GDA Arquitectos

P1: JPJ/KNP
0521842026agg.xml CB885B/Lancaster 0 521 84202 6 April 18, 2005 20:36
This page intentionally left blank

P1: JPJ/KNP
CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME
Concrete Vaulted Construction in Imperial Rome examines the methods and techniques
that enabled builders to construct some of the most imposing monuments of ancient
Rome. Focusing on structurally innovative vaulting and the factors that influenced
its advancement, Lynne Lancaster also explores a range of related practices, including
lightweight pumice as aggregate, amphoras in vaults, vaulting ribs, metal tie bars,
and various techniques of buttressing. She provides the geological background of
the local building stones and applies mineralogical analysis to determine material
provenance, which in turn relates to trading patterns and land use. Lancaster also
examines construction techniques in relation to the social, economic, and political
contexts of Rome, in an effort to draw connections between changes in the building
industry and the events that shaped Roman society from the early empire to late
antiquity.
Lynne C. Lancaster is assistant professor of classics at Ohio University. An architect and

archaeologist, she is a Fellow of the American Academy in Rome and has published
in a variety of journals, including American Journal of Archaeology, Journal of Roman
Archaeology, and Römische Mitteilungen.
i
P1: JPJ/KNP
ii
P1: JPJ/KNP
CONCRETE VAULTED CONSTRUCTION

IN IMPERIAL ROME
INNOVATIONS IN CONTEXT
LYNNE C. LANCASTER
Ohio University
iii
cambridge university press
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
Cambridge University Press

The Edinburgh Building, Cambridge cb2 2ru, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521842020
© Lynne C. Lancaster 2005
This publication is in copyright. Subject to statutory exception and to the provision of

relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.
First published in print format 2005
isbn-13 978-0-511-16068-4 eBook (EBL)

isbn-10 0-511-16068-2 eBook (EBL)
isbn-13 978-0-521-84202-0 hardback

isbn-10 0-521-84202-6 hardback
Cambridge University Press has no responsibility for the persistence or accuracy of urls
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
P1: JPJ/KNP
For Tom
v
P1: JPJ/KNP
vi
P1: JPJ/KNP
I believe that in architecture, as in all art, the artist instinctively keeps the marks
which reveal how a thing was done.
– Louis I. Kahn
vii
P1: JPJ/KNP
viii
P1: JPJ/KNP
CONTENTS
List of Illustrations and Tables page xiii

Preface xix
1 INTRODUCTION 1
A Note on Monuments and Previous Scholarship 2
How to Use this Book 2
The Nature of Roman Concrete 3
Concrete Vaulting during the Republic 3
Structural Behavior of Concrete Vaults 6
Roman Mathematical and Analytical Background 10
Materials, Transport, and Production 12
The Building Industry in Rome 18
The Innovations 21
2 CENTERING AND FORMWORK 22

Assembling the Centering 22
Removing the Centering and Formwork 26
The Use of Brick Linings on the Intrados of Vaults 29
Evidence for Centering Arrangements 32
Barrel Vaults, 32 • Cross Vaults, 34 • Domes and Semidomes, 40
Conclusions 48
3 INGREDIENTS: MORTAR AND CAEMENTA 51

Mortar 51
Lime, 53 • Pozzolana, 54 • Cocciopesto, 58
ix
P1: JPJ/KNP
CONTENTS
Caementa 59
Provenance of Caementa, 64
Conclusions 65
4 AMPHORAS IN VAULTS 68
Types of Amphoras Used 69
Early Examples of Amphoras in Vaults 69
Examples of Amphoras in Late Antique Vaults 75
Conclusions 81
5 VAULTING RIBS 86
Early Development of Vaulting Ribs and Relieving Arches 86
The Vaulting Ribs at the Colosseum 88
The Use of Solid Brick (Bipedalis) Ribbing 91
Ladder and Lattice Ribs in Barrel Vaults 98
Ladder and Lattice Ribs in Cross Vaults 106
Ribbing in Domes 108
Conclusions 111
6 METAL CLAMPS AND TIE BARS 113

The Earliest Use of Tie Bars 115
Tie Bars in the Imperial Thermae 116
Tie Bars at the Forum of Trajan 118
Spacing of Tie Bars 125
The Use of Tie Bars with Lightweight Caementa 126
Iron Architrave Bars and Impost Blocks at Hadrian’s Villa 126
Conclusions 127
7 VAULT BEHAVIOR AND BUTTRESSING 130

Arches and Barrel Vaults 132
Cross Vaults 134
Domes and Semidomes 138
Conclusions 146
8 STRUCTURAL ANALYSIS: HISTORY AND CASE STUDIES 149

Historical Development 149
Basilica Ulpia 156
“Temple of Mercury” at Baiae 156
Pantheon 158
x
P1: JPJ/KNP
CONTENTS
“Temple of Minerva Medica” 161

Conclusions 164
9 INNOVATIONS IN CONTEXT 166

Accumulated Knowledge 166
Evident Need 168
Economic Ability 170
Social/Cultural/Political Acceptability 172
appendix 1. catalogue of major monuments 183
appendix 2. catalogues of building techniques 205

A. Formwork Imprints on Barrel Vaults 205
B. Formwork Imprints on Domes 206
C. Formwork Imprints on Semidomes 206
D. Brick Linings on the Intrados of Vaults 207
E. Uses of Cocciopesto to Protect the Extrados of Vaults 211
F. Lightweight Caementa 213
G. Amphoras in Vaults 215
H. Vaulting Ribs 216
I. Blocks with Tie Bar Cuttings 221
appendix 3. scoria analysis 222
appendix 4. thrust line analysis 225
Notes 231
Glossary 245
Works Cited 251
Index 265
xi
P1: JPJ/KNP
xii
P1: JPJ/KNP
LIST OF ILLUSTRATIONS AND TABLES
Frontispiece. “Temple of Minerva Medica” in the Horti Liciniani.
figures
1. Sanctuary of Jupiter Anxur. Terracina. page 5
2. Diagram of principal parts of an arch. 6
3. Diagram of stress patterns in a beam. 7
4. Section D at Baths of Trajan. Rome. 8
5. Diagram of deformation of vault due to creep. 9
6. Sanctuary of Fortuna Primigenia. Palestrina. 9
7. Sketches of the Sanctuary of Fortuna Primigenia. Palestrina. 11
8. Reconstruction of wooden centering for concrete barrel vault. 23
9. Wooden trusses at San Pietro and San Paolo Fuori le Mura. Rome. 23
10. Marble panel with depiction of wooden amphitheater. 24
11. Detail of plaster cast from Trajan’s Column. Italica, Spain. 24
12. Scarf joints. 25
13. Frame saws. 26
14. Wooden centering. Olympia, Greece. 27
15. Detail of Blackfriar’s Bridge. G. B. Piranesi. 28
16. Diagram of method of lowering the centering of vaults. 29
17. Reconstruction of centering using brick linings. 29
18. Detail of vault built using brick linings. Case a giardino, Ostia. 30
19. Heliocaminus Baths at Hadrian’s Villa. Tivoli. 31
20. Pons Fabricius. Rome. 32
21. Pons Cestius. Rome. 33
22. Engravings of Pons Cestius. G. B. Piranesi. 33
23. Reconstruction of centering frames at Pont du Gard. A. Leger. 34
24. Barrel vaulted room under San Clemente. Rome. 35
25. Common deformation pattern for centering frames. 36
26. Basilica of Maxentius. Rome. 36
27. Basilica of Maxentius. Rome. 37
28. Reconstructions of formwork for a pavilion vault and a cross vault. 38
xiii
P1: JPJ/KNP
29. Detail of cross vault from the Colosseum. Rome. 38

30. Reconstruction of the centering scheme for the Aula at Trajan’s
Markets. Rome. 39
31. Comparison of centering schemes (same scale) for cross vaults at Trajan’s
Markets and the Baths of Caracalla. Rome. 39
32. Diagram of domes showing formwork layouts. 40
33. Reconstruction of the centering at the “Temple of Mercury.” Baiae. 41
34. Formwork impressions on the octagonal dome at the Domus Aurea. Rome. 42
35. Geometry of centering for the octagonal dome at the Domus Aurea. Rome. 43
36. Pantheon section. Rome. 44
37. Reconstruction of Pantheon centering. M. Viollet-le-Duc. 45
38. Reconstruction of formwork for segmental domes. 47
39. Excavation photo of the Serapeum at Hadrian’s Villa. Tivoli. 47
40. Diagram showing the cycle of chemical changes in lime mortar production. 53
41. Quicklime at the Casa del Sacello Iliaco. Pompeii. 54
42. Section of Mausoleum of Helena. Rome. 57
43. Detail of vault covering from the Basilica Argentaria. Rome. 59
44. Reconstruction of horti pensiles at the Domus Tiberiana. Rome. 60
45. Vesuvian scoria at the Forum of Caesar. Rome. 61
46. Section of the Pantheon showing distribution of materials. Rome. 62
47. Mausoleum of Helena with amphoras in dome. Rome. 69
48. Amphora types found in vaults in Rome. 70
49. Cross vault with amphoras. Magazzini “Traianei,” Ostia. 72
50. Plan of the Villa alla Vignaccia. Rome. 72
51. Villa alla Vignaccia with amphoras in walls and vaults. Rome. 73
52. Fallen vault from the Casa de la Exedra. Italica, Spain. 73
53. Diagrams showing uses of amphoras for land reclamation projects. 74
54. Wall containing amphoras. Pompeii. 75
55. Balcony vault of the “Casa di Via Giulio Romano” with amphoras. Rome. 76
56. Circus at the Villa of Maxentius with amphoras in vaults. Rome. 77
57. Sketch of dome of the “Temple of Minerva Medica.” Rome. 78
58. Octagonal Hall at the “Villa of the Gordians.” Rome. 79
59. Semidome at Santa Maura with amphoras. Rome. 83
60. Plan and section of St. Jerome. Cologne, Germany. 84
61. Acropolis gate at Alatri, Italy. 87
62. Porta Rosa in city walls of Velia, Italy. 87
63. Theater of Marcellus. Rome. 88
64. Sanctuary of Hercules Victor. Tivoli. 89
65. Drawing of the Theater of Marcellus. Rome. 89
66. Section through the Theater of Marcellus. Rome. 90
67. Four types of vaulting ribs at the Colosseum. Rome. 90
68. Drawing of travertine ribs at the Colosseum. Rome. 91
69. Section and partial plan of the Colosseum showing location of ribs. Rome. 92
70. Sketch section at bay 34 of the Colosseum. Rome. 93
71. Detail of Flavian ladder rib at the Colosseum. Rome. 93
72. Axon of the Domitianic Vestibule. Rome. 94
73. Plan of lower level of the Domus Augustana. Rome. 95
xiv
P1: JPJ/KNP
74. Sketch plan of cryptoporticus at the Villa of Domitian. Castelgandolfo. 95

75. Cryptoporticus at the Villa of Domitian with ribs. Castelgandolfo. 96
76. Cryptoporticus at the Villa of Domitian with ribs and
coffers. Castelgandolfo. 97
77. Section E at the Baths of Trajan. Rome. 97
78. Drawing of the Trajanic latrine in reconstructed part of the Forum
of Caesar. Rome. 98
79. Trajanic latrine at the Forum of Caesar with travertine impost
blocks. Rome. 99
80. Drawing of the Pantheon with its ribbing system. Rome. 100
81. Exterior of the Pantheon. Rome. 101
82. Plan of the Villa di Sette Bassi. Rome. 102
83. Ladder rib in the cryptoporticus of the Villa di Sette Bassi. Rome. 102
84. Ladder rib at the Villa di Sette Bassi. Rome. 103
85. Plan of the Severan Baths and the Baths of Maxentius on
the Palatine. Rome. 103
86. Lattice ribs at the Severan Baths on the Palatine. Rome. 104
87. Nymphaeum Alexandri “Trophies of Marius.” Rome. 104
88. Detail of rib in outer ambulatory of the Colosseum. Rome. 105
89. Detail of lattice ribbing at the Baths of Maxentius on the Palatine. Rome. 105
90. Detail of nave façade of the Basilica of Maxentius. Rome. 106
91. Detail of ladder rib in cross vault at the Villa di Sette Bassi. Rome. 107
92. Lattice ribs in the cross vaults of the Maxentian substructures on
the Palatine. Rome. 107
93. Dome at the Baths of Agrippa with lattice ribbing. Rome. 108
94. Plans of domes showing restored rib patterns. 109
95. Section of “Temple of Minerva Medica.” Rome. 111
96. Architrave block from the Forum of Trajan with remains of iron
tie bar. Rome. 114
97. Reconstructed section of the Horrea Agrippiana. Rome. 115
98. Cornice block with tie bar cutting from the Horrea Agrippiana. Rome. 116
99. Reconstruction of the portico fronting the Basilica Aemilia. Rome. 117
100. Measured sketch of bucranium block from the portico fronting the
Basilica Aemilia. Rome. 117
101. Bucranium block from the portico fronting the Basilica Aemilia. Rome. 118
102. Section H at Trajan’s Baths showing holes for anchor blocks for
tie bars. Rome. 119
103. West palaestra of the Baths of Caracalla. Rome. 120
104. East palaestra of the Baths of Diocletian. Rome. 120
105. Generic reconstruction of the palaestra portico from an imperial
bath showing tie bar and anchor block. 121
106. Cornice blocks from Trajan’s Forum. Rome. 121
107. Comparison of proposals by C. M. Amici and J. Packer for
reconstruction of the Column Portico at Trajan’s Forum. Rome. 122
108. Comparison of proposals by C. M. Amici and J. Packer for
reconstruction of the Basilica Ulpia. Rome. 122
109. Plan of the Basilica Ulpia. Rome. 123
xv
P1: JPJ/KNP
110. Reconstruction of lintel construction at the Hall of the Doric Pilasters at

Hadrian’s Villa. Tivoli. 124
111. Hall of the Doric Pilasters at Hadrian’s Villa. Tivoli. 125
112. Arcuated lintels at Santa Costanza. Rome. 127
113. Diagram of two proposed design schemes used at the Pantheon. Rome. 131
114. Diagram showing failure patterns of arched structures. 132
115. Diagram showing four factors that affect the stability of an arched structure. 133
116. Drawing showing vaulting configurations at the Market at Ferentino, the
Market at Tivoli, the Domus Aurea, and Trajan’s Markets at Rome. 134
117. Diagram showing the lines of compressive force in a cross vault. 135
118. View of the Aula at Trajan’s Markets. Rome. 136
119. Gallery overlooking the central hall of the Aula at Trajan’s Markets. Rome. 136
120. Frigidarium of the Baths of Caracalla. Rome. 137
121. Aerial view of the Basilica of Maxentius. Rome. 138
122. Plan and section of the Basilica of Maxentius. Rome. 139
123. Diagram of construction of dome built of stone voussoirs and diagram of
stress patterns in an uncracked concrete dome. 139
124. Deformation patterns in a cracked dome and a semidome. 140
125. Step-rings on the Pantheon dome. Rome. 141
126. Large semidome at Trajan’s Markets. Rome. 141
127. Stairs on dome of the “Temple of Mercury.” Baiae. 142
128. Buttressing walls at the “Temple of Mercury.” Baiae. 143
129. Plan of dome of the “Temple of Mercury.” Baiae. 144
130. Drawing of the buttresses of the octagonal dome at the Domus
Aurea. Rome. 144
131. Drawing of the “Temple of Venus and Cupid” in the Sessorian
palace. Rome. 145
132. Buttress at the “Temple of Venus and Cupid” in the Sessorian
palace. Rome. 145
133. Diagram demonstrating the concept of the vector. 150
134. Diagram demonstrating the concept of moment. 151
135. Diagram of stable three-hinge arch and of collapsing four-hinge arch. 151
136. Diagram showing the concept of Hooke’s hanging line. 152
137. Example of a funicular polygon diagram. 153
138. Thrust line diagrams for three reconstruction proposals of the Basilica
Ulpia. Rome. 157
139. Thrust line through dome of the “Temple of Mercury.” Baiae. 158
140. Definition of the Rankine safety factor. 159
141. Thrust line diagram of the Pantheon dome under different conditions. 159
142. Thrust line diagram of the “Temple of Minerva Medica” under
different conditions. 162
143. Drawing of the “Temple of Minerva Medica.” Franz Innocenz Kobell. 163
144. Dome of the “Temple of Diana.” Baiae. 164
145. Drawing demonstrating the steps in making a funicular polygon analysis
of an arch. 226
xvi
P1: JPJ/KNP
maps
1. Map of Rome and environs locating major monuments discussed in
the text. 4
2. Map of Italy showing major volcanic districts. 13
3. Geological map of Rome and environs showing locations of major building
materials used in concrete vaulting. 14
4. Map of Italy showing locations of major sources of timber and limestone. 15
5. Map of the Bay of Naples showing the air fall distribution of the a.d. 79
eruption and the pre-79 eruption of Mt. Vesuvius. 63
tables
1. Recommended Mortar Mix Proportions 55
2. Amounts of Scoria in Vaults in Rome 67
3. Physical Characteristics of Dressel 20 and Dressel 23 Amphoras 71
4. Blocks from Trajan’s Forum with Cuttings for Tie Bars 123
5. Proportions of Barrel Vaults 133
6. Proportions of Cross Vaults on Exterior Walls 135
7. Proportions of Freestanding Domes 140
8. Scoria Analysis 223
color plates
Color plates follow page 154.
I.Major local building stones used in vaulted structures around Rome.
II.Detail of opus caementicium vault from the Porticus Aemilia. Rome.
III. Four types of pozzolana.
IV. Abandoned pozzolana quarry on via Nesazio. Rome.
V. Detail of mortar with white pumice from the Colosseum. Rome.
VI. Plan of the Baths of Trajan. Rome.
VII. Section H at the Baths of Trajan. Rome.
VIII. Scoria and tufo giallo della via Tiberina at Section E at the Baths of Trajan. Rome.
IX. West palaestra at the Baths of Caracalla. Rome.
X. East palaestra vault from the Baths of Diocletian. Rome.
XI. Dome at the “Temple of Minerva Medica.” Rome.
XII. Tor de’Schiavi. Detail of painting by Henrik van Lint and
reconstruction drawing.
XIII. Plans of the Baths of Caracalla and the Baths of Diocletian showing
vaulting ribs. Rome.
xvii
P1: JPJ/KNP
xviii
P1: JPJ/KNP
PREFACE
as a student of architecture at virginia tech I have been fortunate to work with people who
traveling in Italy in 1985, I became fascinated with have provided the intellectual grounding to tackle the
the ancient brick walls that had obviously inspired problems that interest me. After graduating and then
one of my favorite architects, Louis Kahn. He had vis- working as an architect for a couple of years, I de-
ited Rome as a Resident at the American Academy in cided to get a Master’s degree in classical archaeol-
1950, and later much of his work was designed around ogy. I went to Oxford University to study with Jim
themes of brick arches. The arches in Roman archi- Coulton, whose sharpness of mind and interest in
tecture, and particularly the relieving arches inserted problem solving always enabled him to ask just the
into solid walls, captured my imagination. I won- right question to guide me where I needed to go. As
dered what secret reasons the Romans had for scat- a bonus, two other scholars interested in architecture
tering these elements throughout their buildings. At happened to be at Oxford at the time on postdoc-
the time, I was inspired by what I saw as the Roman toral fellowships, Janet DeLaine and Hazel Dodge.
“honesty” in their use of materials, though I now real- Margareta Steinby then came to All Souls for a year,
ize they were probably not remotely interested in this and the architect, Sheila Gibson, who worked with
modernist concept. (I have also come to admit that I John Ward-Perkins for many years, lived in Oxford.
like most Roman buildings as ruins much better than I With this core group and others, we all presented our
would have liked them in their original state.) On my work at a series of architecture seminars where I had
return from the study abroad program, I convinced the benefit of studying with an intense group of schol-
my architecture professor, Dennis Kilper, to supervise ars in my formative years. This was also a time when I
an independent study project on Roman concrete was spending months at a time at the British School at
construction, the final product of which would be an Rome where I came to know Amanda Claridge, who
illustrated paper. In the end, it was based largely (if not has always both inspired and humbled me with her
exclusively) on information from Vitruvius and M. E. intimate knowledge of ancient Rome, and Andrew
Blake, and the illustrations were never completed. In Wallace-Hadrill, who has been a constant supporter
the present work, I hope to have remedied the short- of my work. John Lloyd, a pottery and field survey
comings of that first project begun two decades ago. expert, came to Oxford when I did and became a
xix
P1: JPJ/KNP
PREFACE
great influence on my archaeological education. His can be seen in the color plates, was generated from our
unexpected and much too early death was a blow, early excursions together looking at Roman rocks.
and to his memory I have dedicated Chapter 4 on However, before long my initial chapter on build-
amphoras. ing materials had grown to twice the length of any
This project on vaulting began at Trajan’s Markets. other chapter, and I realized that this was the topic
I knew that I wanted to study some aspect of the of a separate work. The information presented here
monument for my Master’s thesis, and when I went is a fraction of what I collected, and I hope to
to meet with Lucrezia Ungaro, the person in charge present a more detailed analysis of building materials
of the Markets, for the first time in 1989, she sug- in the future. Later, in 2001, I met another geologist,
gested that I concentrate on the vaulting. Eventu- Fabrizio Marra of the Istituto Nazionale di Geofisica e
ally the Master’s thesis was expanded into a doctoral Volcanologia in Rome. His help and encouragement
dissertation on concrete vaulted construction from with the scoria analysis in Appendix 3 were invalu-
Nero to Trajan, the premise of which was to investi- able, and he acted as my “cicerone,” taking me to
gate the fifty-year period leading up to the construc- explore abandoned pozzolana quarries. Finally, in the
tion of the Pantheon dome. This book, the proverbial field of geology, I want to thank a group of scholars
“thesis book,” derives from that study, but the scope with whom I have worked on a mortar dating project
is much expanded. After completing the dissertation, and who have inspired me to pursue more actively
I realized that some of the most interesting questions interdisciplinary projects: John Hale (archaeologist),
were those that dealt with the differences between Jan Heinemier (physicist), Alf Lindroos (geologist),
the construction of the high empire and that of late and Asa Ringbom (art historian).
antiquity. As a publication strategy I decided to con- I have long been interested in the engineering
centrate first on publishing the detailed information aspects of ancient construction and had read struc-
I had collected on particular monuments, such as the tural studies on historical buildings by Robert Mark,
Colosseum and Trajan’s Markets, so that I could re- Roland Mainstone, and Jacques Heyman. I could per-
fer to them later in a more general work on concrete form simple beam analyses from what I had learned
vaulting. During the past decade, the book has been in college, but I aspired to do thrust line analyses
growing in the background as I familiarized myself and discovered that I needed tutoring. In 2002, I re-
with the radically different world of late antiquity in ceived notice of the First International Congress of
an effort to understand how the construction industry Construction History in Madrid, where Heyman was
changed. billed as the keynote speaker, so I decided to attend.
This book is more “interdisciplinary” than the dis- In the end, we met and he very kindly agreed to
sertation, and I would not have been able to tackle read an early draft of what is now Chapter 8. He also
some of the more technical issues were it not for introduced me to one of his former students who
help from other experts, especially in the fields of was an organizer of the conference, John Ochsendorf
geology and structural engineering. In 1996, I at- at MIT. With John’s help and many exchanges over
tended a lecture in Rome on the rocks of ancient e-mail and telephone, I learned how to make my own
Rome by the geologist Marie Jackson, who has since thrust line analysis, and I have presented the basic steps
helped tremendously with my geological education. in Appendix 4 for anyone brave enough to have a go
The emphasis on geology throughout this book, as at it. Thus, I have been fortunate to meet experts in
xx
P1: JPJ/KNP
PREFACE
other fields whose broad visions and generous spirits industry and legal issues; Ted Peña on the amphoras
have aided me in answering questions that I might and trade; Marie Jackson and Fabrizio Marra on ge-
not have otherwise asked. ology; and Jacques Heyman, Robert Mark, and John
Many people contributed to my thoughts and Ochsendorf on structural analysis. Finally, I am grate-
ideas about Roman construction through conver- ful to my former Ohio University student, Mandy
sations and excursions. Margareta Steinby has been White, for her careful reading of and insightful com-
a constant source of information, inspiration, and ments on the penultimate draft of this manuscript.
support over the years. Mark Wilson Jones has of- One complication for a project such as this one
fered many insights about the design process and is the need for numerous permessi for various mon-
has provided a refreshing counterpoint to my fo- uments around Rome, and many friends and col-
cus on construction. I have benefited greatly from leagues have been instrumental making the arrange-
Robert Coates-Stephens’s knowledge of ancient and ments. Lucrezia Ungaro, Roberto Meneghini, and
medieval Rome. Peter Rockwell, a sculptor/scholar, Riccardo Santangeli Valenzani of the X Ripartizione
taught me to carve a Corinthian capital when I was AA.BB.AA of the Comune di Roma have been won-
in Rome in 2001, an invaluable experience that pro- derfully supportive colleagues in all my endeavors. I
vided me with a different perspective on the con- also am indebted to support I have received over the
struction process. I also have benefited greatly from years from those at the Soprintendenza Archeologica
insights and on-site visits with Jim Packer to the di Roma including Irene Jacopi, Giangiacomo
Forum of Trajan and am grateful for his support Martines, Cinzia Conti, and Maria Letizia Conforto.
of this project. Others with whom I have had en- I owe special thanks to Rossella Rea, who is in
lightening conversations include: Jane Aiken, Larry charge of the Colosseum, for her encouragement of
Ball, Heinz Beste, Elisabetta Bianchi, Mario Como, my work there. With regard to obtaining permessi, I
Lucos Cozza, Clayton Fant, Shawn Graham, Michael am especially grateful to Maria Pia Malvezzi at the
Heinzelmann, Henner von Hesberg, Shawna Leigh, British School at Rome for arranging an often com-
Giovanni Manieri Elia, Archer Martin, John Oleson, plicated schedule of visits and sometimes working her
Bob Ousterhout, Betsey Robinson, Rabun Taylor, own miracles during the early research stages of this
Bill Wallace, and Roger Wilson. project. I also thank Anne Coulson of the American
Two colleagues deserve special thanks for reading Academy for help in arranging permessi during my
substantial portions of the manuscript and for shar- Rome Prize year in 2001.
ing their own (often unpublished) research with me, Financial support both for the research and the pro-
Carla Amici and Janet DeLaine. I have had the great duction of this book came from a variety of sources.
joy of engaging in many long conversations and on- The M. Alwyn Cotton Foundation has been gener-
site visits to monuments with each, and I consider ous in funding my year in Rome in 1996–1997 and
them my closest intellectual companions. I also owe providing a subvention to pay for the color plates.
debts of thanks to others outside of my own area For my second research year in Rome in 2001–2002,
of expertise who have taken the time to read and I owe thanks to the American Academy in Rome
comment on various chapters of early drafts of this for awarding me the Phyllis G. Gordan Post-Doctoral
manuscript (and are in no way responsible for any mis- Rome Prize Fellowship and to the Graham Founda-
takes in the final one): Susan Martin on the building tion for Advanced Studies in the Fine Arts. I also have
xxi
P1: JPJ/KNP
PREFACE
received generous support from the Department of Finally, without my husband, Tom Carpenter, this
Classics and World Religions, the College of Arts and project and my interest in archaeology would never
Sciences, and the Vice President for Research at Ohio have happened. From the beginning, he taught me
University for travel and for book production costs. the value of reason, evidence, and the positivistic ap-
At a personal level I am grateful to a number proach and has been my most profound intellectual
of people who have helped me along the way: to model. He has given generously from his own re-
my parents for supporting my education and to my search schedule to help me with mine at every level –
grandmother for her long-term planning; to Charles measuring, climbing, hauling me out of holes on a
Knight for giving me a construction job when I was rope, reading too many drafts of this manuscript, serv-
an undergraduate; to Michael and Mariella Stannus, ing as a sounding board, and at the end even helping
friends in Rome who have always been there for me me finish the illustrations as my deadline loomed. I
in sickness and in health; to Bailey van Hook for also am grateful to his remarkable patience with me
her encouragement during hard times; and to Brian during emotionally tumultuous periods and during
Rose for companionship in Rome and for hospital- the past two years when I was married to my key-
ity in Cincinnati during research trips to the Blegen board as I completed the text and illustrations. To
Library. him I dedicate this book.
xxii
P1: NDZ
0521842026c01.xml CB885B/Lancaster 0 521 84202 6 April 17, 2005 14:8
1
INTRODUCTION
C oncrete vaulted structures represent one

