abandoned household cluster in Yanaoqo. Note the doors closed with ch 'ampa blocks, and the collapse of the entryway and apex portions.

Figure 16. A household cluster in Yanaoqo composed of three putuku. Note the putuku in the center which has a convex-walled vault in contrast to the conical vault to the right.

397-

Chavez: Corbel Vaulted Sod Structures

Figure 17. Corbel vaulted structures in the household cluster belonging to Mariano Quispe Bautista, our infonnant andputuku builder, and his wife Paula Caira Larico in Yanaoqo. Aputuku being used for cooking, eating and sleeping. Note the small extension at the right side, and the two stone vessels from the Island of Amantani.

ANDEAN PAST 5 (1998)

-398

Figure 18. Corbel vaulted structures in the household cluster belonging to Mariano Quispe Bautista, our informant and putuku builder, and his wife Paula Caira Larico in Yanaoqo. A putuku built by Mariano's father which was being disassembled in order to rebuild it. Note the adobe gabled structure thatched with totora at the right.

Figure 19. The northern portion of Mariano's household cluster. A corral;attached to the northwest comer walls, in which two of the walls are built of two layers of irregular stone at the bottom followed by ch 'ampa layers on top.

399-

Chavez: Corbel Vaulted Sod Structures

Figure 20. The northern portion of Mariano's household cluster. Mariano standing next to the corral holding a fishing implement he made. Note the adobe wall behind the corral with an additionallayer of ch 'ampa at the top, and the wall surrounding the school of Yanaoqo in the background.

Figure 21. View from Taraco north towards the Ramis River and beyond to the community of Qollana, a rural settlement comprised of many dispersed household clusters.

ANDEAN PAST 5 (1998)

-400

Figure 22. Partial view of Taraco, an example of a nucleated settlement. In the center is the tower of the church bfSan Taraco. Note the two abandoned putuku in the house lot enclosed by tapia walls.

Figure 23. Two corbel vaulted structures within a tapia wall enclosure used for keeping harvested products. The large putuku is part of the.enclosure wall, and the wall in the foreground borders a portion of one of the streets in Taraco.

401-

Chavez: Corbel Vaulted Sod Structures

Figure 24. Exterior view of the corbeled vault seen in the large putuku of Figure 23. Standing next to the doorway of this tapia and adobe built putuku is Francisco Palomino Olivera who was our informant and builder of this structure.

ANDEAN PAST 5 (1998)

-402

Figure 25. Interior view of the corbeled vault seen in the large putuku of Figure 23, showing the technique used in building and closing the adobe corbeled vault. The wood pole set into the north and south walls is used for hanging things.

Figure 26. Abandoned putuku in Taraco. The large putuku was built of tapia and adobe, and the medium-sized one was built entirely of adobe and mortar.

403-

Chavez: Corbel Vaulted Sod Structures

Figure 27. Abandoned putuku in Taraco. Interior view of the corbeled vault of the medium sized putuku. Compare the well defined sides or comers of the corbeled vault here and the purposeful avoidance of them in Figures 24-25.

ANDEAN PAST 5 (1998)

-404

Figure 28. View of the large putuku of Figure 26 built of tapia and adobe.

405-

Chavez: Corbel Vaulted Sod Structures

..'

Figure 29. Drawing of a ch 'ampa-built taklla putuku.

~
r;1

.~
~ ~

> ~
(.A ~

\0 00 '-'

\0

., ,

Figure 30. Plan and section of a circular and a square chullpa in Puno, the fIrst such illustration of chullpa with interior corbel vaults: After . I Squier 1877:353.

I +:o 0\

407-

Chavez: Corbel Vaulted Sod Structures

Figure 31. A corbel vaulted niche about 2 m in height and depth at the site of Huata north of Cusco. Photograph taken by Ann Kendall during a trip to Huata with the author.

> Z ~ ~ > Z ""d > ~ -3 tA ..\0 \0 00 '-"'

Figure 32. View of at least five clustered settlements in the lower valleys of the Qolqa River (in contrast to dispersed settlements in the Lake Titicaca Basin). From an air photograph in Johnson and Platt (1930: figure 28).

I ~ o 00

ARCHAEOMAGNETIC RESULTS FROM PERU: A.D. 700-1500 Daniel Wolfman Museum of New Mexico Richard E. Dodson
Mission Research Corporation