of the ancient Romans’ most original and en-
during contributions to the artistic and architectural
ploit the fireproof nature of concrete and in doing so
created a new aesthetic based on the plastic potential
inherent in the material.1 In imperial Rome, all of
patrimony of the Mediterranean world. A combina- the natural advantages and cultural influences came
tion of factors led to the development of the large together and manifested themselves in imposing con-
spans and curvilinear forms still visible in buildings crete vaulted structures, the remains of which are the
such as the Pantheon and the Basilica of Maxentius. focus of this study.
Rome was endowed with a wealth of natural re- My intention is to examine the changes that oc-
sources in its immediate environs, and what it could curred in the choice of materials and techniques used
not supply for itself it could bring in from afar through in concrete vaulted construction in Rome from the
the development of extensive trade networks. Along time of Augustus to Constantine and to place the re-
with the financial benefits of conquest came the ar- sults in the wider social, economic, and political con-
chitectural, technological, and mathematical exper- text. I document the appearances of particular mate-
tise of the architects, builders, and engineers from the rials and building techniques and examine the reasons
conquered territories. Augustus, in bringing the civil for their use and the ways that use changed over time.
wars to an end, also brought a vision of urban renewal In particular, I am interested in techniques that aided
for Rome that provided incentive for more grandiose in the creation of large and complex structures, such as
schemes than had previously been possible. By that the use of lightweight concrete, brick vaulting ribs,
time, the architects and builders had over a century metal tie bars, and various forms of buttressing. In
of collective experience with concrete construction, some cases, the choices of the builders were affected
but Augustus’s creation of an organizational infras- by external factors such as the availability and the cost
tructure provided a context in which new ideas and of materials or the changes in the infrastructure of the
larger building schemes were possible. As emphasized building industry itself. The interplay between the
by W. L. MacDonald, the fire that devastated much decisions made on the building site and these exter-
of Rome during Nero’s reign in a.d. 64 effectively nal factors can create a window into the complexities
cleared the slate and provided opportunities to ex- of urban and suburban life in Rome.
1
P1: NDZ
a note on monuments and previous combining my own on-site observations with those
scholarship of others to create an overview of the developments.
This would not be possible without the publica-
The monuments included in this study date from the
tion of monographs during the past few decades by
reign of Augustus (27 b.c.), when the resources of
scholars conducting fieldwork on some of the major
the Mediterranean basin became widely available in
monuments in Rome: K. de Fine Licht on the
Rome, to the reign of Constantine, when patron-
Pantheon (1968), the Baths of Trajan (1974), and Sette
age was diverted to the new capital inaugurated at
Sale (1990); C. M. Amici on the Forum of Trajan
Constantinople (a.d. 330). The monuments are for
(1982) and the Forum of Caesar (1991); J. E. Packer
the most part limited to buildings in the city of
on the Forum of Trajan (1997); J. DeLaine on the
Rome and its immediate environs because I am par-
Baths of Caracalla (1997); and J. J. Rasch on a se-
ticularly concerned with the local materials and the
ries of late Roman domed structures including the
economic, social, and political factors unique to the
Tor de’Schiavi (1993) and the Mausoleum of Helena
capital city. Many of them are state-sponsored public
(1998). The engineering works of J. Heyman (1995,
monuments, such as the imperial thermae, basilicas,
1996) and R. Mark (1982, 1990) have been par-
and places of public spectacles like theaters and am-
ticularly influential in my approach to the struc-
phitheaters. Some are imperial residential structures
tural aspects of vaulting. I also have drawn on
such as the palaces on the Palatine, the domed pavil-
numerous articles by archaeologists, geologists, and
ion in the Horti Sallustiani, or the nymphaeum in
engineer/architects working in Rome as well as on
the Horti Liciniani (“Temple of Minerva Medica”).
the invaluable resource of E. M. Steinby’s Lexicon To-
Some structures in the immediate outskirts of Rome
pographicum urbis Romae (1993–2000). Although the
also are included, such as the Villa alla Vignaccia,
study is in the spirit of previous works on Roman
the Villa di Sette Bassi, and the so-called Villa of the
construction such as those by M. E. Blake (1947,
Gordians. Further afield are two imperial villas,
1959, 1973), G. Lugli (1957), J.-P. Adam (1994), and
Domitian’s Villa in the Alban hills and Hadrian’s Villa
C. F. Giuliani (1990), my focus is narrower and my
near Tivoli, both of which demonstrated innovative
inquiry delves deeper into specific issues relating to
vaulting techniques that relate to developments in
the construction of large-scale concrete vaulting.
Rome itself. During the early fourth century, domed
mausolea often located on suburban villas became
popular, and these extramural structures are also ex-
how to use this book
amined. One monument important to this study is
located outside of the immediate environs of Rome. The book is organized so that it can be used by both
The structure, known as the “Temple of Mercury” at general readers and specialists. The material in the
Baiae on the Bay of Naples, is both the earliest pre- remaining sections of this chapter and in the final
served concrete dome and the largest spanned dome chapter (“Innovations in Context”) is intended to
before the Pantheon and hence must be considered in provide general discussions accessible to a wide au-
any discussion of the development of concrete vaults. dience. Each of the other chapters is provided with a
One goal of the present work is to provide a syn- brief introduction to the major issues and a conclu-
thetic study of the concrete vaulting in Rome by sion that includes a broader overview and assessment
2
P1: NDZ
INTRODUCTION
of the material discussed within the chapter. A gen- concrete were laid separately, by hand and trowel.
eral reader can read the first and last chapters of the In both ancient and modern concrete construction,
book as well as the beginning and end of each chap- some type of structure, or centering, is necessary to
ter to get an idea of the issues discussed and their contain and model the wet mortar until it sets and
relevance, whereas the specialist can delve into the gains strength.
details of the arguments presented within the chap- The mortar of the Romans was stronger than the
ters. I also have provided catalogues in Appendix 2 earlier mortar used in Greek architecture because of
listing all of the documented examples of a particular the addition of a local volcanic material called poz-
technique, many of which are not discussed in the zolana, which creates a chemical reaction that results
text. For those who want to pursue the subject fur- in a mortar much more tenacious than simple lime
ther, these tables provide detailed information about mortar. Furthermore, pozzolana mortar is hydraulic
every entry along with bibliographic references. In and sets underwater. Mortared construction was used
addition, I have included in Appendix 1 a catalogue outside of Rome and Italy, but locally available ingre-
of the main monuments discussed in the text and a dients were often substituted. Because each ingredient
map with their locations (Map 1, p. 4). For readers has a unique effect on the final mixture, distinguish-
not familiar with a particular monument, Appendix 1 ing between mortars from different areas is critical.
provides a catalogue with an introduction to each one For example, both O. Lamprecht and R. Malinowski
followed by a list of the relevant vaulting techniques provide useful studies of ancient Roman mortar, but
with cross-references to discussions in the text. A their samples are not from Rome itself.2 In recent
glossary of technical terms used is also provided at years, Italian engineers and geologists, often working
the end of the book. with preservationists, have become more active in the
analysis of mortar and concrete samples from build-
ings in Rome and Ostia, and, in Chapter 3, I have
the nature of roman concrete incorporated these results in an effort to provide the
most relevant information regarding the local mortar.
Roman concrete, or opus caementicium, is different
from what we think of today as concrete. The word
caementa means rough, unhewn quarried stones and
concrete vaulting during the republic
refers to the rubble of fist-sized pieces of stone or
broken brick that were used in the mortar as ag- The development of concrete vaulting during the
gregate. As implied by its name, the concrete in an- Republic has been covered admirably by W. L. Mac-
cient Rome is more akin to a type of mortared rub- Donald and others,3 so in what follows I limit myself
ble (Pl. II) than to modern concrete, which consists to a brief introduction of the major developments
of mortar mixed with an aggregate of much smaller before the time of Augustus. Pozzolana mortar and
stones usually ranging in size from a pea to a walnut. concrete walls probably developed as early as the late
The way that ancient and modern concrete is put third century b.c.,4 but the use of concrete for vault-
in place is also different: Modern concrete is literally ing came somewhat later. One of the earliest and most
poured into place over a network of steel reinforc- spectacular examples of concrete vaulting in central
ing bars, whereas the caementa and mortar of Roman Italy is at the Sanctuary of Fortuna Primigenia at
3
P1: NDZ
map 1. Map of Rome and environs locating the major monuments discussed in the text.
4
P1: NDZ
INTRODUCTION
Palestrina (ancient Praeneste). The sanctuary has been

known since the Renaissance, but the upper sanctuary
was only uncovered after bomb damage during World
War II revealed parts that had been built into mod-
ern structures. The dating of the sanctuary has been
controversial. It originally was assumed to have been
built after Sulla’s occupation of the city in 82 b.c.,5
but G. Gullini in a monograph on the monument
proposed a mid-second century b.c. date, to which
G. Lugli strongly objected.6 A. Degrassi, in a study of
the epigraphic material, supported a pre-Sullan date
of the monument but was unwilling to accept such
an early one and proposed that the monument was
constructed in the last decade of the second century
b.c.7 The weight of the evidence leans toward a late
second century date, which makes it the earliest of a 1. Sanctuary of Jupiter Anxur (first half of the first century b.c.).
View of concrete vaulted platform overlooking the Tyrrenian Sea
series of spectacularly sited, terraced sanctuaries that at Terracina. Fototeca Unione c/o American Academy in Rome,
employed concrete vaulting including the sanctuar- neg. #5139.
ies of Hercules Victor at Tivoli, of Jupiter Anxur
at Terracina (Fig. 1), and of Hercules Curinus near
Sulmona, all of which have been dated to the first By the first half of the first century b.c., con-
half of the first century b.c.8 crete vaulting was firmly established in Rome, as it
Early examples of vaulting in Rome itself are rare, was in the towns of central Italy. The Tabularium,
in part because larger and more impressive imperial which is dated by an inscription to 78–65 b.c.,12
buildings replaced many of them. Traditionally, the was one of the earliest concrete vaulted structures
earliest datable concrete vaulted structure in Rome in the heart of Rome. Like the hilltop sanctuaries,
has been assigned to the remains of a large struc- it served the structural purpose of shoring up the
ture located between the Tiber and Monte Testac- face of the Capitoline. Within its façade of peperino
cio, but once again controversy reigns. In 1934, G. blocks, the Tabularium contained a series of pavilion
Gatti associated this structure with a fragment of the vaults and barrel vaults.13 Some two decades later,
Severan Marble Plan that clearly represents the vis- Rome received its first permanent theater dedicated
ible remains. A partial inscription [—]LIA survives by Pompey in 52 b.c. In breaking a long-standing tra-
on the fragment, and he interpreted it as the Porti- dition within the Senate of not allowing permanent
cus Aemilia,9 which Livy tells us was reconstructed theaters or amphitheaters to be built as places for large
in 174 b.c.10 Recently, this reading of the inscrip- gatherings, Pompey opened the gates for experimen-
tion and the association of it with the remains of the tation in vaulting for the substructures of such build-
Porticus Aemilia has been challenged,11 potentially ings. Some early innovations in vaulting techniques
leaving us with no datable concrete vaulted remains can be found in similar structures, such as the Theater
from second-century-b.c. Rome. of Marcellus and the Colosseum.
5
P1: NDZ
So, what prompted the early development of con-

crete vaulting in central Italy? As seen earlier, the
most spectacular early uses were in the hilltop sanc-
tuaries, but by the first century b.c., vaulting also
could be found in other types of structures such as
the storage/market buildings at Ferentino and Tivoli
and in bath buildings at Pompeii.14 Part of the an-
swer certainly lies in the available natural resources
and in the financial resources generated by conquests
outside of Italy by this time, but cultural influences
also affected the early development. The hillside set-
tings of the terraced sanctuaries were influenced by
Hellenistic Greek types, such as the Sanctuary of
Athena at Lindos on Rhodes (second century b.c.)
and the Sanctuary of Asclepius at Cos (first half of sec-
2. Diagram showing principle parts of an arch and its behavior.
ond century b.c.).15 Incentives to use the new vaulted
construction also came from within Italy itself. Con-
crete vaulting provided both an economical and fire- modern reinforced concrete structures seem to have
proof means of storage for the goods coming from such limited life spans in comparison to ancient ones.
the conquered territories, and it was a particularly The success of Roman concrete structures is often
suitable material for enduring the constant moisture attributed to the strength of the pozzolana mortar.
present in bath buildings that were becoming increas- In fact, this is only part of the explanation. Just as
ingly popular. important is the relationship between the masses and
By the time of the Augustan peace when routes forms making up the structure. Structural form was a
of transport were opened and craftsmen flocked critical factor in the success of Roman buildings. The
to Rome, concrete vaulting had become common, interplay between form and material was ultimately
and during this period the early attempts at more the key to longevity.
sophisticated vaulting techniques began to appear. The arch, which was originally developed for stone
The preceding century had provided the context construction, was the basis for the formal develop-
for the acceptance of vaulting, but once the scale ment of concrete vaulting. Recent findings show that
of the buildings began to grow and the spans be- builders in Rome were using arches of cut stone vous-
came larger, the builders had to deal with structural soirs by the sixth century b.c.16 Voussoirs are wedge-
challenges that had not been relevant in earlier shaped stones that make up an arch (Fig. 2). The radi-
times. ating joints between the voussoirs serve to direct the
weight of the arch and anything it supports toward
the sides and away from the opening under the arch.
structural behavior of concrete vaults
The result is that the arch pushes out at its springing,
Roman concrete vaults are known for their longevity, and this outward thrust must be countered or con-
and many visitors to Rome today often ask why our trolled in some way. If the arch is built into a wall, the
6
P1: NDZ
INTRODUCTION
ning a distance so that the stresses within the material

remain in compression. Tension can develop within
an arch, but it can be controlled by the form, size, and
loading pattern of the arch. The mechanics of arch
and vault behavior and methods of structural analysis
are explored further in Chapter 8.
Concrete vaults take forms similar to arches built
in cut stone, but their behavior is somewhat different.
The forces are not transferred by means of the joints
between individual voussoirs but, rather, through the
mortar between the pieces of caementa, which by the
imperial period were laid in horizontal courses. As
long as the mortar is strong enough to resist any ten-
sile stresses that develop as a result of these factors, the
concrete can act as a solid monolithic block once it
3. Diagram showing the stress patterns in a beam with a point has cured and gained its strength, and lateral thrusts
load applied at center.
will not occur. If too much tension develops then
surrounding masonry acts as a buttress to contain the cracks occur and the vault begins to push outward,
horizontal thrust. or to display lateral thrust, on its supports, just like
The strength of any material is measured in terms the voussoir arch. As long as the thrust is sufficiently
of stress, which can occur as compression (compressive countered the structure will remain stable, but if the
stress) or tension (tensile stress). Compressive stress re- supports cannot resist the lateral thrust the structure
sults when the atoms in a material are pressed together collapses. The success of the Roman builders was in
in the direction of the converging forces. Tensile stress their ability to control the outward thrust of vaulted
results when a material is stretched so that the atoms structures through the choice of form and materials.
are pulled apart in the directions of the opposing The modern understanding of the behavior of
forces. The example of the man on the beam in Fig- Roman concrete has undergone changes during the
ure 3 shows both types of stresses within the beam. As past century. J. H. Middleton, writing at the end of
the beam bends downward under the man’s weight, the nineteenth century, commented that “the Roman
the upper half is in compression because the top sur- concrete vault was quite devoid of any lateral thrust
face is squeezed together and becomes shorter, and and covered its space with the rigidity of a metal
the lower half is in tension because the lower surface lid.”17 This idea of the monolithic concrete vault that
is stretched. At a point in the middle of the beam, has no horizontal thrust was repeated by such no-
there is a neutral axis that is not undergoing tension table scholars as M. E. Blake, J. B. Ward-Perkins, and
or compression. The strength of the beam is its abil- J.-P. Adam,18 but it remained controversial through-
ity to resist the different types of internal stresses that out much of the twentieth century. It is based on
occur under various loading situations. Because both the assumption that concrete made with pozzolana
concrete and stone are very strong in compression and mortar has the strength to resist any internal ten-
weak in tension, the arch provides a means of span- sile stresses that could cause cracks to develop. Both
7
P1: NDZ
With the increased interest in preservation

since the end of World War II, more engi-
neers have become involved in the analysis
of historical structures. As a result of the
analytic approach they bring to the disci-
pline, the traditional view of monolithic
concrete long held among some classical
scholars has been modified to acknowl-
edge that, in spite of the high-quality
pozzolana mortar used by the Romans
builders, lateral thrust often occurred and
had to be countered.
Roman concrete vaults commonly de-
veloped cracks as can be seen in standing
remains of many structures,21 including
such imposing ones as the Pantheon, the
Baths of Trajan (Fig. 4), and the Basilica of
Maxentius. The cracks could occur for a
number of reasons. If the tensile stresses
within the concrete exceed the tensile
strength of the material, cracks will de-
velop. The level of such tensile stresses can
be controlled through the judicious design
of structural form. However, even when
the stresses are normally very low, exter-
nal factors can cause sudden increases. A
common example is a dramatic change
4. Baths of Trajan (a.d. 104–109). Detail of the exedra at section D showing cracks in temperature that results in sudden ex-
in wall supporting semidome (29.5-m span). pansion or contraction, which can cause
the tensile stresses to spike and a crack
W. L. MacDonald and G. Lugli were more circum- 22
to occur. (A similar reaction is observable when a
spect in their assessment of the structural behavior of cold egg is dropped into boiling water and immedi-
concrete vaulting noting that the monolithic qualities ately cracks.) Moreover, concrete is subject to a phe-
actually depend on the size of the vault.19 One of the nomenon called creep, which is slow deformation over
more influential studies affecting the understanding time. In concrete vaulting this usually results in a flat-
of vault behavior has been the extensive documen- tening of the curve of the vault and a spread at the
tation of cracking and deformation in the concrete haunches (Fig. 5). The gradual change in form creates
structure of the Hagia Sophia in Istanbul by R. Van changes in the patterns of stresses within the concrete,
Nice, R. Mainstone, R. Mark, and A. S. Çakmak.20 which can then lead to cracking.
8
P1: NDZ
INTRODUCTION
During the second century b.c., when con-