Introduction In recent years considerable progress has been made in the reconstruction of records of the changes in geomagnetic direction using thearchaeomagnetic method. While numerous results have been reported for the northern hemisphere, until recently few measurements have been made on samples collected in the southern hemisphere. During the summers of 1982 and 1983, a total of 134 archaeomagnetic samples, each consisting of approximately eight individually oriented spe~imens were collected at 37 archaeological sites in Peru (Dodson and Wolfman 1983; Wolfman and Dodson 1986). The majority (103) of these samples were collected at. 22 sites on ,the North Coast between the Casma and Lambayeque River valleys. The largest collections were obtained at the sites of Manchan (in the Casma valley), Chan Chan (in the Moche Valley), and Huaca del Pueblo Batan Grande (in the La Leche Valley). The 24 samples in stratigraphic superposition collected at Huaca del Pueblo Bamn Grande were especially critical to the construction of the archaeomagnetic record for the peripd ca. AD 675 - 1450. Nineteen specimens were collected at eight sites in the Rimac and Lurin drainages near Lima, ten were from six sites scattered in the highlands, and two were from a single site on the south coast. Most of the samples were collected from baked features which date from the time period ca. A.D. 550 - 1550. Consequently, the work thus far accomplished has led to the construction of an archaeomagnetic polar curve for most of this time period which is valid for at least a 500 mile diameter circle centered at about 10 S and 78 W (North Central coast of Peru).

Archaeomagnetism in Geophysics and Archaeology The importance of archaeomagnetic data to several disciplines has been emphasized in a number of recent publications (Schurr et af. 1984; Thellier 1981). In geophysics these data provide the most exact information obtainable about the past changes in the direction, magnitude, and variability of the geomagnetic field prior to direct measurements. This information in turn provides constraints on models of the manner in which the field is generated. A great deal of work has been done to develop geomagnetic records from lacustrine and marine sediments. However, the magnetic data obtained from these sources are poor in comparison to those obtained from baked clays in archaeomagnetic studies. In addition, whereas baked clay samples from archaeological sites can often be readily dated by one or more methods, sediments are very difficult to date"reliably. Of equal importance, archaeomagnetic data (p~icularly directional data) obtained from baked clay provide the basis for an archaeological dating method whose precision in many situations is on the order of:l: 20 to :I:50 years at the 95% confidence level. This high precision allows archaeologists to make inferences concerning culture history and processes that are not possible using other chronometric methods with the exception of dendrochronology. This latter method unfortunately is of limited applicability, whereas archaeomagnetic dating can be applied in most areas around the globe. The accuracy of the archaeomagnetic method (i.e., the calibration of the reference curves) is dependent on a variety of factors, including the independent dating methods used in the area (e.g., dendrochronology, radiocarbon), the refinement of the local cultural chronology, and the number of archaeomagnetic samples collected.

ANDEAN PAST 5 (1998):409-420.

ANDEAN PAST 5 (1998) Paleointensity measurements from baked clays and ceramics have been of considerable interest to geophysicists, but, thus far, have been of far less importance in archaeological dating. Some attempts have been made to obtain paleointensity data from sediments, but the validity of these results is somewhat more dubious than the directional data. The greatest disadvantage of archaeomagnetic directional measurements is the length of time required to collect enough samples to develop records of useful length. In contrast, millennia long cores can often be collected from lacustrine and marine sediments in a few hours. However, the time spent to collect samples which yield high quality archaeomagnetic data is worth the effort. Recently, archaeomagnetic data have been used as the standard against which those obtained from sediments are compared and, to some extent, verified. (Verosub and Mehringer 1984; Wolfman et al. 1982). During the past decade, books and articles have reviewed these topics including lengthy discussions of the history of the method and the theory of the origin of the geomagnetic secular variation (Creer et al. 1990; Merrill and McElhinney 1983). These reviews and other recent developments, including the reporting of basic data, indicate that archaeomagnetic studies, which were formerly of central importance in paleomagnetism (e.g., Thellier and Thellier 1959), have recently undergone a resurgence. Field and Laboratory Procedures Field and laboratory procedures, including configuring and calibrating the curve, with minor modifications follow those discussed by Wolfman (1984, 1990). Most samples were collected from ongoing excavations or from sites that had been excavated in the previous one or two years. A few samples were collected from baked features visible on the surface of several archaeological sites (particularly Chan Chan) and, finally, a very few were collected from the sidewalls of looters' excavations. All samples from excavated sites were collected with the cooperation of either the archaeolo-

-410 gists in charge of the work at the site or others who were familiar with the excavations at the site. These archaeologists were able, in most cases, to provide important information about the approximate age of the collected samples based on radiocarbon results as well as ceramic and architectural styles. Each archaeomagnetic sample consists of . six to ten (but usually eight) oriented specimens encased in a plaster cube 1.062 inches on an edge. The specimens were transported to the University of California, Santa Barbara, where each specimen was measured in six orientations in a horizontal three axis SCT cryogenic magnetometer. The specimen mean direction was computed as the mean of the six measurements. Although the average magnetic moments of the individual samples are not reported on the accompanying table, the specimens are usually strong, with typical remanent magnetizations as large as lxl0-3 emu/cm3, with approximately half the remanent magnetization remaining after demagnetization in a 100 Oersted (Oe) alternating magnetic field. Following measurement of the Natural Remanent Magnetism (NRM), each specimen was remeasured after "cleaning" in alternating demagnetization fields of 50, 100, and 150 Oe. The direction of magnetization of most samples did not change significantly with demagnetization beyond 50 Oe. Normally, the mean direction reported for a sample (see Table 1) is at the demagnetization level at which its angular dispersion is smallest. In a few instances where the direction continued to change significantly with increasing demagnetization, the mean direction for a higher demagnetization level than that of the minimum dispersion is reported. The sample mean direction is computed as the mean of all the specimens in the sample unless a specimen direction differs from the sample mean by more than two angular standard deviations. Such "anomalous" specimens are deleted. In addition, experience has shown that ,samples with alpha-95 (a9S) ~ 4.00 often have aberrant directions. For this reason, the results from such samples are not used in curve construction or for archaeological dating. Of the total of 134 samples collected 68 (or 51%) have a9Svalues of less than four de-

411grees. A high proportion, 56 samples (or 42%) of the samples collected, have an (1.95 ~ 2.0. North and Central Coast results: A.D. 550 1550

Wolfman & Dodson: Peru Archaeomagnetism south to the Lambayeque Valley in the north, a distance of approximately 400 miles. The great majority of these samples dated between A.D. 550 and A.D. 1550. Fortunately, the samples were fairly well spread out in this time period and consequently, a curve for most of the time period A.D. 550 - 1550 has been constructed using selected results from 124 samples from 29 sites collected on the, north and central coasts of Peru as well as 10 samples from 6 sites in the North Highlands. Because the samples for archaeomagnetic dating are collected from baked clay features, it is not surprising that sites on the coast, where many structures were made of adobe, produced far more samples than the highlands where stone was the principal construction material. Determination of the configuration of the curve has been greatly assisted by many stratigraphic sequences of samples, particularly 24 collected at the site of Huaca del Pueblo Batem Grande. Radiocarbon dating provided the primary basis for calibration of the curve. The many radiocarbon results from stratigraphic contexts at Huaca del Pueblo BatemGrande were crucial in this phase of the work. In addition, calibration was assisted by knowledge of the end date of the prehistoric period (A.D. 1532) and the generally accepted date of the conquest of the Chimu Empire by the Inca around 1460-1470. Of the 134 samples collected during the two field seasons, 60 were used for the construction of the current curve. Some points may be found on more than one section of the curve, as they may date to more than one time range. Finally, the length of the curve between A.D. 675 and ca. 1500 is ca. 50, suggesting an average rate of secular variation of ca. 1/10 years, which is in very good agreement with the rate found in other parts of the world. While it is a little premature to assign exact dates to the results, a few approximate dates using the curve as constructed are offered here. Encouragingly, they are in very good agreement with information suggested by other dating methods. For convenience the curve has been divided into three sections: 1. A.D. 675 - 1300. This section of curve starts with two samples (109 and 110) col-

The cultural and historical setting of prehispanic Peru as a whole, and of the North Coast in particular, from ca. 2000 BC to the Spanish conquest is fairly well understood. Throughout much of this century, considerable effort has been made to seriate the successive ceramic styles dating from this time. Radiocarbon dating has provided a framework of absolute dates. However, despite the general understanding of the chronological picture, major differences of opinion still exist concerning the absolute dating of major cultural developments on the north coast of Peru from A.D. 900 to 1500 (e.g., Moseley and CordyColllins 1990; Schaedel 1993). One group of archaeologists, basing their chronology in part on ethnohistorical documents, has argued that the Late Intermediate Period began in this. region ca. AD 1200 (Cavallaro 1988, 1997; Conrad 1982; Kosok 1965; Rowe 1945, 1948; Schaedel 1993; Watson 1986). By this reconstruction, the site of Chan Chan (the capital of the Kingdom of Chimor) was established at this time. Others, particularly those involved in excavations at this site, date its founding at ca. AD 900 (Kolata 1982, 1990; Moseley 1975; Topic and Moseley 1983). This controversy continues to be a major stumbling block to cultural understanding on the north coast as illustrated in the proceedings of a recently published Dumbarton Oaks symposium (Moseley and Cordy-Collins 1990). A major goal of the research discussed in this article is to establish a geomagnetic record for Peru which eventually may help to settle this and other controversies. Previous archaeomagnetic work and analysis of the modem geomagnetic field indicates that archaeomagnetic chronologies can be developed that are valid for areas ca. 500 miles in diameter. Although samples were collected in a wider area in this study, the great majority of them came from the north and central coasts from the Lurin Valley in the