crete vaulting was in its infancy, the builders were
constructing fairly small vaults (typically 5 m or less),
which could have acted monolithically, and evi-
dence from the Sanctuary of Fortuna Primigenia at
Palestrina (second half of the second century b.c.)
suggests that these early builders did not take pre-
cautions to counter lateral thrusts.23 In two places at
the sanctuary, vaults were supported on at least one
side by a trabeated system of columns and architrave
blocks. On the Terrazza degli Emicicli, the concrete
vault (3.7 m span) was built of radially laid caementa
of limestone on the flat upper surface of the traver-
tine architrave (Fig. 6).24 Metal clamps were not
typically used to hold the architrave blocks together,
5. Diagram indicating deformation of barrel vault due to creep.
which suggests that the builders did not expect the
concrete to push laterally against them but, rather,
to bear straight down. A similar condition occurred
6. View of the “Terrazza degli emicicli” of the Sanctuary of Fortuna Primigenia at Palestrina (second
half of the second century b.c.).
9
P1: NDZ
elsewhere in the complex on the Terrazza della divided into either 12 oncia (inches) or 16 digiti (digits).
Cortina (Fig. 7), except there the vault was built of Measuring sticks often had two sets of divisions, one
caementa of the lighter local tuff.25 Both examples had for inches and one for digits.28 The Roman foot var-
coffers in the vault, which G. Gullini suggested were ied somewhat from place to place, but it was usually
intended to lighten the vault while creating a type of about 29.5 cm, which is somewhat smaller than the
ribbing between the coffers.26 In this early example modern foot (30.5 cm).
of concrete vaulting, the builders evidently assumed a The appearance of concrete vaulting comes after
degree of strength in the concrete that later imperial the death of Archimedes (212 b.c.), who provided the
builders did not. mathematical means of estimating volumes of spheres
By the time of Augustus, the builders clearly re- and the areas of conic sections. Such calculations were
alized that once the span of the vaults increased and clearly relevant for concrete construction by the first
the support structure became less massive, they had century a.d. Heron of Alexandria, who gave credit
to take some precautionary measures to counter any to Archimedes, included a section in his Stereometrica
lateral thrust that could develop. They must have explaining how to calculate amounts of materials for
learned (perhaps the hard way) that once cracks de- the curving forms of various types of vaults.29 Heron
veloped in a vault, it began to push out on its support also wrote a treatise called On Vaulting (Camarika),
structure and would collapse if the thrust was not about which Isodorus of Miletus (mid-sixth century
countered. We have little evidence for those exper- a.d.) wrote a commentary.30 Unfortunately, neither
iments that did not work, but by this time builders Heron’s treatise nor Isodorus’s commentary has sur-
had begun to think of ways of reducing the horizontal vived, but the fact that Heron devoted an entire work
thrust, such as choosing lightweight stones as caementa to the subject in the second half of the first century
and using metal clamps to stabilize the stone support a.d. just at the time that concrete vaulting became
structure. the norm in imperial Rome is in itself significant.
Archimedes was famous for shunning the practical
uses of his theoretical discoveries, but the Romans
roman mathematical and analytical
had no such qualms.31
background
Advances in mathematical and geometrical knowl-
With the adoption of concrete, the methods of cal- edge also would have affected the understanding of
culating the necessary materials for building projects the relationships between masses, which govern struc-
changed. For cut stone vaults, the architect would tural form. One of the fundamental principles for un-
have calculated the number of blocks needed, whereas derstanding the behavior of masses is the concept of
for concrete vaults he would have calculated the vol- the center of gravity, another Archimedean contribu-
ume of the vaults and ordered a certain amount of tion. The center of gravity of an object is the point at
lime, pozzolana, and caementa depending on the pro- which the object will balance as if the whole weight
portions of each he intended to use in the concrete of the object is concentrated at that point, as on a ful-
mixture.27 This type of calculation would have re- crum. The development of modern structural theory
quired measuring units (as opposed to number of was ultimately based on this concept (see Chapter 8).
blocks), which for the Roman builders was typically By the first century a.d., Heron was concerned with
in terms of pedes (Roman feet = RF), which could be explaining it. In solving various structural problems
10
P1: NDZ
INTRODUCTION
7. Sanctuary of Fortuna Primigenia at Palestrina (second half of the second century b.c.). Sketches
showing details of areas with colonnades supporting concrete vaults. The architraves are marked with “A.”
in his Mechanics, he was clearly thinking in terms of Heron’s interest in vaulting is representative of the
geometry, the balancing of masses, and the ratios that change in attitude toward vaulting that took place
governed the relationships between bodies.32 If his during the first century. Vitruvius, writing toward the
treatise on vaulting had survived, we surely would end of the first century b.c., barely mentions vaulting
have found similar thought processes as the ones ex- in his treatise, although he does give space to the ma-
pressed in the Mechanics, although there is no evi- terials of concrete in Book 2. By the time Heron was
dence to suggest that the Romans ever developed the writing during the second half of the first century
means to calculate actual thrusts. Archimedes may a.d., vaulting was significant enough to have war-
have provided a way of thinking about arched and ranted its own treatise. The great fire in Rome under
vaulted structures, but ultimately the Romans’ con- Nero in a.d. 64 is often seen as a turning point in
trol of their materials and forms must have come the development of concrete construction, but the
through a combination of experimentation on the turning point it represents is not so much in the im-
building site and the understanding of basic geomet- mediate creation of new vaulting techniques as it is in a
rical principles. new attitude toward design and the control of light and
11
P1: NDZ
space. Concrete offered the potential for new forms seminal works of T. Frank (1924), M. E. Blake (1947),
and combinations of space that eventually prompted and G. Lugli (1957),33 but advances in the mapping
the development of new and innovative construction and dating of the volcanic activity in Italy in the past
techniques. Heron was writing at the time when this half century and particularly in the past decade have
transition was in full swing. Unfortunately for those yet to be integrated into much of the archaeologi-
of us interested in vaulting, Heron’s treatise was lost cal literature. A difficulty that arises for the present-
and Vitruvius lived just a bit too early. day archaeologist interested in the geology of Roman
building materials is that the terminology used in the
standard archaeological works is quite different from
materials, transport, and production
the geological nomenclature, which in itself has varia-
The choice and availability of materials played an im- tions. I use the modern Italian geological names of the
portant role in the creation of large spanned vaults. various volcanic materials employed by the ancient
Rome lies along the Tiber River between two vol- Roman builders and provide the equivalent archaeo-
canic districts, the Monti Sabatini to the north and logical terms in the Glossary. In this study, I avoid the
the Colli Albani to the south (Map 2). These volca- term “tufa,” which has traditionally been adopted by
noes produced a variety of building stones used by the English-speaking archaeologists describing the stone
Romans: dense lavas for road building, heavy tuffs for made of volcanic ash, in favor of “tuff,” which is the
cut stone construction, lightweight tuffs for vaulting, more precise and the preferred geological term, as
and pozzolana for mortar. Sedimentary stones, such “tufa” also can refer to a type of sedimentary stone. I
as limestone and travertine, for making the lime for use the Italian term “tufo” when referring to specific
mortar were found in central Italy and in the Apen- named types of tuff in Italy (e.g., tufo lionato).
nines. In addition, the clay for bricks, which became The critical role played by the supply of building
a fundamental building material during the Empire, materials to a project has recently been highlighted
was abundant in the Tiber and Aniene river valleys. by J. DeLaine, who emphasizes the importance of
The tall fir trees from the Apennines and southern the interplay between geology and topography in the
Italy and the rich forests of hardwoods, such as oak, extraction and transportation of the materials, which
elm, and chestnut, supplied the timbers for scaffold- in turn affects the cost and ultimately the choices
ing and centering and the fuel for lime and brick made by the builders.34 For example, in her analysis
kilns. The Tiber and its tributaries provided an ef- of the cost of the Baths of Caracalla, DeLaine notes
ficient means of transport for materials from inland that about a third of the cost results from building
areas as well as a connection to the port city of materials and their transport.35 The supply network
Ostia where imported materials arrived. The con- involved a variety of people at different levels of soci-
crete vaulted structures of imperial Rome are in part ety, all of whom stood to gain financially in the pro-
a result of a fortuitous geological environment rich in cess of supplying materials for imperial projects: the
natural resources. manual laborers who extracted the material, the prop-
An understanding of the geology of the volcanic erty owners from whose land it was extracted, and the
areas in Italy is important to the study of local build- carters and boatmen who delivered it. As a brief intro-
ing materials. The geological information on which duction, I present here the main building materials re-
archaeologists have typically relied has its roots in the ferred to in this study and provide an overview of the
12
P1: NDZ
INTRODUCTION
way in which they relate to the local trans-

portation networks.
Various types of tuff found locally
around Rome were used for the caementa
in vaulting (Pl. I). The most common type
is a reddish brown variety called tufo lionato
(traditionally called Aniene tufa), which
is a product of the Colli Albani district. It
was quarried extensively along the Aniene
river,36 which provided easy transporta-
tion into the city.37 Outcroppings also oc-
cur to the south of Rome, particularly in
the area known today as Monteverde on
the right bank of the Tiber.38 A less com-
mon type but evidently one more prized
for vaulting because of its light weight is
the yellow tuff known as tufo giallo della via
Tiberina, which is a product of the Sabatini
district. Ancient quarries of this tuff have
been found about 15 km to the north of
Rome along the Fosso di Grotta Oscura,
from which the traditional archaeologi-
cal name, Grotta Oscura tufa, is derived.
Other quarries also occur further to the
west at Fontana del Drago and to the north
at Pian dell’Olmo (Map 3, p. 14).39 The
tufo giallo della via Tiberina often has large
cinderlike scoria and pumice fragments
within the ash matrix, which makes it
map 2. Map of Italy showing the major volcanic districts and the provincial divi-
lighter (1,350 kg/m3 ) than the tufo lion- sions under Diocletian. Shaded circle indicates 100-mile jurisdiction of praefectus
ato (1,600 kg/m3 ). Another Sabatini tuff urbi.
called tufo rosso a scorie nere (traditionally Fidenae tufa) type of tuff usually found along the rims of craters.
also is found in some vaults. It is characterized by large The two types used in Rome were lapis Albanus and
pieces of black pumice and occurs along the Tiber lapis Gabinus, from the craters at Marino on Lago
near Prima Porta and near the ancient city of Fidenae di Albano and at Gabii, respectively. The quarries of
3 both types are located about 20 km from Rome, but
(Map 3 and Pl. I). It, too, was lighter (1,350 kg/m )
than the tufo lionato, because of the scoria in its matrix. those at Gabii are 5–6 km from the Aniene, whereas
Other volcanic stones sometimes used in vaulting the lapis Albanus quarries have no nearby river trans-
include peperino, a denser and heavier (2,250 kg/m ) 3 port available.40 Rarely, a very dense and heavy
13
P1: NDZ
map 3. Geological map of Rome and environs showing locations of major building materials used in
concrete vaulting. Deposits of materials not used by Roman builders are not shown.
14
P1: NDZ
INTRODUCTION
map 4. Map of Italy showing locations of major sources of timber and limestone.
15
P1: NDZ
type of leucititic lava (2,800 kg/m3 ), commonly the south and east of Rome (Map 3, p. 14).41 Major
called selce, was used in vaulting. It is extremely hard deposits of pozzolana rossa up to 10 m thick are located
and difficult to quarry and was used primarily for road near the basilica of San Paolo Fuori le Mura, from
building or as caementa in foundation walls. The main where it gets the name “pozzolana di San Paolo.”
quarries were from the Capo di Bove flow along the Pozzolana also was produced by the volcanoes on the
Via Appia, but small flows also occurred elsewhere Bay of Naples, but there is no evidence that it was
around the Colli Albani craters. Like the lapis Albanus, used in imperial Rome (see Chapter 3).
it would have been transported to the city by road. Travertine (2,450 kg/m3 ) is a sedimentary stone
Occasionally one finds very lightweight caementa that was often used in cut stone construction because
used for vaults in Rome, usually for large or struc- of its attractive creamy white color and its hard and
turally precarious ones. The most common type was a durable nature. Although not typically used in the
reddish to dark brown, vesicular scoria from Vesuvius vaulting itself, it sometimes played a role in the sup-
on the Bay of Naples. This material is often referred porting structure. It is found in great quantities near
to generically as pumice in the archaeological liter- Tivoli (ancient Tibur) and was therefore called by
ature, but it is somewhat coarser and heavier (750– the Romans lapis Tiburtinus. Anyone traveling out to
850 kg/m3 ) than true pumice (600–700 kg/m3 ). It Hadrian’s Villa has no doubt experienced the sul-
would have been shipped up the coast to Ostia and phuric odors of the hot springs of Bagni di Tivoli.
then transported upriver to Rome. This was one of These odiferous springs are the source of the nearby
the only nondecorative building materials imported travertine quarries, which were once part of a basin
from outside the immediate environs of Rome. Some in which the calcium carbonate in the water sup-
local pumices, varying in color from white to yellow plied by the hot springs precipitated creating an 80-m
to gray to black, were used as caementa. Most if not all thick stratum of travertine. These quarries were the
were products of the volcanic districts north of Rome main source of travertine for the ancient Romans and
and would have been transported down the Tiber. are still today among the top suppliers of travertine
The pozzolana used in the mortar is a local prod- worldwide.42
uct of the Colli Albani district. It comes in three Lime is the main ingredient of mortar and is de-
varieties: pozzolana rossa (red), pozzolana nera (black), rived from burning stones containing calcium car-
and pozzolanella (grayish) (Pl. III). Each belongs to bonate (CaCO3 ), usually limestone, travertine, or
a different volcanic event. Stratigraphically the red marble. Limestone was probably the major source
(1,600–1,900 kg/m3 ) is in the lowest layer, followed of lime for Rome during the first three centuries
by black (1,340 kg/m3 ), with the pozzolanella (1,360– a.d., but it was not local to the immediate envi-
1,670 kg/m3 ) on top (Pl. IV). The last was often rons. The nearest sources of limestone would have
quarried in open air and was probably the first to be been the mountain ranges to the northeast of Rome:
exploited. It was, however, an inferior product, and Monti Tibertini, Monti Cornicolani, and Monti
the Romans soon began to use the deposits of poz- Sabini reaching as far as Narni, and those to the
zolana rossa and nera by means of underground tunnel- southeast: Monti Prenestini and Monti Lepini reach-
ing. Access in antiquity was typically gained from the ing down to Terracina (Map 4), all of which lie within
side along valleys such as the Marrana della Caffarella, an 80-km radius of Rome.43 Lime is a processed ma-
Fosso di Tor Carbone, and Fosso delle Tre Fontane to terial. It first had to be fired, which required fuel, and
16
P1: NDZ
INTRODUCTION
then it had to be slaked. The processing resulted in a Rome also was surrounded by forests that provided
material that was more expensive than the pozzolana fuel for firing bricks and lime and for building the
and tuff.44 Pliny the Elder lamented that the chief wooden centering necessary to mold the concrete
reason for the collapse of buildings was from skimp- vaults. For the largest concrete vaulted structures, the
ing on lime,45 and indeed Faventinus, writing in the most prized wood would have been long timbers of
early fourth century a.d., made the point that lime fir, which was considered by both Vitruvius and Pliny
was the most expensive ingredient used in mortar.46 the Elder to be light, strong, and stiff.52 Fir was found
Brick (1,750 kg/m3 ) became an important material in the lowland areas on the west side of Apennines
for vaulting by the end of the first century a.d., and facing Campania and Etruria as far north as Pisa.53 It
the development of the brick industry had a great also was grown in southern Italy in the Sila forest and
effect on the vaulting techniques in Rome. The bricks on the island of Corsica (Map 4, p. 15).54 The avail-
were made in three basic sizes 23 RF (bessalis), 1 12 RF ability of large timbers seems to have declined during
(sesquipedalis), and 2 RF (bipedalis). The earliest brick late antiquity. In the fifth century a.d., Sidonius im-
wall facings in Rome began to appear during the plies that there had been too much timber taken from
late Republic and were made of roof tiles with the the Apennines,55 though R. Meiggs cautions against
flanges knocked off, as can be seen in the facing of placing too much emphasis on deforestation.56
the tomb of Caecilia Metella on the Via Appia.47 Transportation for building materials to Rome was
By the time of Claudius the roof tile manufacturers provided by the Tiber and its tributaries as well as by
were branching out into bricks, and wall facing made an extensive and well-built road system. The Tiber
of triangular bricks sawn from bessales or sesquipedales connected Rome to Ostia on the coast and ran in-
began to appear then.48 Within a century, the brick land as far north as Arezzo. Its tributaries, the Pallia
industry had become a highly developed organization and Clanis Rivers, serviced the area west of the Tiber
involving people from various levels of society. between the Lago di Bolsena and Lago Trasimeno,
The clay used for bricks and tiles was the old marine the Aniene area east to Tivoli, and the Nar and Tania
clay of the Pliocene era (2–13 million years ago) that areas east of the Tiber into Umbria. The points at
underlay the volcanic material of later periods, and it which the rivers and roads intersected often were ser-
was typically accessed along riverbeds that had eroded viced by river ports, as at Otricoli and Narni, to fa-
the more recent materials, laying bare the Pliocene cilitate transport of goods south to Rome.57 These
clay. Stamps on the bricks provide information both tributaries to the east of the Tiber also would have
on the general locations of the clay beds and on people been used for transport of lime and limestone from
involved in the industry (discussed later). Within the the Monti Sabini. The upper reaches of the Tiber
city, the Vatican and Trastevere were known for their above the confluence of the Tania and Clanis were
clays.49 Outside of the city, the clay beds tended to not always navigable, but Pliny the Elder noted that
be located along the Tiber and Aniene river valleys at a system was devised by which an ingenious series of
least as far as 70 km north of the city around present- dams collected water over a period of nine days after
day Bomarzo.50 Ongoing research into locations of which it was released to create a navigable waterway;
kiln sites and the mineralogical and chemical makeup otherwise, the upper Tiber was suitable only for logs
of the clays will hopefully yield further information and rafts.58 Strabo emphasized the importance of the
on the landholdings north of Rome.51 Tiber and its tributaries, the Nar, Tania, and Clanis,
17
P1: NDZ
for supplying timber to Rome during the Augustan the viability of building materials into Rome. The
period, and presumably much of this could simply be cura operum publicorum oversaw the upkeep of public
floated without the need for boats.59 The areas to the property,65 although whether it also was in charge of
south and east of Rome did not have the advantage the construction of new buildings is less clear.66 The
of waterways, but they were serviced by the roads ra- result of Augustus’s reorganization was the creation of
diating out from the city, the Appia, Latina, Labicana, an infrastructure for the supply and maintenance of
and Praenestina (Map 4, p. 15). the city, which provided a level of continuity and cen-
The method of transportation would have affected tralized control that had not existed previously, and
the costs of materials, which in turn could have had an this certainly would have aided in the organization of
effect on their use. Based on evidence from Diocle- labor and the supply of materials to the capital.
tian’s Price Edict, DeLaine has calculated the ratio of During the imperial period, the labor for both new
costs for transportation by means of sea:downstream: building projects and maintenance of existing struc-
upstream:oxcart as 1:3.9:7.7:42.60 Indeed, transport tures continued to be acquired through the letting out
of large timbers was a major factor in their availabil- of bids to private contractors as attested by Frontinus,
ity. In describing the timbers cut from the Sila forest writing from his perspective of water commissioner.67
in south Italy during the Augustan period, Diony- The use of redemptores on imperial building projects
sius of Halicarnassus says that the largest timbers are is also borne out in various funerary inscriptions in
cut as near as possible to the sea or river with tim- which the deceased identifies himself as a contractor
bers further away being cut into smaller pieces on for imperial or public works (see later). A common
site and then transported.61 One of the few lime kiln misconception is that the construction of large im-
complexes to have been excavated that may have sup- perial building projects such as the Colosseum or the
plied the city is located at Lucus Feroniae (c. 10 km imperial baths was made possible by large numbers
north of the Grotta Oscura tuff quarries) near a river of slave laborers taken from conquered territories.
port on the Tiber, which would have provided easy The implication of this assumption is that the gov-
access to the city.62 Likewise, the proximity of quar- ernment did not have to pay for labor other than the
ries and brickyards to river transport would have af- upkeep of the slaves. This assumption, however, has
fected transportation costs. been shown to be a simplistic view of the use of slave
labor in Rome. In 1980, P. Brunt argued that a sub-
stantial amount of nonslave labor was used for build-
the building industry in rome
ing projects in Rome,68 and more recent work on
During the Republic, public building was over- the building industry supports the idea that signifi-
seen by the aediles or the censors who let out cant numbers of the free populace in Rome found
bids for contracts to private contractors, redemptores.63 work on public building projects.69 These contractors
Augustus, as part of his urban renewal program, es- could have staffed their crews with both slave and
tablished commissions to oversee the care of the pub- nonslave labor, but regardless of the social status of
lic buildings (cura operum publicorum), the water sup- the worker, a majority of the crew would have been
ply (cura aquarum), the roads (cura viarum), and the skilled laborers as opposed to unskilled war captives
bed of the Tiber (cura alvei Tiberis).64 The care and used for hard labor.70 Some slave labor could have
maintenance of roads and the Tiber valley ensured been involved, but that labor would have come at a
18
P1: NDZ
INTRODUCTION
cost: both to the contractor who bought and sup- Building contractors often relied on a type of con-
ported (or else rented) the slaves and to the impe- tract called locatio conductio (lease and hire). A common
rial administration who hired the contractor and his type used for building projects was locatio conductio
crew. operis (lease and hire of units of work), in which the
Inscriptions bearing the names of contractors for locator (patron) lets out a job to be completed by the
public and imperial works also reveal that many of conductor (builder). The contract included a final in-
them were freedmen or descendants of freedmen spection (probatio) and an agreed-on price (merces). In
from wealthy senatorial families. One example dat- this type of contract, the builder took on responsibil-
ing from the late first century or early second century ity for the site until the final inspection of the work
is the funerary inscription of the imperial freedman (probatio),76 which released him of responsibility. He
[T. Clau]dius Aug. l. Onesimus, which states that could negotiate for either a task fee for the whole job
he was a contractor for imperial works ([rede]mptor or a task rate based on measured intervals. Another
operum Caesar(is)).71 Another self-identified redemptor, method of hiring was through a contract of locatio
Q. Haterius Tychicus, was a freedman of the power- conductio operarum, in which the locator (laborer) lets
ful senatorial Haterius family.72 He may be the same himself out to the conductor (patron) for a daily wage
person who was buried in the tomb of the Haterii. or piecemeal wage. In this case, the laborer took no
Unfortunately, the cognomen of the deceased in the responsibility for the site,77 but along with less respon-
tomb is not preserved to verify his association with sibility came less pay. A single project could combine
Tychicus. Nevertheless, the iconography of the reliefs various types of hires depending on the nature of the
in the tomb suggests that the deceased may have been job.78
involved in the building trade, as was another freed- The inscriptions on brick stamps provide a glimpse
man of the Haterius family, Q. Haterius Evagogus.73 into the working relationships between people of dif-
Freedmen were bound to their former masters ferent social status in the building industry. Steinby
through the Roman institution of clientela whereby has proposed that the stamps represent a contract of
persons of different social strata had certain obliga- locatio conductio operis, whereby the owner of clay beds
tions for each other’s welfare. In the case of freedmen, (dominus) contracted for the brickmaker (officinator)
however, this relationship was formalized by law, and to make a certain number of bricks that were then
it is often traceable in the epigraphic record through the property of the landowner to sell as he (or of-
the naming convention for freed slaves.74 The bond of ten she) pleased.79 The domini listed in the stamps
clientela also could have been a significant factor for the were typically of the senatorial class, and the officina-
advancement of some freeborn building contractors, tores were from the lower social classes and were often
albeit one that is not so evident from the epigraphic freedmen. Fewer than 19 percent of officinatores were
record. Such connections between the members of slaves.80 There was much money to be made for both
different social strata would have been beneficial to the upper and lower classes through the large im-
both parties, with the senatorial land owners provid- perial building projects,81 and cooperation between
ing contacts to the contractor bidding on large state them ensured that both benefited from the building
projects and the contractors acquiring materials such activity in the city during flush times.
as timber, pozzolana, or bricks produced on senatorial A number of redemptores, such as Q. Haterius Eva-
properties.75 gogus and Ti. Claudius Onesimus mentioned earlier,
19
P1: NDZ
are known to have been officials of the collegium fab- ings may have been related to their activities in the
rum tignariorum, which was an organization composed collegium, but it was not dependent on it.86 DeLaine
largely of builders. Unlike the medieval guilds, which has recently argued for a somewhat more active role
had political power and strict control over their crafts, for the collegium in organizing labor for large imperial
the collegia were primarily social organizations during projects.87 Along with kinship and clientela, the collegia
the first three centuries of the empire. In Rome, the undoubtedly formed another cog in the machinery
various collegia of craftsmen were not under the di- of the building industry and provided a means of ad-
rect control of the state, although one of the advan- vancement and a sense of achievement for the lower
tages of membership included some exemptions from segments of society.
public services as encouragement to practice crafts For a contractor interested in advancing in the pro-
that would benefit the state.82 The collegium of the fession, contracts of locatio conductio operis, in which he
fabri tignarii was the largest of the craft guilds attested took direct responsibility for his work and the build-
in Rome, and its organization dates from the late ing site, would have provided him the most control
Augustan period.83 A faber tignarius was strictly speak- and flexibility. The use of such a contract, however,
ing a carpenter, but numerous inscriptions indicate raises the question of his responsibility for the sound-
that membership was not limited to woodworkers, ness of the structure. The vaulting techniques dis-
and the collegium fabrum tignariorum seemed to have cussed in this study were often used to ensure the sta-
been open to builders of all kinds. Inscriptions listing bility of the building. For large projects, an architect
the members of the collegium in Rome in the late sec- usually was involved. So, who then decided when and
ond century reveal that the membership was as high how to use the various vaulting techniques intended
as 1,330.84 to ensure stability – the builder or the architect? The
Membership in the collegium fabrum tignariorum re- architect designed the building, but the builder put
quired entry fees and dues and was therefore a show the pieces in place. In the Digest of Justinian, the jurists
of some financial success. Of the known officers of dealing with private buildings are particularly con-
the collegium, many were freedmen and a number of cerned with the legal obligations of both the client
them were also Augustales, an honorary priesthood and the contractor in situations of building failure,
that required a certain amount of public munificence but the obligations of the architect do not seem to be
from the holder.85 These were not simply laborers of great concern.88 The redemptor was only respon-
but, rather, men of some means who were intent on sible for building failure until the final inspection. If
raising their status within the community. The fact the building failed a week or even a year after the
that Onesimus advertised himself as a contractor of inspection, it was the fault of the person in charge of
imperial works suggests that this fact in itself con- the probatio who was an agent of the patron,89 pos-
ferred some prestige. J. D. L. Pearse suggests that the sibly the architect. The architect was typically hired
individual members of the collegium fabrum tignariorum in a different manner from the contractor. He was
may well have been a primary source of contractors paid an honorarium for supplying technical skill and
for public works, but he is careful to point out that advice rather than manual labor. He was, therefore,
if this were the case, there is no evidence to suggest not directly responsible for the structure, though he
that the contracts were acquired through the collegium. could be sued for deliberate fraud, which included
The success of the individuals in their business deal- gross incompetence.90 The details of many difficult
20
P1: NDZ
INTRODUCTION
structural problems were probably worked out on- voluntary societies that conferred some prestige on
site with both architect and builder contributing their its members, became an obligatory requirement for
own expertise to the discussion. In the end, we must workers of a particular skill.96 Both membership and
assume that the most innovative buildings were the the movements of the members were strictly con-
result of the collaboration between a visionary patron trolled. Similarly, some professions were made hered-
and a creative architect working with experienced itary so that there was limited flexibility in adapting
builders. one’s work to one’s innate skills.97 The incentive to
By the beginning of the fourth century a.d., the use the building trade as a means of social and eco-
organization of the building industry in Rome had nomic mobility was thus removed as was the sense of
changed, largely in response to the political instabil- pride that came in membership of the various collegia
ity and economic crisis during the mid-third cen- of craftsmen.
tury. When Diocletian took over and established the
tetrarchy, he reorganized the provinces and instituted
the innovations
tax reform. Italy for the first time was included in
the taxed areas. A land tax (iugatio) was introduced, In the final chapter, I employ a framework based on
and the landowners paid their assessed taxes in kind, four criteria that have been used for identifying in-
depending on what they could produce.91 The new novation in agricultural technology: (1) accumulated
tax system became a means of requisitioning building knowledge, (2) evident need, (3) economic possibility, and
materials for the state. The evidence for the chronol- (4) cultural/social/political acceptability.98 So, for exam-
ogy of the implementation of the new tax scheme ple, a brief and simple application of these criteria to
in the various areas of the empire is incomplete, but the early development of concrete vaulting in central
Italy seems to have been divided into provinces by Italy yields the following: The development of the
a.d. 294.92 Within this scheme, the supply of building arch and the discovery of pozzolana-lime mortar con-
material to Rome was the responsibility of the prae- stituted the accumulated knowledge necessary for con-
fectus urbi, whose jurisdiction included areas within a crete vaulting to develop. The desire to create usable
one-hundred-mile radius of the city (Map 2, p. 13).93 flat terracing at hillside sites of religious sanctuaries
The new taxation system had a great effect on con- in central Italy is an example of evident need. The
struction in Rome by increasing state control of ma- wealth coming into Italy during the second century
terials and labor. B. Ward-Perkins traces the effects b.c. from conquests and taxation provided the economic
that the change had on public munificence in Italy possibility for building the increasingly grand sanctu-
and points out that the social mobility of the freed- aries. Cultural acceptability then developed from the
men, who had used the building trade as a means of desire to match the architectural accomplishments of
advancement earlier, was much reduced in the fourth conquered territories in the Hellenistic Greek world,
century.94 Many of the redemptores of earlier times in which hillside sanctuaries, such as those at Cos
were replaced with a system in which labor for build- and Lindos, had gained international repute. These
ing projects was requisitioned by the state as means four criteria also are applied throughout the follow-
of collecting taxes or through the collegia.95 Under ing chapters, but though in a less systematic manner
the new system, the collegia, which had once been than in the final chapter.
21
P1: KIC
2
CENTERING AND FORMWORK
R oman concrete vaults are praised for the