ANDEAN PAST 5 (1998) lected at Pampa Grande (Table 1). Shimada (1994) and others (e.g., Haas 1985) have suggested that a major conflagration took place at the end of the Moche V occupation of this site at the neck of the Lambayeque Valley. The results from the two archaeomagnetic samples collected from widely separated baked features, which have almost identical pole positions, support this. The radiocarbon dates from this site (see Appendix in Shimada 1990:372-382; table 2 in Shimada 1994:4-5) suggest a date of ca. A.D. 675 for this great fire. The configuration of the rest of this section of curve is essentially defined, as noted above, by 24 stratigraphically superimposed samples from Huaca del Pueblo Batan Grande. Thirty-one radiocarbon results (see table 3 in Shimada 1995:183-187), which are in excellent agreement with the stratigraphy at this site and associated ceramic material, provided information for the calibration of the curve. Of particular interest, two samples (131 and 132) from the lowest levels of Huaca del Pueblo Batan Grande, associated with Moche V material, appear to date at ca. A.D. 700.. In those cases where there is some independent chronometric information fQr samples collected at other sites, the archaeomagnetic results provide confirmation. For example, two Middle Horizon s~ples (125 and 126) were collected from the Huaca La Pintada section of Pachacamac in the Loon Valley. Based on the chronology inferred from the Huaca del Pueblo Batan Grande samples, an archaeomagnetic date of ca. A.D. 800 is indicated. This is within the accepted time range of the Middle Horizon, and furthermore, is in reasonably good agreement with a recent radiocarbon date of 1180 :l: 70 B.P. (PUCP-83)for this period at this site (Paredes and Franco 1985:80; Shimada 1991:xxvii-xxx). Three samples (35, 107, and 108) from the site of Huaca EI Corte, 15 Ian northwest of Huaca del Pueblo Batan Grande, are thought to date to the early part of the Late Intermediate Period (LIP) based on associated cultural material and radiocarbon dates (see table 3 in Shimada 1995:190). The archaeomagnetic results from these samples suggest a date in the A.D. 1000-1050 range. In addition, two samples collected at Pacatnamu (91 and 92)

-412 which are thought to date between A.D. 1200 and 1300 have pole positions within the expected interval (Figure 2). Finally, one sample (30) from the site of Sur Chayuac at Chan Chan illustrates the problems often encountered in archaeomagnetic dating. This sample, because it was associated with a few sherds of the Black-white-red style, which dates to the . final phase of the Middle Horizon on the north coast (Collier 1955; Mackey 1983; Proulx 1973; Schaedel 1966), had the potential to shed light on the differences of opinion about . the beginning of the LIP in the Moche Valley which is thought to be contemporaneous with the founding of Chan Chan. Unfortunately, due to the fact that the curve loops back over itself, archaeomagnetic dates of both ca. A.D. 875 (Figure 1), which would favor a long occupation for Chan Chan (see e.g., Kolata 1982, 1990; Moseley 1975; Topic and Moseley 1983), and ca. A.D. 1150, which would favor a short occupation (e.g. Cavallaro 1988, 1997; Kosok 1965; Rowe 1945, 1948; Schadel 1993; Watson 1986) are possible. 2. A.D. 1300 - 1500. The great majority of samples from Chan Chan, Manchan, and Tucume have pole positions in what appears to be the A.D. 1300 - 1500 range. However, due to the paucity of samples collected from stratigraphically superimposed baked features at these sites, the exact configuration of the curve in this time period and its calibration is not entirely clear. Examination of Figure 3 indicates that there are two sets of points in this time period. The earlier set includes samples from Tucume, Cerro de los Cementarios, one from Manchan, and the Ciudadelas and Units S, AT, and AW at Chan Chan. The Unit AW sample (29) is under a wall and therefore earlier than the final construction phase in this unit (as represented by samples 26 and 27, which are included in the second group). The second group of samples includes four from Manchan, two from the final construction phase of Unit AW at Chan Chan, and one from Site 38 in the Casma Valley. The close temporal association of the Manchan and the Unit AW samples is in good agreement with the archaeological evidence. While a variety of configurations could be made to fit the data points, at this time the two loops shown in Ta-

413ble 1 seem to best fit the data. However, this could change as more archaeomagnetic results are obtained. The exact end date of this section of curve is not entirely clear. The calibration shown on Figure 3 was developed assuming that all the samples from Chan Chan (including hearths in Unit AW) and nearly all the Manchan samples predate the Inca conquest of the Chimu kingdom, thought to date to A.D. 1460-1470, based on historic documents. Unfortunately, independent archaeological dating for the abandonment of Chan Chan is lacking. 3. Post A.D. 1500. Only four samples which apparently date later than A.D. 1500 have been collected (Figure 3). Based on associated cultural material, one (from Manchan) is Late Horizon in age and one (also from Manchan) dates to the early.Colonial Period. The third (also from Manchan) was thought to date to the LIP and it is not clear whether the archaeomagnetic result is aberrant or the estimated date a little'too early. W}:rile it is conceivable that the post A.D. 1500 curve is as shown in part. by the dotted line on Figure 4, it is also quite possible that a more complex pattern will emerge. The fourth sample was obtained at Chinchero (iri the highlands) and is almost certainly Colonial in age. Concluding Remarks The work described above has led to the development of the first archaeomagnetic curve anywhere in South America. For the field of archaeology this provides a basis for archaeomagnetic dating in the A.D. 675 1550 time period for the Central and North Coast regions and adjacent highlands of Peru. For geophysics, these important new data for the southern hemisphere should provide further constraints on models of the geomagnetic field. In addition, the curve can be used to test the reliability of geomagnetic. direction data obtained from lake sediments from the same area. Future archaeomagnetic work in Peru will be directed at extending and refining the curve developed for the north and central coasts and

Wolfinan & Dodson: Peru Archaeomagnetism

adjacent highlands as well as augmenting the limited number of archaeomagnetic samples thus far collected from southern Peru. References Cited
Cavallaro, Raffael 1988 Architectural Analysis and Dual Organization in the Andes. Doctoral dissertation, Department of Anthropology, Harvard University. Ann Arbor: University Microfilms International. 1997 Architectural Analysis and Dual Organization in the Andes. In Archaeologica Peruana 2, Arquitectura y civilizacion en los Andes prehispanicos/Prehispanic Architecture and Civilization in the Andes, edited by Elizabeth Bonnier and Henning Bischof, pp. 42-61. Sociedad Arqueol6gica Peruana-Alemana, Reiss-Museum, Mannheim, Germany. Collier, Donald 1955 Cultural Chronology and Change as Reflected in the Ceramics of the Viru Valley, Peru. Fieldiana: Anthropology 4. Field Museum of Natural History, Chicago. Conrad, Geoffrey W. 1982 The Burial Platforms of Chan Chan: Some Social and Political Implicl}tions. In Chan Chan: Andean Desert City, edited by M. E. Moseley and K. C. Day, pp. 87-117. Albuquerque: University of New Mexico Press. Creer, K. M., N. Thouveny, and I. Blunk 1990 Climactic and Geomagnetic Influences on the Lac du Bouchet Palaeomagnetic SV Record through the last 110,000 Years. Physics of the Earth and Planetary Interiors 64:314-341. Dodson, R., and D. Wolfman 1983 Los resultados arqueomagneticos de las muestras recogidas en el Peru en 1982. Report submitted to the Instituto Nacional de Cultura, Lima, Peru. Haas, Jonathan Field Archaeology 12:391-409. Kolata, Alan 1982 Chronology and Settlement Growth at Chan Chan. In Chan Chan: Andean Desert City, edited by M. E. Moseley and K. C. Day, pp. 67-85. Albuquerque: University of New Mexico. 1990 The Urban Concept of Chan Chan. In The Northern Dynasties: Kingship and Statecraft in Chimor, edited by Michael E. Moseley and A. Cordy-Collins, pp. 107-144. Washington, D.C.: Dumbarton Oaks. Kosok, Paul 1965 Life, Land and Water in Ancient Peru. New York: Long Island University Press. Mackey, Carol 1983 La cerAmica chimu a fmes del Horizonte Medio. Revista del Museo NacionaI46:73-92. Lima.

1985 Excavations on Huaca Grande: An Initial View of the Elite at Pampa Grande. Journal of

ANDEAN PAST 5 (1998)


Men-ill, R. T., and M. W. McElhinny 1983 The Earth's Magnetic Field: Its History, Origin and Planetary Perspective. New York: Academic Press. Moseley, Michael E. 1975 Chan Chan: Andean Alternative of the PreIndustrial City. Science 187:219-225. Moseley, Michael E., and Alana Cordy-Collins, editors 1990 The Northern Dynasties: Kingship and Statecraft in Chimor. Washington, D.C.: Dumbarton Oaks. Paredes, B. P., and Regulo Franco 1985 Excavaciones en La Huaca Pintada 0 EI Templo de Pachacamac. Boletln de Lima 7(41):78-84.

-414
Topic, John, and M. E. Moseley 1983 Chan Chan: A Case Study of Urban Change in Peru. IVawpaPacha 21: 153-182. Verosub, K. L., and P. J. Mehringer, Jr. 1984 Congruent Paleomagnetic and Archeomagnetic Records from the Western United States, A.D. 750 to 1450. Science 224:387-389. Watson, Richard P. 1986 C14 and Cultural Chronology on the North Coast of Peru: Implications for a Regional Chronology. In Andean Archaeology: Papers in Memory of Clifford Evans, edited by R. Matos and S. Turpin, pp. 83-129. Los Angeles: UCLA Institute of Archaeology. Wolfman, D. 1984 Geomagnetic Dating Methods in Archaeology. In Advances in Archaeological Method and Theory, vol. 7, edited by M. B. Schiffer, pp. 363-458. New York: Academic Press. 1990 Retrospect and Prospect. In Archaeomagnetic Dating, edited by Jeffrey L. Eighmy and Robert S. Sternberg, pp. 313-364. Tucson: University of Arizona Press. Wolfman, D., and R. Dodson 1986 Los resultados arqueomagneticos de las muestras recogidas en el Peru en 1983. Report submitted to Instituto Nacional de Cultura, Lima, Peru. Wolfman, D., W. Kean, and M. Fowler 1982 Archeomagnetic Results from Modoc Rock Shelter. Paper presented at the American Geophysical Union Spring Meeting, Philadelphia.