impressive distances they were able to span, but
one often forgets that the wooden structures on which
large wooden centerings capable of taking the weight
of the concrete. These wooden structures differed
from wooden roof structures in that they did not
the concrete was first laid is what determined the require clear spans and had to take a much greater
size of the final vaulted structure. Much of the tech- load with minimum deflection, but many of the join-
nology for building large concrete vaults was based ery techniques and the structural principles were no
on woodworking techniques; this aspect of concrete doubt the same. The basic structure for the center-
vaulting has not received much attention, in part, being of arches and vaults would have consisted of a
cause very little remains of these wooden structures. number of arch-shaped frames connected by form-
In this chapter, I examine the evidence that exists work boards (Fig. 8). The frames for the centering
for the wooden centering and formwork and pose could have been preassembled on the ground so that
the questions: How were the centerings constructed? the proper curve could be repeated using a template,
How were they lifted into place? How were they in which case some type of lifting device was neces-
supported? How were they removed without dam- sary to raise the preassembled pieces. For more com-
aging the work below? The sources used to answer plex vaults, the frames could have been constructed in
these questions include the impressions of the boards place, in which case the shapes had to be determined
left in the concrete, pieces of the actual wood (which in situ.1 This was a problem faced by Brunelleschi
are rare), ancient pictorial representations of wooden in the fifteenth century when he had to lay out the
structures (albeit not centering structures), literary deform of the octagonal dome of Santa Maria del Fiore
scriptions of wooden construction for bridges and in Florence,2 and no doubt it was one faced by many
siege towers, and comparisons of centering structures a Roman builder.
from later periods. The development of the triangular truss made pos-
sible the construction of large-scale wooden struc-
tures. A truss is a structure created of a minimum
assembling the centering
of three beams forming a rigid triangle (Fig. 9). It
The construction of the most impressive Roman had the advantage over a simple beam of spanning
vaults was dependent on the builder’s ability to erect great distances using a number of smaller timbers.
22
P1: KIC
In any complex structure involving joinery, the

strength of the joints plays as important a role as the
strength of the wood. Different types of joints are used
in response to different types of loads. The longest
member of a trussed structure was the tie beam, which
often had to be created by joining two shorter tim-
bers together end to end using various types of scarf
joints (Fig. 12). Tie beams of trusses are always in
tension, and an effective type of scarf joint used by
Roman carpenters in such situations was a splayed in-
dent scarf, “trait de Jupiter.” Roman carpenters were
using this type as early as the mid-first century b.c.,
as shown by the examples found joining two timbers
8. Reconstruction of wooden centering for concrete barrel vault.
Trusses were in use at least by the first century a.d.,

as shown by a relief found under the Palazzo della
Cancelleria (Fig. 10),3 and probably earlier.4 Vitru-
vius, writing in the first century b.c., described what
was probably a truss as a typical roof structure for
large spans.5 The depiction of Apollodorus’s famous
bridge over the Danube on Trajan’s Column demon-
strates a complex understanding of trussed construc-
tion in the early second century a.d. (Fig. 11).6 Other
large spanned structures known from Rome and else-
where, such as the Odeum of Agrippa at Athens
(c. 28-m span), the Diribitorium at Rome,7 the aula
regia of the Domus Flavia (c. 30-m span), the Basilica
Ulpia (26-m span), and the Constantinian basilica at
Trier (28-m span), surely used trussed construction
in their roofs. The Basilica Ulpia probably provided
the model for those trusses in the fourth-century
churches, San Pietro (24-m span) and San Paolo Fuori
le Mura (24-m span), of which depictions do survive
(Fig. 9).8 9. Wooden trusses.
23
P1: KIC
10. Marble panel found under the Palazzo della Cancelleria with a depiction of a wooden amphitheater
showing truss construction (first century a.d. [?]).
of elm in a shipwreck at La Madrague de Giens (60– many of which were replaced during the centuries,11
40 b.c.).9 Fontana’s drawing of the trusses of old San employed metal bands to create a composite tie beam
Pietro shows bands used to reinforce a scarf joint in as well as to attach the queen posts to the tie beam.
the tie beams (redrawn in Fig. 9), as was often recom- As early as the first century, Roman shipbuilders used
mended in eighteenth–nineteenth-century building iron bands to reinforce the scarf joints in the deck
manuals.10 The trusses at San Paolo Fuori le Mura, beams of the Caligulan ships found at Lago di Nemi.12
11. Detail of plaster cast from Trajan’s Column showing Apollodorus’s bridge over the Danube
(106–113 a.d.).
24
P1: KIC
12. Scarf joints used to join two timbers end to end. The two examples on the right are particularly
useful for tie beams for trusses because their configuration creates a “lock” that resists tension (after
Adam 1994: fig. 230).
Many of the joints used for centerings, unlike per- for its iron mines, by the late second century b.c.
manent wooden structures, had to allow for disassem- gave the Romans access to iron ores particularly
bly. Various types of pegs and bolts would have been high in magnesium, which was more easily con-
very useful for putting together centering that had to verted into steel through the absorption of car-
be taken apart. An example of an iron eyebolt (10 mm bon atoms (carburization).16 In the Augustan period,
dia, 85 mm long) has been found in a deposit of im- Strabo noted that iron was produced in Noricum, and
plements at a Roman fort at Newstead in Scotland Horace associated it with the highest quality swords.17
dating from the first/early second century. It has a Alhough the Roman smiths did not understand the
riveted washer (33 mm dia) at one end and an eye physics behind the principles of carburization and
(47 mm ext. dia, 30 mm int. dia) at the other, so that quenching (plunging hot iron into liquid) for harden-
connections could be made by slipping a straight piece ing, an analysis of iron tools found in Germany from
through the eye.13 In this way, disassembly could have the second or third century a.d. revealed that they
been facilitated. A rare example of a wrought iron had mastered both processes so that they could con-
threaded bolt nut (413 mm square × 160 mm high) sistently harden specifically chosen parts of a tool.18
dating to a.d. 180–260 was found at Niederbieber Such advances would have affected the efficiency of
in Germany,14 so the technology existed, though woodworking in general.
the regular use of connectors requiring such ad- In addition to the access to high-quality steel, the
vanced craftsmanship seems unlikely. In fact, Heron of development of the frame saw would have aided in
Alexandria recommended using lashing ropes rather the production of the long formwork boards that cov-
than nails and pegs when constructing lifting struc- ered the centering structure. Saws had been in use for
tures, such as cranes that had to carry great weights.15 centuries in Egypt and Greece, but the earliest depic-
By the first century a.d. and probably somewhat tions of the frame saws, which put tension on the
earlier, a number of developments had occurred in blade thereby allowing for cutting in both stroke di-
the production of iron tools, which would have aided rections, date from the early imperial period.19 The
in the efficient construction of wooden centering advent of the single-man frame saw and the two-man
and formwork. The establishment of friendly ties pit-saw would have substantially reduced the work
with Noricum (roughly modern Austria), famous time for the sawing of long boards. Figure 13 shows
25
P1: KIC
13. Frame saws. Drawing on the left is a detail from a wall painting at the Bottega del falagname at
Pompeii (VI.7.8–9) showing two workers using a pit-saw to cut planks. The two drawings on the right
show two different types of one-man frame saws depicted on imperial period reliefs in the Antiquario
Comunale at the Capitoline (top: inv. #2743).
an illustration from a painting at Pompeii of two men and coordination. Once the vaults were built, access
using a pit-saw and two examples of different types to the frames from above was no longer possible, and
of single-man frame saws depicted on stone reliefs.20 the centering had to be lowered or dismantled from
The pit-saw was a type of frame saw particularly use- below. If the vault was not very high, the centering
ful for cutting long boards. One man stood below frames could simply have been propped up with ver-
the board (hence “pit”) and the other stood above tical members that were tapped away once the vault
the board. The blade had teeth facing both directions gained its strength (Fig. 14), or the centering frames
so that it cut on both up and down strokes, thus in- could have been lowered with shoring and props.
creasing the efficiency. Another innovation related to For higher vaults such as those in the imperial baths,
saws was the setting of the teeth so that they were this may not have been practical. Alternative methods
bent outward alternately to one side and the other, could have included the use of ropes and pulleys at-
which created a wider groove for the saw blade and tached to the surrounding scaffolding to lower parts of
allowed the sawdust to escape, as noted by Pliny.21 the centering one at a time or even the construction
Saw-setting keys for bending the teeth were found of a tall, movable platform to aid in the process.23
at Pompeii.22 These advances in iron and wooden No ancient source gives a general rule of thumb
technology would have increased the efficiency of for how long the concrete should cure before the
building wooden centering so that larger and more centering was removed. Modern building manuals
complex wooden structures became easier and less suggest anywhere from four days to two weeks, but
time-consuming to build. these figures are complicated by the variety of dif-
ferent grades and types of modern concrete used.24
In determining when the formwork can be removed
removing the centering and formwork
from concrete, theorists agree that there is a relation-
The removal, or striking, of the centering once the ship between heat output and strength gain. Concrete
concrete was laid involved a great deal of planning that is curing puts out the most heat during the first
26
P1: KIC
14. Wooden centering used for a modern concrete barrel vault near Olympia, Greece (November
1994). Note the oblique supports that will ultimately be knocked away to lower the centering.
twenty-four hours and continues to generate some of the concrete did not cool too quickly and produce
heat for another three days. The rate of strengthen- cracks.25 The Romans surely did not understand the
ing is dependent on an adequate supply of water to modern theory for formwork strike time but, given
continue the chemical reactions within the concrete the empirical methods by which the Romans learned
(see further, Chapter 3). For very large vaults in an- to build, they were probably rather conservative and
tiquity, one can imagine that the period the form- allowed the concrete plenty of time to gain strength
work was left in place was somewhat closer to two before removing the centering.
weeks than to the four days it took the concrete In more recent history, a common technique for
to cool. easing the centering away from the vault before taking
When the centering and formwork was in place, it it apart has been to place a series of wedges under the
served a number of functions in addition to providing structure so that it could later be lowered by knock-
support. The formwork absorbed moisture from the ing out the wedges as shown in Piranesi’s engraving
freshly laid concrete and aided in the retention of of the construction of Blackfriar’s Bridge in London
the water, which is critical for the curing process of (Fig. 15).26 The amount of force necessary to move
the mortar (see Chapter 3). Striking the formwork the wedge is proportional to the load acting on it;
too early could allow for excessive evaporation that therefore, placing many wedges as high as possible in
could result in a weaker structure. The formwork also the structure is advantageous when the time comes to
acted as a type of insulation so that the outer surface ease the centering so that the weight bearing down
27
P1: KIC
15. Detail of engraving of Blackfriar’s Bridge showing wedges used to lower the centering (G. B.
Piranesi 1764).
on them is minimized. The more wedges used, the could have been used to deal with this problem.
less force required to dislodge each one.27 An alterna- Modern builders use mineral or vegetable oils (as well
tive method was to set the frames on wooden blocks as other substances not available to the Romans) to
in which a recess had been carved on the bottom coat the formwork and act as a release agent. Wax is
center portion. When the time came to lower the another substance used today that the Romans could
centering, the builders would saw the block apart on have used, although modern builders usually consider
either side of the recess so that the central portion wax to be pretreatment that helps to preserve the
dropped down (Fig. 16).28 This method would have formwork and to prevent the releasing agent from
required that the block be located some distance from soaking into the wood.30 The preservation of the
the supporting wall so that there was room to move formwork for reuse is an important economic factor
the saw back and forth. Unfortunately, there is little in today’s concrete industry and probably was for the
direct evidence from ancient times to verify how the Roman builder as well. Postantique builders some-
Roman builders accomplished the task. times used reed mats as a barrier between the wood
Easing the formwork away from the vault would and the lime mortar, as can be seen from the im-
have often been complicated by the wooden form- pressions on some of the reconstructed vaults at the
work boards sticking to the hardened mortar, as Tabularium and the Emissarium at Albano.31
exemplified by a piece of chestnut formwork still em- Removing the centering was a dangerous process
bedded in the pavilion vault (2.80 × 3.90 m) of one of and had to be carefully coordinated and executed.
the towers of the Aurelian Walls.29 Various methods The concern for the striking process is apparent in the
28
P1: KIC
sesquipedales with a grid of bessales covering their joints

(Fig. 17). When the wooden centering and form-
work were removed the brick linings remained ad-
hered to the intrados of the vault, although the larger
bricks have often fallen or been removed, leaving vis-
ible only the grid of bessales. By using the brick lin-
ings as a barrier between the formwork boards and
the mortar of the concrete, the boards would no
longer have become stuck to the concrete as it set.34
Traditional explanations of these linings have as-
serted that they were used to save time or money.35
In the nineteenth century, A. Choisy suggested that
the brick linings were used to reduce the amount
of wooden formwork needed for the centering of
the vaults, thereby reducing the cost. He proposed a
model in which a layer of bipedales is placed over a
16. Diagram of method of lowering the centering of vaults
(modified by author from Fitchen 1961: fig. 12). light wooden centering. Another layer of bricks was
then bound to the first layer with mortar creating a
writings of the ninth-century author of the Diegesis rigid skin on which the concrete could be laid; in
who, in describing the construction of the Hagia
Sophia in Constantinople, noted that the disregard in
lowering the centering caused damage to the founda-
tions of the original Justinianic church. He claims that
during the removal of the scaffolding of the second
dome, the church was filled with water to a depth
of five cubits to prevent damage to the floor and
foundations.32 The story is clearly apocryphal, but
it does represent a concern for the problems encoun-
tered in striking the centering for vaults.
the use of brick linings on the intrados

of vaults
In the early second century, a technique appears in
Rome that would have solved the problem of the
formwork adhering to the concrete (Appendix 2d).33
A lining of bricks was placed over the wooden form-
work before the concrete of the vault was laid. The
linings usually consisted of a layer of bipedales or 17. Reconstruction of centering for barrel vault using brick
linings of bipedales and bessales.
29
P1: KIC
this way, the centering could be moved when the

concrete had only been laid halfway up the vault and
reused for another vault while the first was still under
construction.36 Many of the linings, however, do not
have two solid layers of bricks, and it is doubtful that
the linings could ever have functioned as Choisy de-
scribed. His proposal was influenced by his familiarity
with a technique used in Spain and southern France in
the eighteenth and nineteenth centuries called “tim-
brel vaulting” that allowed vaults to be built with very
minimal centering,37 but it is quite different from the
Roman examples of brick linings.
G. Cozzo pointed out the problems inherent in
Choisy’s solution and argued that the linings were
used as a backing for stuccowork, as he believed
18. Case a giardino, Ostia (III.9.22–3) (c. a.d. 128). Detail of
the stucco adhered better to the brick than to the
vault built using brick linings. Arrows indicate where a thin layer
concrete.38 However, the locations in which the brick of lime paste was first spread over the bricks before the mortar
linings occur, such as on the vaults of praefurnia and and caementa were placed on top.
small stairs, do not support the idea that they were
intended for places with heavy stucco. Moreover,
Vitruvius’s statement that tiles reject the initial layer alone, many of which would have been used any-
of plaster because of the dryness from being fired in way to support the bricks.40 The most obvious ben-
a kiln contradicts Cozzo’s proposal. Vitruvius notes efit of the linings is that they helped to preserve the
that a layer of lime must be applied to bond the plas- formwork and facilitated the easing process, and in
ter to the tile, and in the same passage he refers to the course of doing this they also may have saved
slaked lime as being like “glue.”39 A similar use of the builders both time and money, thereby increasing
lime paste appears to have been employed to bind the efficiency.
bricks to the concrete core in some examples of brick By the Hadrianic period, many of the vaults with
linings (Fig. 18). brick linings also have upright bricks set into the
As suggested by Choisy, the use of whole bipedales mortar usually in correspondence to the edges of
or sesquipedales could have reduced the necessary the bessales of the inner layer. Typically these upright
formwork somewhat, but often broken bricks were bricks run in horizontal rows at the joints between the
used on small and irregularly shaped vaults and would bessales, but occasionally they also are set vertically, as
have required continuous planking for support. in the cross vault of the Large Baths at Hadrian’s Villa.
Therefore, the decision to employ this technique was In cases in which they do not run in rows, they still oc-
not governed exclusively by the desire to reduce the cur in relation to the bessales, usually just above them,
amount of wood. The bricks could not have been as can be seen at the Heliocaminus Baths at Hadrian’s
reused, and as a result, using the brick was proba- Villa (Fig. 19). This example is somewhat unusual in
bly more expensive than using the wooden planks that it also has provision for T-shaped metal bars,41
30
P1: KIC
19. Heliocaminus Baths at Hadrian’s Villa (a.d. 125–133). View of semidome showing the upright bessales
set in correspondence to the flat bessales, some of which are still visible. Inset (lower left) shows detail of
hole where the impression of a T-shaped bar that held a suspended ceiling can be seen. The bars were
formed by creating a circular hole (like an eye bolt) at one end and inserting a cross piece through it
to create the “T,” which was then embedded in the concrete to provide purchase. The short vertical
impression was made by the “eye” and the longer horizontal impression was the cross piece that ran
through the “eye.” On the right is a reconstruction of the vault.
the imprints of which are still visible, that once held The pattern of use of the brick linings over time
a suspended ceiling, but in other examples as well appears to be related to the fluctuations of the brick
there is a strict relationship between the placing of industry. The linings first appeared in the Trajanic
the bessales and the uprights. In all cases, the uprights period at Trajan’s Markets and the Trajanic latrine at
seem to have been used to aid in the laying of the the Forum of Caesar when the brick industry was at its
caementa on vaults with brick linings. peak in terms of the number of producers listed on the
31
P1: KIC
20. Pons Fabricius (62 b.c.). View of the intrados showing the holes used to support the wooden
centering. The bottom hole has a slanting upper surface indicating that it held a diagonal brace for the
beam in the upper hole.
brick stamps. They then disappeared after the reign of the vault itself. Different vault forms would have
of Caracalla when brick stamps with text ceased to each had their own logistical problems. In the fol-
appear. When the stamps reappeared on bricks under lowing examples, the evidence for the arrangement
Diocletian, the technique of using brick linings was of centering structures is discussed for the three most
not reinstituted. The brick industry seems to have basic types of Roman vaults: the barrel vault, the cross
undergone a rebirth in the late third century but, vault, and the dome.
even so, reused bricks are commonly found in the wall
facings of the period, and evidently such a profligate
Barrel Vaults
use of whole bricks for the linings on vaults was not
an option at this time. Depending on the span and height of the vault, the
centering could have been supported from ground
level or from corbels or holes higher up in the struc-
evidence for centering arrangements
ture. Holes and corbels often were used for bridges.
The evidence for centering arrangements can be de- The two bridges leading to Tiber Island both have re-
duced from the formwork imprints left in the vault mains of their centering supports. The Pons Fabricius
(Appendices 2a–2c),42 holes or corbels built into has a combination of two holes, one above the other
walls as support for the frames, and even the form with a cutting between them that suggests the use of a
32
P1: KIC
diagonal strut (Fig. 20). The Pons Cestius, by con-