Proulx,DonaldA.

1973 Archaeological Investigations in the Nepena Valley, Peru. Department of Anthropology, Research Report 13. University of Massachusetts, Amherst.
Rowe, John H.

1945 Absolute Chronology in the Andean Area. American Antiquity 10:265-284. 1948 The Kingdom of Chimor. Acta Americana 6:26-59. Schae4el, Richard P. 1966 Incipient Urbanization and Secularization in Tiahuanacoid Peru. American Antiquity 31:338344. . 1993 Congruence of Horizon with Polity: Huari and the Middle Horizon. In Latin American Horizons, edited by Don S. Rice, pp. 225-261. Washington, D.C.: Dumbarton Oaks. SchUIT. K., H. Becker, and H. C. Soffel

1984 Archaeomagnetic. Study of Medieval Fireplaces at Mannheim-Wallstadt and Ovens from Hen-enchiemsee (Southern Germany) and the Problem of Magnetic Refraction. Journal of Geophysics -56:1-8. Shimada,!. 1990 Cultural Continuities and Discontinuities on the Northern North Coast, Middle-Late Horizons. In The Northern Dynasties: Kingship and Statecraft in Chimor, edited by Michael E. Moseley and A. Cordy-Collins, pp. 297-392. Washington, D.C.: Dumbarton Oaks. 1991 Pachacamac Archaeology: Retrospect and Prospect. In Pachacamac: A Reprint of the 1903 Edition by Max Uhle, pp. xv-Ixvi. Philadelphia: The University Museum of Archaeology and Anthropology, University of Pennsylvania. 1994 Pampa Grande and the Mochica Culture. Austin: University of Texas Press. 1995 Cultura Sican: Dios, riqueza y poder en la costa norte del PerU. Lima: Banco Continental Thellier, E. 1981 Sur la Direction du Champ Magnetique Terrestre, en France, Durant les Deux Dernier Millenaires. Physics of the Earth and Planetary Interiors24:89-132. Thellier, E., and O. Thellier 1959 Sur I'intensite du Champ Magnetique Ten-estre dans Ie Passe Historique et Geologique. Annales de Geophysique 15:285-375.

415-

Wolfman & Dodson: Peru Archaeomagnetism

Peru 675-1050

Figure 1. Archaeomagnetic curve for Peru, A.D. 675-1050.

ANDEAN PAST 5 (1998)

-416

75

1250

()

Peru 1050-1300

igure 2. Archaeomagnetic curve for Peru, A.D. 1050-1300.

ANDEAN PAST 5 (1998)

-418

Peruvian Curve AD. 675-1550

Figure 4. Summary ofarchaeomagnetic curves for Peru, A.D. 675-1550.

Table 1. Archaeomagnetic Results for Peruvian Samples (continued on following page)


jt

-I::\0 I n/N H a95 1.2 1.5 1.7 2.2 0.9 1.0 1.8 2.8 1.8 1.3 1.7 1.1 2.6 2.8 1.4 2.3 1.5 2.2 1.9 2.7 2.0 3.0 2.7 1.3 2.6 1.8 3.5 Lat Long

Site

Feature

Horizon

110 109 132 131 125 126 134 129 128 133 116 35 107 119 117 118 114 97 115 30 37 32 92 91 106 113 105

Pampa Grande Pampa Grande Batan Grande Batan Grande pachacamac pachacamac Batan Grande Batan Grande Batan Grande Batan Grande Las Ventanas Huaca Corte, Huaca Corte Batan Grande Batan Grande Batan Grande Batan Grande C. Cementerios Batan Grande Chan Chan (SC) Batan Grande C Cementerios pacatnamu Pacatnamu Batan Grande Batan Grande Batan Grande

floor 1, s B, m 2 floor, s 16 floor 5, r 40, q 72 floor 3, r 40, q 28A floor (courtyard La Pintada) floor 5 em below 125 oven, f 7, r 34, q 48 oven, f 5, r 33, q 78 85 floor 3, r 35, q 90 floor 3, r 39, q 67 floor, huaquero pit floor, platform mound 1 floor 1, huaquero pit, m 2 floor 9, r 29, q 76B floor 8A, r 29, q 76A floor 7, r 29, q 71 floor 7, r 29, q 76D pit, NE corner p 2, m T floor 5, r 29, q 62-73 pit N1-E96.98, u 2 pit, t 3, r 14, F 49 oven, s III, a 1, su Q, F 25 floor, H-1, F 35 wall, H-1, F 36 floor 5, r 29, q 62 floor 2 & wall, r 31, q 68 floor 2 & wall, r 31, q 67

MH
MH MH MH .MH MH MH MH MH MH

(-

(c (c (?) (?)