trast, employs travertine corbels, though these
probably belong to the fourth-century a.d. recon-
struction and, therefore, represent a much later
practice (Fig. 21). All the original corbels have on
the underside the remains of two grooves cut so that
they are deeper towards the outer end of the block.
G. B. Piranesi reconstructed the recesses as lodgings
for a diagonal brace for the centering structure, but
he showed the corbels as each containing one single
recess rather than two grooves (Fig. 22). At both
bridges, the centering frames would have been at
least partially disassembled for removal. A. Leger’s
reconstruction of the centering in relation to the
corbels at the aqueduct bridge, the Pont du Gard
(early first century a.d.), provides an idea of how
such corbels would have been used with diagonal
braces (Fig. 23).
The bridges were both built of cut stone as opposed
to concrete, but evidence survives in concrete vaulted
21. Pons Cestius (fourth century a.d. reconstruction [?]). View structures to indicate that similar methods involving
of the underside of the travertine corbels with cuttings.
22. Pons Cestius. Details of engravings by G. B. Piranesi (from Le antichità romane IV [1748: left pl. XXIII,
right pl. XXIV]). Circles added by the author to indicate Piranesi’s interpretations of the cuttings shown
in previous figure.
33
P1: KIC
The centering for barrel vaults was particularly sus-

ceptible to deformation from the weight that built
up against the haunches as the vault was being con-
structed. As the material (whether it be concrete or
cut stone) was laid onto the centering, its weight be-
gan to push the sides of the wooden structure inward,
which in turn forced the crown upward (Fig. 25). The
author of one nineteenth-century building manual
noted this tendency and recommended putting rocks
on the crown to keep it from rising.45 This phe-
23. Reconstruction of centering frames in relation to the corbels nomenon may account for the irregular form seen
on the Pont du Gard outside of Nimes, France (first century a.d.). in the barrel vaults (24.5-m span) of the Basilica of
Leger 1875: pl. II.
Maxentius (Fig. 26). These are among the largest con-
crete barrel vaults in ancient Rome. Earlier vaults
corbels were used. In a Flavian structure now under with similar spans were typically domes and cross
the Church of San Clemente, a series of square holes vaults, which behaved somewhat differently than bar-
was built into the upper part of the walls that sup- rel vaults. The dome had the advantage of double cur-
ported the concrete barrel vaults. F. Guidobaldi in his vature, so as long as the concrete was laid evenly as
report on the findings under San Clemente suggests work progressed upward the pressure on the centering
that the holes were used to support the beams of a was distributed circumferentially and the problems of
mezzanine level, but I would argue otherwise.43 In deformation were minimized. For cross vaults, the
most of the surviving rooms the impressions of the area at the haunch was minimized, so the weight in-
butt joints of the wooden formwork align above the crease began to accumulate higher up in the structure.
central hole, suggesting that they were used to sup- The builders of the Basilica of Maxentius may have
port the centering with formwork about 10 RF long run into one problem of pushing the limits of barrel
(Fig. 24). A similar situation has been documented vaulting: The increased surface area at the haunch of
by K. de Fine Licht at Sette Sale, the cistern sup- a very large barrel vault put unusually great pressure
plying Trajan’s Baths (see Appendix 1.13).44 Other on the centering frames early in the construction pro-
vaulted buildings in Rome employed permanent sup- cess. Whatever the cause of the deformation seen in
ports in the form of travertine or marble corbels. The Figure 26, it seems to have had an effect on the laying
clearest example occurs in a barrel vault along the of the brick arches along the façade as well (Fig. 27).
front of the Nymphaeum Alexandri located in Piazza
Vittorio Emmanuele. In this case, the corbels did not
Cross Vaults
support a mezzanine floor and could only have been
used for the construction process. As discussed fur- The formwork for a cross vault was much more
ther in Chapter 5, these corbels aligned with a type complicated than that for a barrel vault, which might
of brick ribbing in the vault. The correspondence account for the rarity of preserved examples before
between them suggests that both corbels and ribbing the end of the first century a.d. The earliest surviving
were related to the centering frames. cross vaults occur in some of the rooms facing onto
34
P1: KIC
24. Barrel vaulted room under San Clemente (late first century a.d.). Upper image shows inset of
photo of wall and vault. Arrows indicate the impressions of butt joints of centering boards over the
holes in the sidewalls. Lower image show author’s reconstruction of the way the holes could have been
used to support centering during the construction of the vaults.
the octagonal room of the Domus Aurea, although for a cross vault because of the way the frames intersect
they were probably used earlier in the Augustan (Fig. 28). For a pavilion vault, the frames would be
period at the portico in front of the Basilica Aemilia placed along the diagonals and intersect at the crown,
(c. 14 b.c.–4 a.d.). The earliest type of intersecting and then formwork would span between the frames
vault built by the Romans was the pavilion vault, as from corner to corner,46 whereas for a cross vault,
seen at the Tabularium (78–65 b.c.). The centering the frames would have been oriented orthogonally
for a pavilion vault was much easier to construct than along the major axes of the intersecting barrel vaults
35
P1: KIC
so that they could be laid out as a portion of a circle.

Some evidence for this arrangement is preserved in
the continuous series of cross vaults at the Colos-
seum. In places where the formwork imprints are
still visible, the butt joints of boards are located above
the piers, which suggests that there was a support of
some sort underneath at that point (Fig. 29). One
also can see in Figure 29 that the lowest part of the
vault actually consisted of a travertine block cut to
the form of the curve. This would have allowed the
formwork for the concrete to begin higher up so
that there was room to lower it when the time came
to remove it.
In building cross vaults, the size of the vaults and the
support system for the frames would have been major
25. Diagram showing common deformation pattern for center-
criteria in deciding how to lay out the centering. For
ing frames during construction. As the load increases along the
haunch, it pushes down on the frames causing them to rise at the cross vaults built in a series, the centering would have
crown. been laid out so that a major barrel vault ran along
26. Basilica of Maxentius (a.d. 306–312). View of the easternmost barrel vault showing the curve of
the vault, which deviates from the geometric ideal.
36
P1: KIC
the main axis and secondary barrel vaults were built vaults grew in scale. Proposals by both J. DeLaine and
up to it. This was surely the case for the central hall R. Taylor assume that the centering frames for the
(8.5-m span) of the Aula at Trajan’s Markets (Fig. 30). transverse barrel vaults were built first and then “sad-
The transverse vaults are not consistent in span from dles” were built between them to support the main
one side of the central hall to the other, so they clearly barrel vault (Fig. 31).47 Both proposals are opposite
were not continuous from one side to the other and from the scheme proposed above for the cross vaults at
must have been built up against the major barrel vault. Trajan’s Markets where the longitudinal vault is built
The centering frames of the major longitudinal vault first but for good reason: As the span of the vault in-
would have rested on the travertine corbels, whereas creased, the factors affecting the centering changed.
the centering for the smaller transverse vaults sprang In the case of the frigidarium, if the main barrel vault
from a higher level so that the crowns of all vaults were built first the centering frames would have to
were roughly even. span the full 22 m if they were to be supported on
As the size of the vault increased, the criteria for the the projecting columns as proposed. This would have
arrangement of the centering changed. A comparison required a large truss but not as large as some ear-
of the Aula scheme with centering reconstructions lier ones. More critical would have been the 17-m
for the frigidarium of the Baths of Caracalla provides distance that the formwork had to span between the
some insight into the problems encountered when the frames. By contrast, building the transverse vaults first
and supporting them on trussed frames spanning be-
tween the columns would have allowed for the trans-
verse frames to be set up much closer so that shorter
formwork boards could be used.
Another factor that would have come into play
for very large centering frames was the weight of
the wood used to build them. DeLaine uses a fig-
ure of 17 tonnes for the centering frames specified for
the construction of a seventeenth-century barrel vault
(c. 24-m span) at San Pietro. This would have been
just beyond the proposed capacities for two of the typ-
ical two-armed Roman cranes (6–8 tonnes each).48
The wood specified for the San Pietro truss, however,
was chestnut (475 kg/m3 seasoned). If fir (430 kg/m3
seasoned) were used for a similar sized structure, the
weight would decrease by 10 percent or more and
bring the frame from 8.5 tonnes per crane down into
the appropriate range of about 7.5 tonnes per crane.
The way that the centering was constructed for
27. Basilica of Maxentius (a.d. 306–312). Detail of outer face of the 8.5-m span cross vaults at Trajan’s Markets is
the easternmost barrel vault showing a disturbance in the pattern not necessarily evidence for the way the 22-m span
of the brick voussoirs, which may indicate that the deforma-
tion shown in the previous figure occurred during construction cross vaults of the imperial thermae were constructed.
because of deformation of the wooden centering.
37
P1: KIC
28. Reconstructions showing the difference in the centering and formwork of a pavilion vault and a
cross vault.
29. Colosseum (post-a.d. 217). Detail of cross vault from level 2. Arrow indicates the impression of a
butt joint between two formwork boards. The location in the center of the supporting pier suggests
that the centering frame was located in this spot and probably was supported on the projecting cornice.
Note the block of travertine forming the lower curved part of the vault. This would have allowed the
centering for the concrete portion to start at a level above the cornice so that it could be lowered later.
38
P1: KIC
30. Aula at Trajan’s Markets (a.d. 106–113). Reconstruction of the centering scheme.
31. Comparison of centering schemes (same scale) at the Aula at Trajan’s Markets (a.d. 106–113) and
the frigidarium at the Baths of Caracalla (a.d. 212–216).
39
P1: KIC
As the scale changed, the factors that governed the

design also changed. This comparison exemplifies the
types of decisions with which the builders were faced
in designing the wooden structures for large vaults.
Domes and Semidomes
The construction of the centering for domes and

semidomes is more complicated than that of most
other types of vaults because of the double curvature
that must be approximated using straight planks of
wooden formwork. The formwork could have been
arranged in two ways: horizontally (parallel to the cir-
cumferential lines) or radially (along the meridians)
(Fig. 32). When the formwork was arranged hori-
zontally, the ends of the planks usually aligned along
meridional lines of the dome and were supported by
radiating centering frames placed at the ends of the
boards. Because each horizontal plank had to be rela-
tively short in order to approximate the curve, this ar-
rangement typically required many centering frames
converging at the center of the circle defining the
dome. When formwork planks were arranged radi-
ally, they usually occurred in tiers, and the ends of
each tier of planks fell along a single circumferen-
tial ring, which in turn had to be supported in some
way. This support ring could be carried by fewer ra-
dial frames than would be necessary to support the
ends of the short horizontal planks. The earliest doc-
umented example of radially arranged formwork for
a dome occurs at the caldarium of the Large Baths at
Hadrian’s Villa.49 Other examples occur in domes and
32. Diagram of dome showing the difference between horizon-
semidomes dating from the second to the fourth cen- tally arranged formwork boards (top) and radially arranged boards
turies, such as the Tor de’Schiavi and the Octagonal (bottom).
Hall of the “Villa of the Gordians.”50 Unfortunately,
little evidence for formwork exists from the first and The earliest preserved concrete dome is that of the
early second centuries to give an indication of the “Temple of Mercury” at Baiae (Augustan, 21.6-m
typical formwork arrangement for large vaults with dia),51 and a recent photogrammetrical study of its
double curvature. intrados reveals significant deformations from the
40
P1: KIC
33. “Temple of Mercury” at Baiae (late first century b.c.). Reconstruction of the centering. The pattern
of deformation documented by F. Rakob is shown in exaggerated form by the dashed amoeba shape,
and the support locations proposed by both Rakob and J. J. Rasch are shown in the plan view.
41
P1: KIC
arc defining its ideal form, which suggests that differently in relation to the deformation pattern and
the builders encountered difficulties in constructing interprets the deformations as a mistake in laying out
the centering for such a large structure. The vari- the frames and as subsequent sagging between the
ations in form occur in six unequally spaced sec- frames because of the lack of a central support tower
tions with a maximum displacement from the ideal (although how the frames were otherwise supported
arc of 22 cm (Fig. 33). F. Rakob, who undertook is left unclear).53 Unfortunately, the imprints of the
the study, believes that the deformations were caused formwork do not survive to verify the validity of ei-
by mistakes in laying out both the shape and the ther proposal. Regardless of the reasons for the defor-
position of the centering frames. He reconstructed mations, which could have been because of problems
the centering with eight radiating frames supported in laying out the curves, in laying out the frames, or
by a central tower under the oculus and scaffolding in providing adequate support, the builders had not
along the outer wall. The radiating frames would yet mastered difficulties of constructing large domes.
have been connected by horizontal members that The first issue to be addressed was the layout and
supported the ends of radial formwork. I show a construction of the wooden support system.
modified version of his proposal in Figure 33.52 J. J. The earliest surviving attempt at constructing a
Rasch, who has made photogrammetrical studies of dome in Rome itself appears in the octagonal room
late Roman domes, proposes an alternative explana- in the Esquiline Wing of the Domus Aurea. The vault
tion for the deformations that Rakob found in the is not a true dome, but it is one of the few central-
Mercury dome. He also suggests a scheme with eight ized vaults preserved between the construction of the
radiating frames, but he arranges the frames somewhat Mercury dome at Baiae and the Pantheon dome. It
34. Domus Aurea (a.d. 64–68). View of octagonal room showing the formwork impressions remain-
ing in the vault. Arrows indicate butt joints between formwork boards. Dashed line indicates where
geometry of formwork changed. Impression of clamp (?) is visible directly above central butt joint.
42
P1: KIC
also preserves the imprints of its formwork and pro-

vides some evidence for the way the centering would
have been arranged. On seven facets of the eight-
sided vault the formwork boards span from one cor-
ner to the next with no break (max. 5.79 m), but,
on the southeast facet, the impressions of a butt joint
can be seen six boards from the impost approximately
2.15 m from the left corner (Fig. 34). The uppermost
board was apparently attached to the board above it
by means of a clamp on the outside of the wooden
structure, which has left its impression in the con-
crete. Presumably, the ends of the boards were resting
on some sort of intermediary support at this point.
The clamp impression and the butt joints do not oc-
cur at the center of the panel, which suggests that at
least two intermediate supports were used along the
straight sides of the vault.
An eight-sided vault had the advantage of having
straight sides so that constructing a form with double
curvature was avoided, but there was still the problem
of merging the octagon into a circle at the oculus.
The formwork imprints may provide some clue as
to how this was accomplished. The butt joints along
the groin are visible up to a certain point. They then
stop and the pattern of the boards changes (Fig. 34).
The imprints become less clear, but the geometry of
the vault appears to change so that a triangular area
merges the eight-sided figure into a sixteen-sided one,
which more closely approximates a circle (Fig. 35).
A corresponding sixteen-part division of the oculus
roughly coincides with location of the butt joints and
clamp. Guglielmo De Angelis d’Ossat has noted that
the profile of the intrados of the vault corresponds to a 35. Domus Aurea (a.d. 64–68). Hypothetical reconstruction of
geometry of centering for octagonal room.
circular arc only along the groins,54 which implies that
the main forms were at the corners and that the frames
in between were secondary supports. This example By the time the Pantheon was built some sixty years
provides some idea of the way the builders dealt with later, the construction of centering and formwork for
the complicated geometrical issues encountered in domes had clearly been mastered, but there is little di-
laying out the formwork. rect evidence for the details. Various proposals, some
43
P1: KIC
merely suggestive and others quite detailed, have been

made.55 My intention here is not to add to the ex-
isting hypotheses but, rather, to ask how one would
go about distinguishing “possibility” from “proba-
bility” and to comment on some of the factors that
make the Pantheon centering such an extraordinary
achievement.
One of the perennial questions regarding the cen-
tering of the Pantheon dome is whether it was sup-
ported from the ground and if so to what degree. At
one extreme is the idea of the centering consisting
of a virtual forest of supports and, on the other, is
the idea of the most minimal “hanging” centering
propped against a ring at the oculus with no ground
support at all. A third possibility is one that falls some-
36. Pantheon (a.d. 118–128). Geometry of dome based on a
where between the two extremes and employs a cen- sphere within a cylinder.
tral tower for support under the oculus. In evaluating
the proposals, the first question to ask is whether the had to be supported in some way at the oculus. The
maximum distance spanned in any of them is within two possibilities are that a central tower was used as
the known capacity of Roman trusses. In either, the support against which the frames could rest or that
central tower proposal or the central ring proposal, the radiating centering frames all rested against a cen-
the centering frames themselves would have been ar- tral ring at the oculus and balanced each other. Rasch
ranged radially in some fashion in order to determine has argued that the idea of a large central tower for
the appropriate form, so the maximum distance any dome centering was developed in the second century
frame would have had to span was the distance from for the Pantheon by Apollodorus, who based the de-
the spring of the dome to the edge of the cornice of sign on the siege towers described in his treatise on
the oculus. This comes to about 26 m, which is ap- siege warfare.56 However, Apollodorus’s siege towers
proximately the same as the span of the Basilica Ulpia were no doubt modeled on earlier ones developed by
trusses (Fig. 36). This correspondence is unlikely to Hellenistic Greek engineers. Vitruvius describes one
have been casual, given the close connection in time designed by Diades, an engineer of Alexander the
and techniques between the two projects. The known Great, that was 120 cubits high (53 m) by 23.5 cubits
and tested precedents in wooden construction proba- wide (10.4 m) at the base, which would have been
bly governed to some degree the ultimate size chosen more than large enough to fit under the dome of the
for the dome. Thus, the solution based on the for- Pantheon (Fig. 36).57 So, by the second century, the
est of support structures is unnecessarily conservative idea of a central tower for dome centering was one
given the technology of the period. that may well have had a trial run in the previous cen-
If continuous support from the ground was not tury, and the scale had been achieved much earlier,
necessary, the next question is whether the frames albeit in a military context. As seen earlier, Rakob
were supported at all from the ground level. They proposed that a tower was already used at the dome
44
P1: KIC
proposed a variation on it. He suggests that twenty-

eight radiating frames were installed in two phases and
were positioned by means of twenty-eight cranes set
up on top of the rotunda wall, each holding a frame in
place until the horizontal struts bound them all into
place.59
Of the two options – central tower versus cen-
tral ring – I would suggest that the tower scheme
is the more probable. The construction of such a
tower has precedents in military technology as men-
tioned earlier, and in the context of construction sim-
ilar types of towers were used for lifting very heavy
weights by the mid-first century a.d., as described
by Heron of Alexandria.60 The tower would have
provided a means of raising and lowering materials
for the concrete as well as materials for building the
wooden structure itself. Additionally, it could have
been equipped with ladders or even elevators for
workers, such as those used at spectacles for men, sets,
and animals.61 Moreover, it simply would have been
safer. The schemes involving no central tower seem to
me more complicated than necessary – in short, possi-
ble but not probable. The dome of the Pantheon was
twice as large as any earlier known dome, and the
37. Pantheon (a.d. 118–128). Plan (below) and section (above) most likely scenario is that they would have aimed
showing reconstruction of centering from Viollet-le-Duc 1868: toward a balance between economy and a degree of
p. 473, fig. 4. “L” marks where ring at oculus is located.
safe conservatism using devices such as the tower and
truss, which had been previously tested.
of the “Temple of Mercury” at Baiae, and something The use of coffers in the Pantheon dome would
similar was likely used for the octagonal room at the have complicated the construction and particularly
Domus Aurea. the decentering process. Centering and formwork
The alternative to the central tower is a scheme in had to be taken apart from below, either directly off
which the radiating frames are propped against a cen- the hardened vault or after being eased. In the case
tral ring at the oculus, the principle being that once of coffers, not only did they have to be indepen-
all the frames were in place they were connected with dent from the centering structure but also the steps
horizontal struts to create a whole that was in equilib- within each coffer form had to allow for removal by
rium. In the nineteenth century, M. Viollet-le-Duc containing no acute angles that would cause them to
proposed such a scheme, although he provided very get stuck in the concrete. In the case of the Pantheon,
few details (Fig. 37).58 More recently, R. Taylor has the centering, formwork, and coffers comprised such
45
P1: KIC
a complex network that the centering frames were number but, seen in context, the number takes on
likely constructed in such a way that they could be added significance.
disassembled and removed in parts. Dowels or eye- The Pantheon dome was a singular experiment in
bolts could have been used for this purpose. Because the ancient world. No other dome in Rome or in
of the angles of the coffer boxes, they would have the Roman world ever approached its scale or the
been removed individually; they could not have been simplicity of its conception. The complexity of its
eased all together because the edges of the boxes are construction, by contrast, may be one explanation
at different angles and could not have slid straight of this singularity. It displays a variety of advanced
down.62 construction techniques and for that reason appears
The unusual number of twenty-eight coffers in many times during the course of this study, but the
each of the five concentric rows presented an added aspect of its construction that involved the greatest
difficulty in laying out the centering. The governing daring, ingenuity, and logistical coordination was the
order of the Pantheon’s structural system was based temporary wooden structure on which the concrete
on a sixteen-part geometry, which was very easy to of the dome was laid and of which no traces remain.
lay out for a circular form by simply using a com- The Hadrianic period also produced a unique
pass and straightedge to divide the circle first into series of segmental vaults, in which a dome or
quarters, then eights, and finally sixteenths. Divid- semidome was divided radially into a series of wedges,
ing a circle into twenty-eight evenly spaced parts which were either all convex or else alternated be-
is not possible with a straightedge and compass.63 tween flat and convex (Fig. 38). The earliest examples
As Wilson Jones has recently pointed out, twenty- are the latter type and occur at Hadrian’s Villa at the
eight was a special number, being one of only four vestibule of the Piazza d’Oro and the semidome of
numbers known in antiquity for which the sum of the Serapeum (Fig. 39). The centering for such seg-
the factors equals the number.64 It also represented mental vaults would have required formwork boards
the lunar calendar, thus contributing to the celes- that were arranged radially rather than horizontally.
tial symbolism of the building. One has to be care- The fact that the earliest example of radially arranged
ful in assigning such significance in hindsight,65 but formwork in a dome also occurs at Hadrian’s Villa
the Pantheon is the one building in antiquity for in the caldarium of the Large Baths is probably not
which one can safely attribute some geometrical and coincidence. The construction of one would have in-
numerical significance to the form of the building. fluenced the other. F. Brown first suggested the con-
The design is modeled on a sphere within a cylinder nection between the vaults and Apollodorus’s famous
(Fig. 36), which was also the device that Archimedes retort to Hadrian reported by Dio Cassius: “Be off
had carved on his tomb as representing one of his and draw your pumpkins ().”67 He sug-
greatest mathematical discoveries – his theorem de- gested that the pumpkins of the quote were actually
termining the 2:3 relationship between the volume referring to these segmental vaults. In fact, most pre-
of the sphere and the cylinder.66 Whoever designed served examples belong to the Hadrianic period. Ex-
the Pantheon seems to have been well aware of the amples occur at the Horti Sallustiani (11.2-m span),
symbolic significance of numbers and geometry. If a room in a Hadrianic bath at Otricoli (9-m span),
not for this background, the five rows of twenty- the “Temple of Venus” at Baiae (26.8-m span), and
eight coffers may be considered simply a convenient semidome (10.5-m span) of a room of a bath at the
46
P1: KIC
38. Reconstruction of formwork for two types of segmental domes.
“Villa of the Gordians.”68 The centering for such Domes and semidomes became particularly popu-
vaults would have required intricate and elaborate lar during the late third and fourth centuries, so more
carpentry, and the fashion for building them may well preserved evidence exists from this period than from
have been inspired by the tastes of the emperor him- earlier times. The formwork imprints are visible on a
self. They do not seem to have been as common in number of examples, and these allow us to understand
later periods.69 better how some of these vaults were built. The domes
39. Serapeum at Hadrian’s Villa (a.d. 125–133). Excavation photo of segmental semidome showing
remains of the concrete dining couch below. Fototeca Unione c/o American Academy in Rome,
neg. #5984.
47
P1: KIC
with preserved formwork imprints from this period with the removal of the formwork. They along with
are built exclusively with radially arranged formwork the ribs in many of the later Roman domes would
whereas the semidomes were built with either hor- have had the additional advantage of regulating the
izontally or radially arranged boards (Appendices 2b creep.
and 2c). The builders ran into problems in the construc-
Rasch’s photogrammetrical documentation of late tion of the centering for early domes, but by the late
Roman domes reveals that the board lengths tended Empire they had refined and regularized the process,
be cut in standard lengths, which implies a fairly which in part, explains the proliferation of domes
systematic approach to the construction of the during the fourth century. As discussed in Chapter 5,
centering.70 Rasch argues that the radially arranged this regularization is also related to the development
formwork was used so that each tier of boards could of ribbing that was used for cross vaults, domes, and
be removed before the next tier was built and sug- semidomes of this period. The construction of the
gests that the horizontal courses of bipedales that wooden forms was crucial to the success of a project,
occur in some domes, such as the Pantheon (Fig. 46, and until this was mastered the concrete itself was of
p. 62), Mausoleum of Helena (Fig. 42, p. 57), and the secondary importance. The various forms of domes,
“Temple of Minerva Medica” (Fig. 95, p. 111), mark semidomes, and segmental vaults used at Hadrian’s
the end of a “pour” and the point at which the form- Villa are as much works of fine joinery as they are of
work was moved up to the next level.71 Ever an ad- any advances in concrete technology per se.
herent to the principle of Ockham’s Razor, I would
argue that this idea promotes more problems than it
conclusions
solves – again, possible but not probable. The repeated
process of easing, striking, dismantling, and reassem- Advances in wooden technology were necessary be-
bling would have been time-consuming. It would fore concrete vaults could become larger and more
have allowed for the reuse of wood, but it also could complex in form. Hand in hand with wooden con-
have caused a logistical nightmare for the organiza- struction came the technology for processing iron
tion of the masons and carpenters who would have and making tools to work the wood. Knowledge of
had to alternate constantly, resulting in unproductive carburization and quenching to harden the iron was
periods for both groups. In fact, leaving the form- known since Etruscan times, and by the first century
work in place for longer provides the added advantage a.d. the Romans were able to produce consistent re-
of helping the mortar retain the moisture needed to sults. The iron mines of Noricum also provided a ma-
acquire its full strength. More important is the ten- terial particularly well suited to the creation of steel
dency of concrete to “creep” or deform over time, for blades. The development of the frame saw also
especially during the first six months after being laid reduced the labor time for cutting long boards and
(see further, Chapter 3). Removing the formwork made building large vaults more efficient in terms
all at once would allow any initial movements to be of time and expense. Moreover, advances in joinery,
spread throughout the mass. The courses of bipedales probably developed in the context of making ships
that run horizontally through the dome may have and siege towers, provided the accumulated knowledge
been used as an organizational tool to mark the end for more complex forms of carpentry necessary for
of a layer of work, but they need not be associated the construction of large vaults.
48
P1: KIC
The use of the brick linings along the intrados of have changed as the spans increased. The progressive
vaults was probably introduced as a means of increas- weight of the concrete itself on the centering also
ing efficiency. The number of wooden planks in the became an issue as demonstrated by the deformed
formwork could have been reduced somewhat, and profiles of the barrel vaults at the Basilica of Maxen-
they would have been protected from the damaging tius. As the weight increased on the centering while
effects of the caustic lime so that they could be reused. the concrete was being laid, the wooden structure it-
The linings also would have made the removal of the self could deform in ways that did not occur in small
centering easier and reduced the amount of time and vaults. Finally, the process of removing the centering
labor involved. Most revealing is the close connection of large vaults was complicated by the complexity and
between the development of this technique and the weight of the components. Whereas in a small vault
fluctuations of the brick industry. The linings first ap- one could remove the struts and then let the pieces fall
peared when the number of brick producers was at its to the ground, in a vault such as the Pantheon dome
highest under Trajan, and they disappeared altogether one had to be much more careful and have an or-
after the death of Caracalla, when the sudden absence ganized plan for removing the centering. In general,
of stamped bricks suggests a dramatic decline in pro- the construction of centering, especially for domes,
duction. When stamped bricks reappeared in Rome appears to have become more efficient and standard-
under Diocletian and Maximian, however, the linings ized over time as shown in the evidence gathered by
did not. J. J. Rasch.
As the size of the vaults increased, the construc- The extensive use of cross vaults and domes did
tion and removal of the centering became more cum- not occur until at least the first century, and the late
bersome, and advances in woodworking tools, lifting appearance of these types of vaults must in part be
devices, and organizational skills became critical to because of the complexity of building the wooden
the success of the project. For a series of small bar- structures to support them. Laying out the curves and
rel vaults, as at the Republican sanctuaries at Tivoli, arranging the parts so that they could be efficiently
Palestrina, or Terracina, the wooden boards could eas- removed involved more sophisticated planning than
ily be reused from one vault to the next so that great had been necessary for barrel vaults. The evident need
quantities of sawed boards were not needed. For a to build these more complicated vaults on a regular
large vault, many more boards had to be used at the basis seems to have come from the construction of
same time. Efficient sawing techniques would have bath buildings in particular. The use of cross vaults
provided more incentive to build larger vaults. An- for the frigidaria solved problems of lighting the in-
other issue that arose was the size and weight of the terior spaces by leaving open lunettes that formed
beams for the centering structure, which in turn af- clerestory windows above the outer rooms, whereas
fected the type of lifting devices necessary on-site. As circular rooms were particularly well suited for the
beams became larger, the use of lighter and stronger hot rooms of baths since they provided radiant heat
wood such as fir and pine would have been desir- equally to all parts as described by Vitruvius.72 The
able. As shown earlier in the comparison of center- penchant for circular domed spaces during the fourth
ing arrangements at the Aula at Trajan’s Markets and century is largely because of the cultural acceptability
the frigidarium of the Baths of Caracalla, the means factor (explored further in Chapter 9), but the ad-
of putting the centering structure together would vances that had been made in the wooden technology
49
P1: KIC
for building vaults of double curvature during the pre- tassels along the edges.74 The effect of the draped
vious three centuries provided the accumulated knowl- cloth over radiating support struts is remarkably sim-
edge that allowed for the transitions in architectural ilar to the pumpkin domes. Iconographic studies of
form that occur in the fourth century. Eventually, the vault decoration by both K. Lehman and D. Hemsoll
construction of centering for cross vaults and other have shown that the theme of canopies as ceilings
complex forms was avoided completely by the use of goes back to an example described by Athenaeus of
pitched brick vaulting, as can be seen in a couple of the dining pavilion set up before the procession of
the Honorian towers of the Aurelian walls (a.d. 401– Ptolemy II in Alexandria.75 The segmental semidome
403) and by the use of tubi fittili (see Chapter 4), as seen at the Serapeum at Hadrian’s Villa provides a useful
the fifth-century Church of San Stefano Rotondo.73 comparison, as the function of the space is known to
Segmental domes, or “pumpkin” domes, were the have been a dining hall from the semicircular concrete
latest vault type to appear and required the most com- dining couch that remains (Fig. 39). The vault itself
plex centering. Once the radial formwork for domes was once decorated with colorful mosaics, only a few
was mastered, the construction of the segmental tesserae of which remain, that could have imitated the
domes would have required only simple but perhaps woven decorations of tapestries.76 Egyptian themed
time-consuming additions to the basic framework. statuary found in the area, such as representations of
They were the most complicated of the wooden Isis, Canopus, and the Nile, would have further rein-
centering, but they offered great decorative poten- forced the Alexandrian theme.77 Given the context
tial, which perhaps made the new form worth the of the early examples of the segmental domes and
added effort. The remains of the decoration of the semidomes, one wonders if the initial impetus for
semidome of a niche in the circular room at the the complex wooden centering that was required to
Large Baths at Hadrian’s Villa shows that the undu- build them was inspired by the desire to create the
lating form often alluded to hanging canopies. The semblance of a light and impermanent structure used
painted decoration, which has deteriorated in recent by Hellenistic kings. The Roman builders, however,
times, was drawn in the eighteenth century when it had the ability to create the same effect in a solid and
was still legible and showed a tapestry complete with permanent structure of concrete.
50
P1: NDZ
3
INGREDIENTS: MORTAR AND CAEMENTA
T he first task after setting up the centering

was to mix the mortar and choose the caementa.
The best-quality Roman concrete during the impe-
(H2 O) hardens and gains strength through its contact
with carbon dioxide (CO2 ) in the air as the water
evaporates; as a result, the mortar at the center of a
rial period was extraordinarily strong and durable. mass does not develop the same degree of strength
This is in part due to the addition of pozzolana but as that in contact with the air. Unlike the quartz
also to the high-quality lime that was available. In sand, which is inert, the pozzolana plays an active role
addition, the choice of caementa played a significant in the chemical transformation of the mortar during
role in the stability of the structures as did the care in the hardening process. Pozzolana contains both silica
placing them within the mortar and ensuring that the (silicon dioxide, SiO2 ) and alumina (aluminum ox-
mixture was very compact. In the following sections, ide, Al2 O3 ), which through the eruptive process are
I examine the individual ingredients of the concrete converted into soluble forms allowing a chemical re-
mixture to show how they interacted with each other action to take place when mixed with water (H2 O)
and how and why the builders chose the varieties that and slaked lime (Ca(OH)2 ).2
appear in the extant remains. Today the most common type of hydraulic mor-
tar used for modern concrete is made with Portland
cement. It, too, goes through a chemical process in-
mortar
volving silica and alumina, but the cement powder
The mortar used by the Romans employed poz- is made by mixing the raw materials of limestone
zolana, a volcanic ash that imparted added strength (rich in calcium carbonate) and clays (rich in sili-
and hydraulic qualities (the ability to harden under cates and aluminates) and then firing them together
water) that were lacking in the simple lime mortar in a kiln to produce a material that is reactive when
used by the Greeks. Recent studies show that the mixed with water. This manufacturing technique was
resistance to compression of pozzolana-lime mor- only patented in 1824 by Joseph Aspdin, who named
tar is five to eight times stronger than that of lime it after the high-quality building stone quarried in
mortar.1 A simple lime mortar made of siliceous Portland, England. For both ancient lime-pozzolana
quartz sand (SiO2 ), slaked lime (Ca(OH)2 ), and water mortar and modern Portland cement mortar, the key
51
P1: NDZ
was to create the soluble form of silica and alumina during which time the mortar gains 10–20 percent
through a firing process. The ancients relied on the of its ultimate strength. After the initial development
heat of volcanic eruptions to produce a naturally re- of the crystalline rods, the hydrated silicates in the
active material, pozzolana, whereas modern manu- pozzolana begin to form an amorphous gel that is
facturers rely on kilns to create man-made reactive transformed into long hollow tubes of hydrated cal-
material. Today pozzolana is sometimes mixed with cium silicate glass. The progressive enmeshing of these
Portland cement, especially in volcanic areas such as tubes, which continues for years, provides the major-
Italy where the natural material is abundant. ity of the ultimate strength of the mortar.7
Pozzolana-lime mortar develops its strength slowly Mortar made with pozzolana, in fact, acquires
over a period of time. It sets and loses its fluidity greater strength when it is cured in water rather than
within hours but only reaches its expected strength air. Rapid drying can substantially reduce its strength,
during a long period of curing. The initial set (the loss although higher temperatures increase the strength as
of plasticity) for hydraulic mortar (with pozzolana or long as the mortar is kept sufficiently moist.8 Today,
Portland cement) can occur in 3–4 hours, whereas pours of concrete often are covered with canvas or
the final set (the ability to resist pressure) occurs af- sand and kept damp for a week to ensure that there
ter about 10–12 hours.3 Once the mortar begins to is enough water to complete the chemical reactions
set, the chemical reactions taking place cause it to necessary to give the concrete its strength.9 Vitruvius
become warm. The heating process usually reaches seems to be unaware of the beneficial effects of curing
a peak within 4–8 hours and then slowly declines the mortar underwater when he recommends that if
over the next couple of days.4 Vitruvius refers to the concrete cannot be poured directly into a coffer-
the phenomenon when he explains: “Then the mois- dam for harbor construction, a concrete block made
ture starved heat latent in these types of ingredients with sand from Cumae (pozzolana) should be con-
[lime and pozzolana], when satiated by water, boils structed out of the water and allowed to dry for two
together and makes them combine.”5 After the con- months before lowering it into the water.10 Frontinus,
crete has set and cooled, it begins the period of cur- writing a century later, was aware that temperature
ing, which for hydraulic mortar involves more than affected the quality of the mortar when he recom-
simply “drying out.” Unlike simple lime mortar, the mended that mortar work (structura) be restricted to
process is largely independent of the carbon dioxide the period between April 1 and November 1 with a
in the air, but it does require adequate quantities of break for the hottest part of the summer; however,
water for the chemical reaction to take place through- his advice was not always followed as recent evidence
out the mass.6 The granules of pozzolana are partic- shows that the builders at the Baths of Trajan were
ularly effective in the process because they are very laying brick in February.11 Frontinus says nothing
porous, resulting in large surface areas with which regarding the beneficial effects of curing the mor-
with the water and lime can react (Pl. III). Mod- tar in water. By the third century, however, Dio
ern concrete takes about twenty-eight days to gain a Cassius notes specifically that the pozzolana found
majority of its ultimate strength, but it continues to between Misenum and Pozzuoli becomes petrified
gain strength for months and even years. During the when mixed with lime and water as long as it remains
setting and hydration process, hexagonal crystalline in liquid,12 which suggests an improved knowledge of
rods form causing the initial hardening of the mortar, the material over time.
52
P1: NDZ
Shrinkage and creep are two interrelated character-

istics that affect behavior of all types of concrete over
time.13 As concrete cures and eventually loses its mois-
ture content it shrinks, and this contraction can result
in increased internal stresses. Creep is the permanent
deformation of a material under long-term loading.
Once the concrete creeps, it does not return to its
original form. For modern concrete, creep is greatest
during the first six months, and then it proceeds at
a decreasing rate for two to five years. The amount
of creep is determined by the strength of the con-
crete, and concrete mixtures that use weaker mortar
or more fragile materials for the caementa are more
likely to display a higher degree of creep. The shape
of a concrete vault a year after the centering is re-
40. Diagram showing the cycle of chemical changes in the process
moved is usually somewhat different than the form of
of lime mortar production.
the original wooden centering. Such deformations
also can occur during the construction process be- as it takes on water and can increase in volume up
fore all the structural elements of the building are to 3.5–4.0 times the original volume once it be-
put in place, as was the case with the noticeable comes putty.15 The slaked lime could then be used
deformations in the vaults of the Hagia Sophia in to make either simple lime mortar by mixing it with
Istanbul.14 sand and more water or hydraulic mortar by mixing
it with pozzolana (or crushed terracotta) and more
water.
Lime
The quality of the lime used for mortar was de-
The production of lime for mortar is a two-step termined by the type of stone and the duration and
process: firing and slaking. First the limestone or an- temperature of the firing. The firing time is affected
other calcium-rich stone, such as marble or traver- by the form of the kiln, the heat of the fire, the type
tine, is fired in a kiln to produce quicklime, which and size of stones, and their placement in the kiln.
comes out of the kiln as a very lightweight version Temperature and duration of firing is different for
of the original stone. Before the quicklime can be each type of stone, so if a particular type of stone is
used for mortar, it has to be put through a slaking fired for too short a time or at too low a tempera-
process in which the fired stones are combined with ture the resulting quicklime is not fully calcined. A
water (Fig. 40). During this process, the quicklime fine balance must be maintained for the highest qual-
(CaO) combines with the water (H2 O) to create cal- ity material.16 For this reason, the best-quality lime
cium hydroxide (Ca(OH)2 ), or slaked lime. It first is produced from similar stones fired together. This
disintegrates into powder (the most common form is also one advantage of the stones being fired near
today) after absorbing one third its volume in water, the quarry so that the workmen tending the kilns are
but becomes putty as more water is added. It expands familiar with their material. The quality of the lime
53
P1: NDZ
and the resulting mortar of the fourth century and

later is often inferior to that of the first and second
centuries, and this might be in part explained by the
fact the spolia of various types of stones were fired
together for making the lime at this time, as can be
seen from the numerous lime kilns in late antique
Ostia and in Rome itself.
The quicklime is very volatile before it has been
slaked, and care has to be taken to keep it protected,
as it is susceptible to “air slaking,” which is when it
reacts with the carbon dioxide and the moisture in the
air. If the quicklime becomes air slaked, it then loses
its potency and becomes inert. This phenomenon
was recognized in Renaissance times, as shown by
Alberti’s comment that the quicklime should not be
allowed to lie around for too long after firing because
if “exposed to breezes, the moon, or the sun, espe-
cially in summer, it will very soon turn to ashes and
become useless.”17 This advice for quicklime is not
to be confused with Pliny’s comment that lime paste
(calx intrite, or slaked lime) improves with time.18
During periods of intense building, there could
well have been a warehouse system for acquiring aged
41. Casa del Sacello Iliaco at Pompeii (I.6.4) showing quicklime
slaked lime. During the eighteenth century, the sup- stored for use in plaster of wall paintings that were in progress at
pliers to Rome offered both slaked and unslaked lime, the time of the eruption, a.d. 79.
but they added 13 percent to 25 percent to the cost of
the quicklime if it was supplied already slaked.19 In an-
Pozzolana
cient times, the lime was undoubtedly slaked on-site
for small private projects, as shown by the remains at Pozzolana is a modern generic term for volcanic ash
the Casa del Sacello Iliaco at Pompeii, where chunks used in the building trade to make mortar, and it
of quicklime were found stacked ready to be slaked applies to a type of material produced by volcanoes
for the plastering of the walls, which was only partly throughout the world. The modern term thus cov-
completed at the time of the eruption (Fig. 41).20 ers a wider range of materials than what the ancient
For large imperial projects in Rome, there is little ev- writers came to call “pulvis puteolanus,” or dust from
idence for whether the quicklime was slaked on site Puteoli (modern Pozzuoli) from which the modern
or elsewhere, although the high quality of the lime name is derived.21 In fact, pozzolana can range from
used in the first and second centuries in Rome sug- dust-sized particles to walnut-sized chunks (Pl. III).
gests that there existed a developed supply network By the second century b.c., the Romans discovered
to ensure the best quality. that the volcanic ash from the Bay of Naples made
54
P1: NDZ
a very strong mortar that would set under water. The table 1. Recommended mortar mix proportions
earliest documented use of the exportation of pul-
Mortar type Proportion Ancient source
vis puteolanus outside of the Bay of Naples occurs in
the breakwater of the harbor at Cosa, which prob- lime:sand 1:2 Cato Agr. 25
1:2 (river) Vitr. De arch. 2.5.6
ably dates to the late second or early first century 1:3 (river/sea) Plin. HN 36.175
b.c.22 Pozzolana also was produced by the volcanic lime:harena 1:3 Vitr. De arch. 2.5.5
districts to the north and south of Rome (Sabatini fossicia 1:4 Plin. HN 36.175
and Colli Albani), and the Roman builders recog- lime:pulvis 1:2 Vitr. De arch. 5.12.8–9
nized this product as an effective material for creating puteolanus
strong mortar, but they did not equate it with the lime:crushed 1:1:2 Vitr. De arch. 2.5.7
terracotta:river 1:1:3 Plin. HN 36.175
same type of material found near Puteoli.23 They
sand
referred to it by a different name, harena fossicia,
lime:sand (for 2:5 Vitr. De arch. 8.6.14
sometimes translated as “pit sand.” By the imperial a well) 1:2 Faventinus 4
period, the builders in Rome were using the lo- 1:2 Palladius 1.9
cal harena fossicia (i.e., pozzolana rossa, pozzolana nera,
and pozzolanella) exclusively in their buildings and
were not in the habit of importing the pulvis pute- with river sand. One possibility is that the pulvis pute-
olanus from the Bay of Naples. However, as late as olanus was more expensive than the pit sand so that it
the mid-first century a.d., Pliny the Elder noted that vied with the lime in terms of cost. This was surely
the emperor Claudius had pulvis puteolanus brought the case for the harbor works at Caesarea Maritima,
up from Puteoli to build the foundation for the light- where pozzolana was imported from the Bay of
house at his harbor at Portus,24 and archaeological Naples. In trying to save on the use of pozzolana,
evidence appears to support his claim.25 This may the builders in Caesarea Maritima made use also of
indicate that the builders still believed that the pul- simple lime mortar, a cost-saving device that eventu-
vis puteolanus was more effective for underwater ally contributed to the demise of the harbor structures
construction.26 there.28
In giving advice on how to make a good mor- Of the local “pit sands” used in Rome from the
tar, ancient authors do not always agree on the ap- second century b.c., changes occurred in the pattern
propriate proportions of lime to sand/pozzolana (Ta- of use over time as the builders and quarry work-
ble 1). The proportions given by both Vitruvius and ers became more aware of the locations and efficacy
Pliny the Elder indicate that if pit sand (harena fos- of the local pozzolanas. Mortar in Rome during the
sicia) is used instead of river sand (harena fluvialis),27 Republican period employed grayish pozzolana of
then the amount of lime is reduced (Vitruvius: from low quality, presumably pozzolanella, which could be
one third to one fourth; Pliny: from one fourth to one quarried in the open air (Pl. II).29 Van Deman noted
fifth). Vitruvius goes on to advise that when using the that mortar using the higher quality red pozzolana be-
pulvis from the Bay of Naples for hydraulic works, gan to appear in the Late Republican and Augustan
the amount of lime should be increased from one periods.30 An abandoned quarry of red pozzolana,
fourth to one third. One wonders why the amount now located under the basilica of San Sebastiano on
of lime is increased to the same proportion as used the Via Appia, was reused for tombs in the Late
55
P1: NDZ
Republican period implying that its life as a quarry level vaults, which may have been intended to lighten
began earlier.31 the mixture (Pl. V). This detail reveals the level of at-
Both E. Van Deman and M. E. Blake pointed to tention being put into the mixing of the mortar at
the “earthy quality” of the mortar in Rome up until the time.
the time of Augustus.32 The strata of red and black The clean white appearance of the mortar during
pozzolana deposits around Rome are separated by a the late first and second centuries suggests that the
layer of soil that is often difficult to distinguish from pozzolana was washed to remove dust and dirt be-
the red pozzolana (Pl. IV).33 The “earthy quality” may fore it was added to the mortar.37 A comment by
be the result of not distinguishing clearly between the Alberti in the fifteenth century may shed some light
layer of red pozzolana and the overlying layer of soil. on how this was accomplished in ancient times. He
In fact, Vitruvius’s advice on how to determine the noted that any pit sand that leaves water muddy when
most appropriate pit sand to use for building prob- stirred is not good.38 The tendency for dirt to remain
ably reflects this problem. He noted that it must be suspended in the water while the pozzolana sinks is
sharp angled so as to make a noise when rubbed be- a phenomenon that would have aided in the wash-
tween the fingers and that it should be clean enough ing and purification of the pozzolana. Such methods
not to leave a stain when pounded in a white cloth.34 often were employed in metal ore processing (called
Because much of the pozzolana consists of broken “jigging”), in which the material was washed with a
scoria, which is bubbly in texture, the pieces do of- sieve so that the heavier particles go through the sieve
ten have sharp edges from the broken bubbles. His to the bottom.39 Once the pozzolana was washed, it
advice came at a time when the mortar was begin- could have been further crushed if finer granules were
ning to lose the earthy quality of earlier mortar and needed.40
probably reflects a growing awareness of the impor- A visual inspection of the mortar used in the mon-
tance of avoiding the layer of soil that is adjacent to the uments in Rome from the first through the fourth
pozzolana. centuries shows that red pozzolana was always more
By the time of Augustus, the quality of the mor- common that black. Palladius, writing in the fifth
tar had improved. The pozzolana seems to have been century a.d., ranked the red pozzolana as the best;
coarsely sieved,35 although the reddish color of the however, black pozzolana is particularly noticeable in
mortar suggests that it was not washed of dust-sized the finest mortar of the Trajanic and Hadrianic peri-
particles. Throughout the Julio-Claudian period, the ods, especially at Trajan’s Markets and in the Pantheon
mortar retained the reddish color, although Blake dome, which has predominately black pozzolana.41
noted the presence of more black, gray, and white The use of so much black pozzolana is unusual, and
pozzolana during this period. By the Flavian period, its occurrence in the Pantheon dome would suggest
the mortar took on a much cleaner white speckled that it was considered a high-quality ingredient.
appearance.36 The whiteness of the lime matrix of the After the second century, the very clean white
mortar suggests that additional processing was occur- mortar diminished somewhat, though Van Deman
ring to clean the pozzolana before it was mixed with notes that the pozzolana continued to be washed. The
the lime. The mortar at the Colosseum is particu- mortar continued to be of a much higher quality than
larly unusual for the pea- to grape-sized pieces of it had been during the early first century.42 By the
whitish pumice mixed into the mortar of the upper late third century, when building recommenced in
56
P1: NDZ
Rome after a long period of turmoil, there was a