650) 650) 650)

LIP (_. LIP (LIP

900) 950)

(- 1050)

LIP (LIP (LIP (-

1150) 1150) 1150) LIP (c 1075) LIP (- 1150) LIP LIP


LIP LIP (- 1250) (- 1400)

LIP LIP

LIP (LIP LIP

1400)

8/8 8/8 8/8 8/8 8/8 8/8 8/8 7/8 8/8 8/8 8/8 7/8 9/9 8/8 8/8 7/8 8/8 8/8 7/8 8/8 6/6 8/8 8/8 8/8 7/7 8/8 8/8

50 100 0 0 100 100 100 50 100 0 50 100 50 50 50 100 100 50 50 100 150 50 50 50 50 50 100

81.5 81.5 81.9 83.0 83.9 82.2 70.7 78.1 80.7 73.9 74.0 78.0 75.5 79.7 77.4 75.2 74.8 80.6 75.5 77.2 77.9 78.3 78.2 77.5 77.4 75.8 74.4

111.2 113.1 133.8 127.4 167.9 177.8 167.0 155.2 175.0 162.2 169.7 165.5 168.5 169.0 164.4 166.1 165.0 170.2 176.7 173.8 174.4 175.0 163.9 159.7 161.4 156.4 155.7

103 Batan Grande 98 Batan Grande 84 Batan Grande


102 101 33 36 19 100 89 Tucume (HM) Tucume (HM) C. Cementerios C. Cementerios
Chan Chan (GC)

floor 13, r 19, q 43 pit, f 2, r 28, q 48 wall 87, f 10, r 19, F 91


floor floor oven, oven, floor floor (top), N side of huaca 2 (S side of huaca) F 27, s III, a 1, su Q F 31, s III, a 5 3, entranceway u D

LIP LIP LIP


LIP LIP
LIP (- 1500)

8/8 8/8 8/8


8/8 7/7 8/8 7/8 7/8

50 50 50
100 100 100 50 50

1.9 2.6 1.3


1.8

78.1 77.4 78.8


78.6 79.2 79.7 81.9 81.3

164.5 165.1 162.9


169.5 161.6 163.7 166.2 172.5

e:,

2.6
2.4 1.0 2.2

LIP
LIP LIP

(- 1500)
.

C f} C '.
""I

Tucume Huaca Campos

99 Tucume 83 Batan Grande 82 Batan Grande 5 Manchan 120 Santa Ana 26 Chan Chan 29 Chan Chan (CP)
58 27 Chan Chan Chan Chan

floor, 60 em below surface


floor, q VIII-C-56 wall 90, f 7-7A, r 24 (SE corner) floor 5-5A, w 78, r 19 (SE corner)

pit 80010, u 024, 1 7


pit, site 38 (Collier & Thompson)

pit (south), u S, b 2
pit, u AW, F 3 (under wall) wall (SW), u AT, wall, u Q, F 1 bricks, burnt post hole (NE corner) pit (NE), u S, b 2

21 Chan Chan 28 Chan Chan (R)

LIP LIP LIP LIP LIP LIP LIP LIP LIP LIP LIP LIP

I I I I I I

LH LH LH LH LH LH

8/8 8/8 7/8 8/8 8/8 8/8 9/9 8/8 8/8 9/9 10/10 6/8
8/8

50 50 0 50 50 50 100 50 50 200 100 50


100

1.5 2.0 3.7 1.4 1.9 1.4 2.4 3.0 3.4 1.9 1.3 3.0
3.4

80.6 79.3 83.0 79.0 79.6 84.0 82.5 80.9 79.8 79.7 80.4 81.0
81.8

175.2 173.0 182.7 171.5 168.7 174.5 173.3 167.7 171.6 168.2 164.8 165.9
162.4

;:

C
::!

<1:> .....

f;;'

::!

Table
#

1. Archaeomagnetic Results for Peruvian Samples (continuedfrom previous page)


Site Feature Horizon n/N H (X.95 1.7 2.7 2.3 1.4 1.1 2.8 0.5 1.9 1.8 2.7 1.1 2.8 Lat Long

22 23 20 6 9 7 12 8 59 127 63 10

Chan Chan Chan Chan Chan Chan (GC) Manchan Manchan Manchan Manchan Manchan 011antaitambo Huaca de Lurin Chincheros Manchan

pit, u AW, F 2 pit, u AW, F 3 floor, u H75-P4 pit 00820, u 221, 1 3 pit 60020/21 (B), u 213, 1 2 pit 00000 , u 250 (surface) pit 60081/82/83/84, u 211, 1 3 pit 60020/21 (A), u 213, 1 2 floor, s I, q D/11, 1 4D wall floor, structure 1
floor, Ii 022, (surface)

LIP / LH LIP / LH LIP / LH LH LH LH


.LH
'