marked difference between the quality of the mor-
tar used in public and private structures. The large
imperial projects such as the Baths of Diocletian or
the reconstruction of the Curia display a good qual-
ity mortar with red and black pozzolana of varying
granule sizes. The private structures examined by T.
Heres (largely at Ostia), by contrast, were built with a
very friable mortar. Under Maxentius, the pozzolana
used consisted predominantly of red granules as large
as a centimeter in diameter, which is only possible
because of the wider mortar joints used in late brick-
work. This suggests less care in sieving than earlier,
and there are often pieces of extraneous materials,
such as pot shards or brick, that were not seen under
Diocletian. In general, there was a decline in the qual-
ity of the mortar used in state-sponsored monuments
after Diocletian. Under Constantine, the quality of
the mortar once again improved to yield a strong,
well-sifted binder with both red and black pozzolana
present.43
A recent study on the mortar of the Mausoleum of
Helena shows that in the fourth century the builders
were using red and black pozzolana along with poz-
zolanella, but that the choice was governed by the
location within the building.44 The domed structure
was built in two phases with an apparent change of de-
sign between them (Fig. 42). The walls of the phase-
1 building were constructed using pozzolanella that
was probably excavated from the nearby catacombs.45
Black pozzolana was found in only one sample from
the original structure, where it was used for the 42. Mausoleum of Helena (a.d. 326–330). Section showing two
phases of construction. The numbers along the intrados of the
mortar of the setting bed of a course of bipedales dome indicate the length of the formwork boards in RF.
near the top of the outer wall. Another course of
bipedales with a similar type of mortar was observed added in phase 2.46 It was structurally the most dar-
at the base of the wall. The builders used a better- ing part of the building and would have required
quality mortar for the courses of bipedales. Red poz- a tenacious mortar. These fourth-century builders
zolana was reserved for the mortar used in the con- used the local pozzolanella quarried on-site whenever
struction of the upper part of the dome, which was possible and saved the better pozzolanas (black and
57
P1: NDZ
red) for areas that required stronger mortar. These that a six-digit thick (c. 11 cm) layer of lime and
choices probably reflect a concern for minimizing crushed terracotta in the proportions of 1:3 be laid
costs. as a setting bed for marble slabs, mosaic, or opus spi-
catum (brick paving laid herringbone fashion).51 The
use of multiple layers of cocciopesto along with other
Cocciopesto
materials was recommended by Vitruvius for added
Cocciopesto is the term applied to mortar made with protection. He suggested that bipedales be laid over a
crushed terracotta, although it often contains poz- layer of cocciopesto with their joints sealed with a mix-
zolana as well.47 The term opus signinum is often also ture of lime and oil and then a second setting bed
used to describe this type of mortar, but C. F. Giuliani of cocciopesto be laid to receive the final layer of mo-
has argued that this is actually a misuse of the an- saic or opus spicatum.52 Such multiple layers of protec-
cient term.48 For clarity’s sake, I use cocciopesto here, tion are often found. The nature of the layers varies,
though in the modern literature they are virtually but the most common configuration is a setting bed
interchangeable. of cocciopesto on which is set a layer of opus spica-
The addition of crushed brick or terracotta creates tum covered with another layer of cocciopesto, which
a hydraulic mortar similar to pozzolana-lime mortar. forms the setting bed for a layer of mosaic (Fig. 43).
Both Vitruvius and Pliny recommend mixing crushed These multiple layers have sometimes been taken as
terracotta with mortar to make it stronger (Table 1, evidence for two different phases of paving,53 par-
p. 55). The firing of the clay, which, like pozzolana, ticularly when a layer of mosaic is covered, but the
is rich in silica, also produces a soluble silica com- consistent occurrence of it in so many different places
ponent. The hydraulic properties increase with the suggests that it was a common method employed as
firing temperature, so terracotta fired at higher tem- protection from the elements and should not nec-
peratures, such as roof tiles, create a more effective essarily be seen as indicative of a phase of repaving
hydraulic mortar. Because terracotta is less porous (Appendix 2e).
and has less surface area than pozzolana, which is The protection of the exterior surfaces of vaults was
very vesicular, the chemical reaction (and likewise the an important element in their development, especially
rate at which the mortar gains its strength) is slower in situations in which cracks occurred. Cocciopesto
than that of mortar made with pozzolana.49 If the was the fundamental ingredient used, but it also
crushed terracotta mortar remains in contact with wa- was combined with other types of coverings, such
ter for long periods, the hardening process continues, as roof tiles of terracotta (Trajan’s Markets, Basilica
and the mortar eventually develops a high degree of of Maxentius), marble (Pantheon), and even bronze
resistance. (Pantheon), which would serve to divert water even if
Cocciopesto is most commonly known as a material the vault itself developed cracks. The most specialized
to line cisterns, but it is also the most common ma- situations were those where vaults supported “hang-
terial used to protect the extrados of vaults exposed ing” gardens, or horti pensiles. In the Neronian addi-
to the elements. It was in use from at least the second tions to the Domus Tiberiana, the builders devised
century b.c.50 and during the imperial period was of- a system to protect the vaults from the moisture and
ten used as a setting bed for other paving materials that roots of the trees in the horti pensiles. The extrados
covered vaulted terraces. Vitruvius recommended covered with a 33-cm layer of cocciopesto that had a
58
P1: NDZ
43. Basilica Argentaria in the Forum of Caesar (a.d. 106–113). Detail of extrados (upside down). From
top to bottom: coarse layer of cocciopesto, opus spicatum, finer layer of cocciopesto, mosaic tesserae of leucititic
lava, finest layer of cocciopesto.
10 percent incline for drainage. On top of this was of the “Temple of Mercury” at Baiae and the Pont
then constructed a raised floor consisting of low sus- St. Martin outside of Aosta. In Rome itself, the cae-
pensurae made of pillars of three stacked bessales that menta were laid in roughly horizontal courses by the
supported a single layer of bipedales (Fig. 44).54 This Augustan period, as at the Theater of Marcellus, and
is the earliest documented example, but later exam- by the first century they were laid in very regular
ples of the system also have been found elsewhere on rows, as at the Colosseum.
the Palatine and under the garden areas at the Baths The general practice of grading caementa so that the
of Trajan (see Appendix 2e.8).55 It seems to have be- heavier ones were used in foundations and the lighter
come a standard method of protecting vaults that were ones were used in vaults appeared by the Augustan
used to support elevated gardens. period,56 although only later during the first century
did it become systematic. During the first century
b.c., the caementa in vaults in Rome often consisted
caementa
of refuse from earlier buildings.57 By the imperial pe-
In the earliest examples of concrete vaulting from riod, the builders were using a number of very spe-
the first and second centuries b.c., the caementa were cific lightweight stones in vaulting. The most com-
usually set radially on the centering as if they were mon were tufo giallo della via Tiberina (1,350 kg/m3 ),
voussoirs. Examples occur at the Republican sanctu- Vesuvian scoria (750–850 kg/m3 ), and pumice (600–
aries at Palestrina, Tivoli, and Terracina (Fig. 1, p. 5). 700 kg/m3 ). Van Deman cited the use of light cae-
Outside of Rome, the practice persists sporadically menta in vaulting as a contribution of the Neronian
into the Augustan period as can be seen in the dome period,58 but in fact examples of the use of imported
59
P1: NDZ
44. Domus Tiberiana (first century a.d.). Reconstruction of horti pensiles. Inset shows detail of suspensurae:
A = garden soil, B = suspensurae of bipedales set on stacked bessales, C = cocciopesto, D = concrete vault.
Vesuvian scoria exist from the first century b.c. of thought that was going into the construction of
(Appendix 2f). C. M. Amici has noted the use of this innovative structure.
scoria in the original vaults at the Forum of Caesar During the Trajanic period, one begins to see that
(Fig. 45).59 After this early use of the scoria it does not careful choices were being made in the type of cae-
become typical until the second century, but the sys- menta used for particular parts of buildings and for dif-
tematic use of tufo giallo della via Tiberina began with ferent types of vaults. The Baths of Trajan are the earli-
the construction of the Colosseum in about a.d. 70 est of the major Trajanic monuments, and though lit-
and continued throughout the Flavian period, when it tle remains today, there are tantalizing fragments that
was extremely common in imperial and public build- hint at the types of decisions the builders were mak-
ings around Rome.60 That the builders of the Colos- ing regarding the material used for caementa (Pl. VI).
seum were particularly interested in creating lighter That the builders were making distinctions between
vaults is also suggested by the inclusion of small pieces the use of tufo giallo della via Tiberina and tufo lionato
of white pumice in the mortar mixture of the upper- is clear from the remains of the exedra of section H,
level vaults (Pl. V).61 This is a very unusual use of where a marked change occurs between the caementa
pumice, as it was not used as caementa but, rather, as used at the top of the wall (brick and tufo lionato) and
part of mortar mixture, but it does indicate the level the springing of the vault (tufo giallo) (Pl. VII). The
60
P1: NDZ
tufo giallo della via Tiberina are the cross vaults of the
Aula (8.5-m span), which are structurally the most
precarious because they are raised on a series of four-
teen travertine piers. The decision to use tufo giallo
instead of tufo lionato reduced the weight of the vaults
by about 10 percent.63 The contemporary cross vaults
at the Basilica Argentaria also were built of tufo giallo
with pieces of tufo rosso a scorie nere added. The lat-
ter type also was used in small amounts at section E
of the Baths of Trajan. These two examples of tufo
rosso a scorie nere are tantalizing because their presence
alongside the tufo giallo della via Tiberina indicates that
materials from different volcanic flows north of Rome
were being mixed. This could indicate that the quar-
ries were located where the flows were adjacent to
each other so that material from both flows became
mixed. Alternatively, it could be because of mixing
at a central distribution warehouse, as seems to have
been the case for bricks from different figlinae, or of
adding refuse from older structures.64
The Vesuvian scoria, first used in the vaults at the
Forum of Caesar (Fig. 45), reappeared in the vaults
in Rome at the Basilica Ulpia65 and the Baths of
45. Forum of Caesar (mid-first century b.c.) Vault employing
Trajan (section E on Pl. VI). In both the Basilica Ulpia
Vesuvian scoria. Arrows indicate the remains of a vault in an
adjacent room also employing the scoria. and section E at Trajan’s Baths, the Vesuvian scoria is
combined very systematically with tufo giallo della via
section through this wall gives a clear picture of the Tiberina in alternating rows (Pl. VIII).66 The use at
thought involved in the choice and distribution of the Basilica Ulpia is particularly interesting because
the different types of caementa. The remains of the of the precarious nature of the vaulted structure. The
other large semidomes from the Baths of Trajan em- vaults in which the scoria was used make up the aisles
ploy only the tufo giallo, although during the excava- (6.2-m span), which were supported by colonnades.
tion at the base of the apsidal wall of section L, large In this case, the use of the lightweight caementa was
amounts of scoria(?) on the ground were reported, surely intended to reduce both the load on the marble
which probably belonged to the upper part of the architrave/frieze blocks that supported them and the
semidome.62 potential horizontal thrust of the vault. The addition
Careful decisions about caementa also were made at of the scoria to the tufo giallo would have lightened the
other Trajanic monuments. For example, the major- load of the concrete on the architrave/frieze blocks
ity of the vaults of Trajan’s Markets are built of tufo by almost 20 percent. The disadvantage of adding the
lionato and broken brick. The only vaults that employ scoria is that it created a more brittle form of
61
P1: NDZ
concrete, which was susceptible to

cracking, and by placing the scoria in
such regular courses the builders in-
advertently created planes of weakness
within the structure, which is shown by
the lines of breakage through the scoria
layers now visible in the fallen pieces.
The Trajanic examples of the com-
bined use of scoria and tufo giallo della
via Tiberina are the direct precursors to
the most renowned use of graded cae-
menta, the Pantheon. Here, tufo giallo
and scoria were laid in alternating lay-
ers in the upper part of the dome,67
whereas tuff and brick were used in
the haunch. The walls employed a suc-
cession of heavier materials proceed-
ing toward the ground level (Fig. 46).68
Structural analyses of the Pantheon
have shown that the use of lightweight
caementa in the dome does reduce the
horizontal thrusts on the rotunda walls,
although the dome could have stood
even without it. The efficacy of em-
ploying scoria and pumice at the crown
of vaults is an issue explored further in
Chapter 8.
After a hiatus in the construction of
large vaulted structures in Rome dur-
ing the second half of the second cen- 46. Pantheon (a.d. 118–128) Section showing distribution of materials.
tury, different patterns of use appear in
the choice of caementa when large-scale
building is renewed under the Severans in the early were called for in some of the large or structurally pre-
third century. The last documented example of the carious vaults at the Baths of Caracalla, the Vesuvian
use of tufo giallo della via Tiberina and scoria is in the scoria was again used, but it was often employed alone
large barrel vault of the Hadrianeum (c. 18-m span) rather than mixed with tuff (Pl. XIII).70 The most
69 consistent use of the scoria occurs in the two palaes-
dedicated in a.d. 145. Under the Severans, the tufo
giallo was no longer used in vaults and was replaced trae. The large exedrae (c. 26.8-m span) opening
by the heavier tufo lionato. When lightweight caementa onto the palaestrae employ a very distinct gradation
62
P1: NDZ
map 5. Bay of Naples showing the air fall distribution of the a.d. 79 eruption of Vesuvius in relation to
that of the pre-79 eruption on which Pompeii was built.
of materials: brick in the lower third, tufo lionato in the he mentioned. The Basilica Julia, like the Basilica
middle third, and scoria in the upper third (Pl. IX). Aemilia, had a series of cross (?) vaults (5.1-m span)
The vaulted porticoes around three sides of each supported on an arcaded structure, which is a typical
palaestra also employed scoria alone in the upper three situation in which to find the use of the lightweight
quarters of the vault, whereas brick was used in the scoria.
lower quarter. Like the Basilica Ulpia vaults, these At the Baths of Diocletian (a.d. 298–306), a dif-
were supported on a colonnade, so the light material ferent type of lightweight material, pumice, was used
here would have been used to reduce the weight on for the caementa in the vaulting. Most examples are no
the architraves. Both monuments also employed iron longer visible, but reports from previous centuries in-
tie bars in these vaults to ensure stability (Chapter 6). dicate different types of pumice, described variously
The latest building in this study that employed as yellowish-white, black, and white (Pl. XIII).72 The
the Vesuvian scoria is the reconstruction of the only example visible today occurs in the vaults of the
Basilica Julia after a fire destroyed it in a.d. 283. porticos of the east palaestra. Large pieces of dark gray
A piece of vaulting lying at the southernmost cor- pumice were used at the crown, whereas tufo lionato
ner of the building has caementa consisting exclu- was used in the haunch (Pl. X).73
sively of scoria. L. Canina reported in 1860 that Soon after the Vesuvian scoria was replaced by
remains of the fallen vaulting belonging to the other types of pumice, tufo giallo della via Tiberina
Diocletianic restoration were found on the ground.71 reappeared for the first time since the mid-second
He did not describe the material in them, but the century a.d. in the vaults of the Basilica of Max-
present chunk of vaulting must belong to those entius (a.d. 306–315). Recent excavations and cores
63
P1: NDZ
taken above the barrel vaults also have revealed the the river port of Pompeii at the mouth of the Sarno
use of gray pumice there.74 Later, in the dome of was the major point of exportation, as indicated by
the “Temple of Minerva Medica,” tufo giallo della Strabo.77
via Tiberina was combined with a grayish pumice at The most common use of Vesuvian scoria in Rome
the crown (Pl. XI). (See also Arco di Malborghetto, occurs after the a.d. 79 eruption when Pompeii was
Appendix 1.38). destroyed, but as yet there is no indication that the
79 eruption produced the same type of scoria, so
presumably the scoria was excavated from under the
Provenance of Caementa
79 deposits.78 The mineralogical analysis of the sam-
Most of the materials used for caementa in the vaults ples shows that the one pre-79 sample, from the
in Rome were local, either products of the Colli Forum of Caesar, has mineralogical properties some-
Albani or the Sabatini volcanic districts; however, the what different from the later samples, which sug-
scoria used in the Pantheon dome was analyzed by gests that the quarry source changed. The other sam-
Gioacchino De Angelis d’Ossat during repairs to the ples also fall into groups according to time period.
dome in the 1930s and determined to be a prod- The Basilica Ulpia and Trajan’s Baths samples have
uct of Vesuvius. The question arises whether or not similar mineralogical compositions that are slightly
the other examples of similar-looking brown scoria different from both earlier and later samples, sug-
were imported into Rome as well. Recent analyses gesting that all the Trajanic material was coming
on samples from the Forum of Caesar, the Baths of from the same quarry. The samples from the third-
Trajan, the Basilica Ulpia, the Baths of Caracalla, and century projects at the Baths of Caracalla display
the Basilica Julia reveal that in all cases the material their own characteristics, suggesting yet a different
was from Vesuvius (Appendix 3). quarry.
Vitruvius mentioned a type of lightweight mate- Less work has been conducted on the pumice
rial from Vesuvius that he called pumex pompeianus, or found in the fourth-century buildings, but in at
“pumice from Pompeii.”75 This material (often called least one case, the “Temple of Minerva Medica,”
cruma or ferruggine in modern literature) was produced Gioacchino De Angelis d’Ossat determined that the
by minor eruptions dating to the tenth/ninth cen- pumice used there has similar characteristics as prod-
turies b.c. that formed the hill on which Pompeii ucts from the Sabatini system (an abundance of sani-
was built.76 The builders must have often excavated it dine crystals and acidic chemical composition). He
when digging their foundation trenches, which then described pumices of various colors produced by this
led to its use in the walls there, and the excess could system and suggested that the pumice at Minerva
have been exported. However, much material also Medica was acquired from the nonlithified tuffs north
could have come from the hinterland. A mapping of of Rome such as tufo rosso a scorie nere and tufo gi-
the distribution of these eruptions shows that they allo della via Tiberina. His son, Guglielmo De Angelis
mainly affected the area to the east of Vesuvius and d’Ossat, described the yellowish pumice from the
that Pompeii lies along the outer limits of the activity Baths of Diocletian as being from the “vulcani laziali,”
(Map 5, p. 63). Vitruvius’s association of the material which refers to the craters north of Rome,79 although
with Pompeii does not necessarily indicate that it was whether this was verified scientifically by his father is
typically excavated there but, rather, may indicate that unclear.
64
P1: NDZ
conclusions part due to the firing of different types of stone to-