LH
LH LH C C

6/8 8/8 7/8 9/9 9/9 9/9 8/9 9/9 7/8 7/7 7/7 6/8

50 50 50 50 50 50 50 100 100 50 100 50

84.2 83.2 82.3 80.6 83.8 84.7 82.5 85.0 84.8 85.7 82.9 87.0

170.2 178.0 167.8 173.7 164.5 173.7 171.7 140.1 133.6 122.5 164.8 141.8

~ ~ ~ ~
Iood

>00

.VI

~ '-"

..... \0

Key: # Site Horizon = Our sample number. = Archaeological site from which the sample was taken.. = Age of feature from which sample was taken (approximate dates A.D. are listed if available, a "c" indicates that the feature has been radiocarbon dated). = Number of specimens used to calculate the mean magnetic direction of the sample. = Total number of specimens in the sample (n :::; N). = Peak alternating field (in Oersteds) used to "clean" the specimens.

n N H
(X.95

Lat Long

= Radius

of the circle of 95% confidence (in degrees) about the mean magnetic direction of the sample.

= Latitude of the "virtual" geomagnetic pole position corresponding to the magnetic direction = East longitude of the "virtual" geomagnetic pole position

421Addresses of Authors

Addresses of AUthors

Frank Asaro: B70 MSI08B, LawrenceBerkeleyLaboratory, 1 CyclotronRoad, Berkeley, California94720. E-mail: F_Asaro@lbl.gov Wilton BarrionuevoOrosco: PP.JJ. Manco Capac L-A-16, Santiago, Cusco, Peru. Brian S. Bauer: Universityof Dlinoisat Chicago, Dept. of Anthropology, 1007 West Harrison Street, Harrison, Dlinois60637. E-mail: bsb@uic.edu Ran Boytner: Departmentof Anthropology,341 Hains Hall, Universityof California, Los . Angeles, California90095-1553.
E-mail: rboytner@uc1a.edu David L. Browman: Washington University, Dept. of Anthropology, Campus Box 114, St. Louis, Missouri 63130-4899. E-mail: dlbrowma@artsci. wustl.edu Richard L. Burger: 12 Chestnut Lane, Woodbridge, Connecticut 06525.

E..mail: Richard.Burger@yale.edu

Sergio J. Chavez: Department of Sociology/Anthropology, Central Michigan University, Mt. Pleasant, Michigan 48~59. . E-mail: Sergio.Chavez@cmich.edu Jose Ccencho Huamani: Jr. Inca Roca 614, Lima 28, Peru Carolyn S. Dean: Department of Art History, D-207 Porter College, University of California at Santa Cruz, Santa Cruz, California 95064. E-mail: carolyn_dean@macmail.ucsc.edu Richard E. Dodson: Mission Research Corporation, 735 State Street, PO Drawer 719, Santa Barbara, California 93102-0719. E-mail: dodson@mrcsb.com Michael D. Glascock: University of Missouri, 223 Research Reactor, Columbia, Missouri 65211. E-mail: glascock@reactor.murr.missouri.edu Catherine Julien: Dept. of History, Western Michigan University, Kalamazoo, Michigan 49008-5020. E-mail: catherine.julien@wmich.edu Vincent R. Lee: Design Associates, Box 107, Wilson, Wyoming 83014. E-mail: sixpacmanco@compuserve.com Johan Reinhard: The Mountain Institute, Main & Dogwood Streets, P.O. Box 907, Franklin, West Virginia 26807.

E-mail: reinhard@sympatico.ca
Jack Rossen: Ithaca College, Ithaca, New York 14850. E-mail: jrossen@ithaca.edu

ANDEANPAST 5 (1998)

-422

Guido Salas Alvarez: Departamentode Geologia, UniversidadNacional de San Augustinde AJequ~a,CamllaI203,Arequ~a,Peru. KatharineJ. Schreiber: Universityof California, Dept. of Anthropology,Santa Barbara, California93106. E-mail: schreibk@sscf.ucsb.edu Izumi Shimada: SouthernIllinois University, Dept. of Anthropology,Faner Hall, Carbondale, TIlinois 62901-4502. E-mail: ishimada@siu.edu Fred Stross: LawrenceBerkeleyLaboratory, 1 CyclotronRoad, Berkeley, California 94720. E-mail: FHStross@lbl.gov John R. Topic: 83 Robinson St., Peterborough, Ontario K9H 1E9, Canada. E-mail: jtopic@trentu.ca Paul R Trawick: Departmentof Anthropology,211 Lafferty Hall 0024, Universityof Kentucky,Lexington, Kentucky40506. E-mail: pbtrawl@pop.uky.edu Dwight T. Wallace: 66 Elm Street, Albany, New York 12202. E-mail: CWallace66@aoJ.com
.

Daniel Wolfman: died, Santa Fe, New Mexico, November25, 1994.

Das könnte Ihnen auch gefallen