gether once spolia began to supply the lime burners.
The best-quality Roman mortar occurs during the The concern displayed in the Theodosian Code for the
late first and early second centuries a.d. By that time, supply of lime to Rome during the fourth century
the builders were distinguishing between what was is a reflection of supply problems that had not been
pozzolana and what was simply soil. Although we encountered in earlier times.80
have no direct evidence for the processing of the poz- Waterproofing the extrados of vaults was accom-
zolana, they seemed to have both sieved and washed plished by a special type of mortar, cocciopesto, made
it by this time. The removing of the soil resulted in with the addition of crushed terracotta. Both terra-
a higher proportion of reactive material in the mor- cotta and pozzolana would produce a hydraulic mor-
tar. The mortar of the late empire especially in pub- tar, but the greater density of the terracotta made it
lic buildings could be very hard and of good qual- a better waterproofing agent. Pozzolana mortar, by
ity, but the very clean quality is no longer present. contrast, was preferred for structural conditions be-
This is presumably because less effort was put into cause it gained its ultimate strength quicker because of
the washing and sieving of the pozzolana and may the greater reactive surface area provided by the inter-
indicate that whatever infrastructure for the supply nal vesicles. In the special cases in which vaults were
and processing of materials that may have existed dur- used to support gardens above, the added protective
ing the early second century did not survive into the measure of a raised floor of bipedales supported on low
fourth century. suspensurae was used to create an additional barrier to
The extraordinary quality of the mortar during protect against dampness and root damage.
Trajanic times was in part related to the high level The choice of caementa used in Rome reflects a
of imperial and public building during this period. development similar to that of the mortar with the
This was a time of confidence particularly after the systematic use of lightweight caementa appearing in
Dacian Wars when the mines of precious metals came the Flavian period and advancing under Trajan. Tufo
into the hands of the Romans and funded much of giallo della via Tiberina was clearly the preferred ma-
the building in Rome. With so much building there terial from the reign of Vespasian, especially during
were obvious reasons for people to invest in the build- the late first and the second centuries. The sudden
ing industry, as reflected in the evidence from brick and ubiquitous presence of the tufo giallo in imperial
stamps. Direct evidence for the economy of lime and structures of the Flavian period may indicate a con-
pozzolana is less forthcoming, but the effort put into nection (direct or indirect) with the landowners from
the processing of the materials must reflect an atti- which the tufo giallo was quarried, though there is
tude present in the city at this time on both the part little direct evidence for the nature of such a connec-
of the government and of the private entrepreneurs tion. Large building programs such as those initiated
involved. One advantage of an established and “well- under the Flavians and Trajan would have provided an
oiled” supply system for lime was that workers who incentive for exploitation by the landowners. The ab-
knew the material fired similar stones together at the sence of tufo giallo della via Tiberina during the second
optimum temperature and duration, thereby produc- half of the second century is easily explained by the
ing the best-quality material. The lower quality one lack of major public buildings in Rome during the
finds in the fourth century and later was probably in economic hard times brought on by the plague that
65
P1: NDZ
ravaged Rome during the 160s. The fact, however, was shipped up the coast to Ostia before being sent
that it is not used in the Severan works on the Palatine upriver to Rome. One can imagine that the scoria
or in the Baths of Caracalla is surprising. used in the anomalous example at Ostia was taken
Close proximity of quarries to rivers would have from a load on its way to Rome. Furthermore, if
been a great advantage and may in part explain why the vault dates to the Trajanic/Hadrianic period as
the tufo lionato was always more common than tufo do the buildings on either side of it, the scoria could
giallo della via Tiberina as a material for caementa. The have been part of a shipment to one of the cluster of
ancient quarries of tufo lionato at Tor Cervara and buildings from the first half of the second century that
Salone are both located at points where the bends employed it.
of the Aniene come into direct contact with the de- The examples of Vesuvian scoria from the period
posits of tufo lionato so that the transportation between after the a.d. 79 eruption are all imperially sponsored
quarry and river was minimized (Map 3, p. 14). Tufo structures: the Baths of Trajan, the Basilica Ulpia, the
giallo della via Tiberina is an older unit than the tufo li- Pantheon, the Hadrianeum, the Baths of Caracalla,
onato and was therefore often found buried deeper and the reconstruction of the Basilica Julia, suggest-
and quarried underground in tunnels as at Grotta ing that it was an imperial commodity for impor-
Oscura. The mouth of the quarries at Grotta Os- tation. The 79 eruption devastated the area to the
cura opened toward the west (away from the Tiber) south and east of Vesuvius, and archaeological ev-
and onto the Fosso di Grotta Oscura, which then led idence suggests that the redevelopment of the area
down to the Tiber.81 Both the depth at which the did not occur until the early second century. Mile-
tufo giallo was typically found and the proximity from stones dating to the Hadrianic period show that the
the river’s edge may explain to some degree why tufo roads were starting to be reestablished by this time,
giallo was never as common as tufo lionato for caementa, and some poor tombs indicate a bit of activity during
even though it was better suited to the task. the second century.82 This accords well with what
The Vesuvian scoria is the only material used for is known about the typical development of topsoil
the caementa of vaulting that was regularly imported after an eruptive event, which takes a minimum of
from outside the environs of Rome, and this nat- twenty years.83 The area had once been known for
urally gives it both a rarity and a certain prestige its fertile soil, and many a wealthy Roman owned
value lacking in the other materials found locally. It property in the territory. We have no idea what hap-
was presumably shipped to Ostia and then to Rome. pened in terms of land ownership after the eruption,
In fact, one example of its use in vaulting at Ostia but the slow return of the topsoil may have pro-
exists, although it is an anomaly. It is found in a vided some incentive to exploit the volcanic material
fallen chunk of vaulting located next to the Hadrianic known to lie under the 79 deposits. The use of the
Horrea (I.20.1) and appears to have come from what- scoria in imperial structures beginning with Trajan
ever structure lay between it and the Trajanic Horrea could have been the result of some sort of imperial ini-
dei Mensores (I.19.4). This area has not been exten- tiative for redevelopment, though if so it is an undo-
sively excavated and much is still underground, so cumented one.
the original context is unclear. This is the only ex- The lightweight scoria (750–850 kg/m3 ) would
ample of any type of scoria or pumice used in vault- not have made good ballast material, which tends
ing at Ostia, and its use here is probably because it to consist of denser stones weighing 2,500 kg/m3 or
66
P1: NDZ
table 2. Amounts of scoria in vaults in Rome been skimmed from a shipment of scoria destined for
Rome.
Scoria Tons@850 # Ships@ # River Boats
Monument (m3 ) kg/m3 350 tons @ 200 tons The switch from imported Vesuvian scoria to local
pumice and the reappearance of the tufo giallo della
Basilica 1,980 1,683 5 9
Ulpia via Tiberina in the early fourth century suggests that
Pantheon 398 338 1 2 the resources to the north of Rome, which had been
Hadrianeum 1,380 1,173 4 6
ignored for a century and a half, were being exploited
1
again. The incentive for the use of the local pumice
Baths of 9,400 7,990 23 40
Caracalla and the reexploitation of dormant quarries may lie
Basilica 4,370 3,715 11 19 in Diocletian’s tax reform of a.d. 298, which obliged
Julia Italian landholders to pay their taxes in kind depend-
1 ing on what they had to offer.86 This year also marks
The figure for the Baths of Caracalla is taken from DeLaine 1997:
126 table 11. the beginning of the construction of the Baths of
Diocletian. The new tax scheme would have pro-
more;84 however, when compared to the weight of vided the incentive for local landowners to exploit
wheat (700 kg/m3 ), which was the prime shippable any natural resources they had available to fulfill their
commodity in the empire, it is not an unreasonably new obligation to the state. That Vesuvian scoria was
light cargo. We have no idea of its marketable value still used early in Diocletian’s reign when the Basilica
per unit volume, but the imperial builders in Rome Julia was reconstructed suggests that the switch from
must have placed a high enough value on its efficacy in imported to local was not because of broken supply
building large vaults to warrant the space it was given lines during the mid-third-century crisis but, rather,
on ships. A comparison of the amount of scoria used to some more immediate change. The first half of the
in monuments in imperial Rome and the tonnage of fourth century was a time when materials were not
typical seagoing vessels and riverboats provides some always easy to come by, and the various resolutions
idea of the scale of the import. The typical capacity issued to praefectus urbi recorded in the Theodosian
of the grain ships coming into Rome was 350 tons Code point to the problems of this period.87 The
(though it could be much larger) whereas that of the landowners around Rome had not been subject to
smaller riverboats, or codicariae, was about 200 tons taxes before the reforms of Diocletian, so creative
(Table 2).85 That it was, in fact, shipped is suggested exploitation of the products on their landholdings
by the one anomalous example of the Vesuvian scoria would have been one way of dealing with their new
in a fallen vault at Ostia (I.19.4), which must have tax burdens.
67
P1: KNP/KIC
4
AMPHORAS IN VAULTS
To the memory of John Lloyd, who taught me to appreciate the importance of undecorated pottery
T he use of amphoras in the concrete vaults

of buildings around Rome is a phenomenon that
has been recognized for centuries largely because of
A number of explanations have been proposed for
the use of the amphoras in vaults: that they reduced
the weight of the vault and thereby reduced its hori-
the ruined state of some monuments that has left the zontal thrusts,2 that they reduced the amount of other
amphoras exposed. The most renowned example is materials used,3 that they aided in the curing of the
the Mausoleum of Helena, which by the sixteenth concrete,4 and that they increased the resonance of
century was dubbed the “Tor Pignattara” from the sound in a space.5 The first explanation is the most
visible remains of the amphoras (or “pignatte”) in its common and generally accepted one. It was proposed
partially fallen dome (Fig. 47). The monument made by Alberti in the fifteenth century in his De re aed-
such an impression in the past that it now provides the ificatoria when he noted that he was “impressed by
name for the surrounding suburban area. This tech- the attempt to reduce weight by placing in the hips
nique of placing amphoras in the vaults has thus been [haunches] empty earthenware water vessels, which
long recognized, but it has sometimes been equated are cracked and turned upside down to prevent them
with or confused with another vaulting technique in from collecting water and gaining weight.”6 His ex-
which specially made terracotta tubes (tubi fittili) were planation has been repeated through the centuries un-
inserted into each other to form the permanent cen- til the present day and has taken on mantralike quality,
tering of the vault. They are, in fact, two quite sep- but a systematic investigation into the types of am-
arate techniques. The amphoras are reused material phoras used, the contexts, and the actual structural
within the vault whereas the tubes are made specif- effects has never been undertaken. The examples of
ically to act as the permanent centering. The use of amphoras in vaults around Rome fall into two main
tubi fittili is a technique that only became common groups: those that first appear in the mid-second cen-
in Rome in the fifth century and later and is, there- tury a.d. outside of Rome and those (the majority)
fore, beyond the scope of the present study.1 In the that appear in the fourth century both in and around
following discussion, I focus exclusively on the phe- Rome. In what follows, I try to bring together the
nomenon of inserting reused amphoras into the core available evidence in an effort to understand why the
of the vaults. technique was used, whether it was related to the use
68
P1: KNP/KIC
AMPHORAS IN VAULTS
47. Mausoleum of Helena “Tor Pignattara” (a.d. 326–330). Remains of the dome showing amphoras
and lattice ribbing.
of lightweight caementa, and, finally, how its purpose as for cocciopesto in a crushed form, for fill material in
changed over time. shard form, or for drainage tubes in the case of the
long thin Africana 1 amphoras.10 Wine amphoras are
never found in the vaults, presumably because they
types of amphoras used
could be rinsed and reused.11
The types of amphoras found in vaults are limited to In studies that mention the presence of amphoras in
those that contained oil or fish products and could not vaults, the type of amphora is rarely identified though
easily be rinsed and reused. Almost all examples are dimensions or drawings are sometimes provided, so
either Dressel 20 or Dressel 23 amphoras (Fig. 48), I have had to rely on published measurements and
both of which contained oil from the province of photographs to make some of the identifications. Be-
Baetica in Spain. Other types occasionally found em- cause the Dressel 20 and 23 amphoras are so common
bedded in vaults include one example of an Alma- and are fairly similar in form, I include some basic in-
gro 51c amphora from Portugal (third–fifth centuries formation on the differences between them that can
a.d.)7 that has recently been found in the vaults of be used as a means of identification (Table 3, p. 71).12
the Basilica of Maxentius8 and a number of exam-
ples of Africana 1 amphoras from Tunisia (third–early
early examples of amphoras in vaults
fourth century a.d.) that were built into an addition
to the “Casa di via Giulio Romano” at the base of The two earliest known occurrences of amphoras in
the Capitoline.9 Since the oil amphoras were difficult vaults both belong to the Hadrianic period and oc-
to wash, they were often recycled for other uses such cur outside of Rome proper.13 One is the Magazzini
69
P1: KNP/KIC
places, indicating that there were no other measures

taken to lighten the vaults. The number and distribu-
tion of the fragments of vaulting with amphoras lying
around the Villa alla Vignaccia suggest that the am-
phoras were used throughout the complex wherever
they would fit into the fabric of the building (Fig. 50).
A telling detail occurs in one structure there. A hole
in the wall reveals that the amphoras occur not only
in the vaults but also in the fabric of the walls. Im-
pressions left from two interconnected amphoras that
were built into the wall can still be seen (Fig. 51).
The location of these amphoras below the level of
the vault implies that they were not necessarily in-
tended to reduce the weight of the vault.
The use of the amphoras for sound resonance is
unlikely at both the Magazzini “Traianei” at Ostia
and the Villa alla Vignaccia. The former in particu-
lar was a utilitarian storage structure that would not
have required special acoustical treatments. In fact,
this explanation is probably based on a misinterpre-
tation of Vitruvius when he recommended the use
of bronze resonating pots placed below the seats in
theaters to increase the amplification.15 Later, how-
ever, the practice of building terracotta pots into walls
did become a method of increasing the resonance
48. Amphora types found in vaults in Rome.
of spaces in churches where chanting priests prac-
“Traianei” at Ostia (I.20.1) and the other is the Villa ticed the liturgy,16 but this is unrelated to the Roman
alla Vignaccia at the fourth mile of the Via Latina.14 examples.
At Ostia, two Dressel 20 amphoras were placed in the An unusual example of the Dressel 20 amphoras
haunch of each corner of the cross vaults (c. 4–6-m occurs in a vault in Baetica, the region of Spain
span; Fig. 49). At the Villa alla Vignaccia, Dressel 20 where the amphoras were produced. They are in
amphoras are visible in both standing and fallen vaults the fallen vault of a semidome (c. 7.5-m dia) in the
(5–11-m spans) scattered throughout the present-day Casa de la Exedra at Italica (Fig. 52).17 This is the
park in which the monument is located (Figs. 50–51). only example known from the area of Baetica. Ital-
These early examples at Ostia and the Villa alla ica is located about eight km from Hispania (mod-
Vignaccia occur in vaults that do not have particu- ern Seville), which was the port city through which
larly large spans and have no obvious need for special many of the Baetican amphoras passed on their way
structural precautions. The caementa used in both cases to seafaring ships headed for Rome and elsewhere.18
is tufo lionato, which was readily available near both The Dressel 20 amphoras were produced nearby along
70
P1: KNP/KIC
AMPHORAS IN VAULTS
table 3. Physical characteristics of Dressel 20 and 23 amphoras
Characteristics Dressel 20 Dressel 23
Max. Height (cm) 80 62

Max. External Diameter (cm) 55 45
Avg. Capacity quadrantalia 3 1–15
lt 78 26–39
lbs (oil)1 216 72–108
Diameter of Mouth Opening (cm)2 8.4–11.7 5.8–7.0
Form of Rim – concave with groove to form – slopes inward with no
lip for setting a stopper internal groove
Type of Foot Closure – hole at bottom closed with – exterior spike formed as part
wad of clay on interior and of body of pot and interior
spike added on exterior has small hollow at bottom
Handles – attached to shoulder below – attached directly to rim or
rim just below it
External Markings – until a.d. 260 painted – none
inscriptions provided
information on the weight
of the vessel, contents,
people who packaged it,
place where it was
packaged, consular date
Date mid-first–fourth century3 mid-third–fifth century4
1
The Roman pound (libra) equals 0.327 kg.
2
These figures are taken from the Cabrera 3 shipwreck and therefore represent third-century forms: Bost et al. 1992: 119.
3
Manacorda 1977: 134–7; Keay 1984: 403 (unstamped Dressel 20); Peña 1999: 86.
4
Manacorda 1977: 137–40. Evidence from kilns along the Guadalquivir Valley (Remesal Rodrı́guez 1983: 115–31) and from the Cabrera 3
shipwreck (Bost et al. 1992: 125–6, figs. 5–6) confirm that the two types were produced and shipped together at least during the later part of
the third century.
the Guadalquiver valley and would have been readily The use of the amphoras in vaults has typically
available. The proximity to the manufacturing area been seen as a newly invented technique of the im-
of the pots may explain their presence in this con- perial period and as an independent phenomenon in
text, though one might expect to find other exam- the history of building construction, but the plac-
ples in the area. L. Roldán Gómez, who specializes ing of the amphoras in vaults is best viewed as part
in the construction techniques of this area, has noted of a long history of the reuse of amphoras for land
the Casa de la Exedra employs unusual construction reclamation projects.20 Outside of Italy, the practice
techniques for Italica and may have been particu- of burying large numbers of amphoras for land recla-
larly influenced by the construction techniques from mation projects in marshy areas dates back at least
Rome.19 to the fourth century b.c., when it was used at the
71
P1: KNP/KIC
49. Magazzini “Traianei” at Ostia (I.20.1) (c. a.d. 126). Detail of a cross vault showing Dressel 20
amphoras used in the corners.
50. Villa alla Vignaccia (c. a.d. 130). Plan showing numbers and locations of Dressel 20 amphoras at the
site.
72
P1: KNP/KIC
AMPHORAS IN VAULTS
51. Villa alla Vignaccia (c. a.d. 130). View of section C. Arrows indicate amphoras. “A” marks amphora
impressions within the wall.
52. Casa de la Exedra at Italica, Spain (second century a.d.). Detail of fallen apsidal vault (c. 7.5-m
span) showing the use of Dressel 20 amphoras, which were manufactured nearby. Note also the use of
smooth river rocks as caementa.
73
P1: KNP/KIC
known example is the mid-first century a.d. Castra

Praetoria amphora deposit published by H. Dressel.
The amphoras, consisting of various types made over a
span of seventy-nine years, were stacked upside down
between three and six rows high. They were all put in
place at the same time, with some of the earlier ones
in the upper rows and the later ones in the lower rows.
Dressel pointed out that they were found just outside
the Servian walls and suggests that they were used as a
means of land reclamation to fill the ditch ( fossa) that
ran along the front of the walls in this stretch between
the Porta Collina and the Porta Viminalis.25
Following the model of the buried amphoras, the
technique of embedding the amphoras into concrete
seems to have started at ground level before work-
ing its way up to the vaults. For example, at Aosta
(Augusta Praetoria) in 1839, amphoras were found
built into the foundation walls of a large public build-
ing probably dating from the first century.26 Because
foundations were typically made heavier than the su-
perstructure, this use suggests that the technique was
more likely a means of discarding the pots and saving
on materials than any attempt to lighten the structure.
53. Diagrams showing typical uses of amphoras for land reclama-
Another example occurs in an enclosure wall of an or-
tion projects. Top: amphoras used to protect floors from water
infiltration. Middle: amphoras and sand fill placed in foundation chard/vineyard at Pompeii (I.20.5), where the upper
trench to increase stability and drainage, especially in unstable part of the partially fallen wall consists largely of long
soil. Bottom: amphoras placed horizontally in trench to create
drainage channels.
narrow amphoras set upright side-by-side (Fig. 54).27
Again, the use of the amphoras seems to have been
a means of saving on materials and disposing of the
Greek colony at Marseilles.21 It is found in Republi- pots rather than any type of technological innova-
can times in Italy for a variety of uses: to create stable tion. Both of these examples were variations on the
foundations for buildings and roads, to facilitate soil techniques of land reclamation cited earlier, but they
drainage, and to reduce moisture problems on the were steps in the process that led eventually to the
ground floor of buildings (Fig. 53). Examples can be use of the amphoras in concrete vaults in the second
seen throughout northern Italy in the Po valley and century.
in the Veneto,22 as well as at Fondi near Terracina.23 The use of the amphoras also has been explained as
Amphora deposits also have been found buried at an aid in the curing of large masses of concrete by cre-
Ostia apparently to facilitate drainage as early as the ating interior voids.28 From a modern point of view,
second or first century b.c.24 In Rome itself, the best- this explanation makes more sense for simple lime
74
P1: KNP/KIC
AMPHORAS IN VAULTS
Alberti was aware of the importance of water when

he noted that “ . . . it could be said that nothing is
better for a vault than for it to absorb plenty of wa-
ter and never go thirsty.”30 If the Roman builders’
main concern was to aid the curing process so that
the mortar would be stronger, the amphoras would
have been set upright so that they could be filled with
water, but they are typically set upside down. If the
main purpose of introducing the amphoras into the
mass of the concrete was to create a stronger mortar,
one must wonder whether the builders would have
risked introducing voids that could have weakened
the structure and defeated the original intention.
examples of amphoras in late

antique vaults
After the two second-century occurrences, amphoras
do not appear again in vaults from Rome until the late
third/early fourth century. From this period there are
54. Wall of garden enclosure at Pompeii (I.20.5) showing remains ten known examples, and half of these occur out-
of amphoras used to fill wall (first century a.d.). side the city walls (Appendix 2g). By far the most
common amphora types found are the Dressel 20
mortar than for pozzolana mortar. For simple lime and its successor, the Dressel 23. One example of an
mortars, which harden on contact with air, delayed Almagro 51c amphora has been found at the Basilica
hardening within a large mass is a phenomenon that of Maxentius. All three types were found together
can sometimes prevent the structure from acquiring on a mid-third-century shipwreck off the coast of
its full strength. The amphoras would introduce air France, which suggests that they were typically com-
into the concrete mixture, but the wall of the vessel ing to Rome together in the same shipments.31 The
also would act as a barrier preventing the absorption only example of African amphoras used in vaulting
of the carbon dioxide. The pozzolana mortar used occurs at the “Casa di via Giulio Romano” at the
in Rome, by contrast, was dependent on water to base of the Capitoline, where an arcade was added to
activate the chemical process that gave the material the front of the building, and its vaults were full of
its strength (see Chapter 3). The comments by Dio Africana 1 amphoras (Fig. 55). The vaults were not
Cassius in the third century noting that the mixture large, and the choice to use the long narrow form
of lime and pozzolana became petrified as long as may have been governed by the ability to fit them
it was kept in liquid suggests that by this time the into the available space.
importance of moisture for the curing process was The vast majority of amphoras in vaults from any
well understood.29 Certainly by the fifteenth century period are in the vaults of the circus at the Villa of
75
P1: KNP/KIC
Circus of Maxentius had simply used tufo giallo instead

of tufo lionato, the savings in weight would have been
6.95 percent. A combination of tufo giallo and scoria,
as in the Basilica Ulpia, would have given a 13.9 per-
cent weight reduction.35 A similar calculation for the
dome of the Mausoleum of Helena, which is perhaps
the most well known of all the examples of amphoras
in vaults (Figs. 42 and 47), reveals that the amphoras
(c. 180) would have reduced the weight of the vault
by only 1.33 percent, whereas using tufo giallo alone
or with scoria would have reduced the weight from
4.14 percent (tufo giallo) to 8.26 percent (mix of tufo
giallo and scoria). Again, the use of amphoras is the
least effective means of reducing the weight.
By contrast, the amphoras could have provided a
tangible reduction in the amount of the other ma-
terials necessary, that is, pozzolana, lime, and tuff.
For example, using six to ten thousand Dressel 23
amphoras36 at the Circus of Maxentius would have
resulted in only a 1.25–2.08 percent savings in the
amount of concrete required (roughly equivalent to
55. Detail of the balcony vault (third century a.d.?) added to the weight savings), but this represents 272–453 m3
the façade of the “Casa di via Giulio Romano” (early second of materials, which translates into 62–86 single-yoke
century a.d.) at the base of the Capitoline. Remains of Africana cartloads of lime, 600–1,000 of tuff, and 359–595 of
1 oil amphoras are visible in the vault.
pozzolana, for a total of 1,021–1,681 cartloads.37 In
Maxentius at the fourth mile of the Via Appia. The comparison to the total number of cartloads of ma-
vaults supporting the seating of this structure had terials arriving at the site this is miniscule, but in real
anywhere from six thousand to ten thousand Dres- terms of cost and labor savings it is not an insignificant
sel 20 and 23 amphoras built into them (Fig. 56).32 amount.
Since the Circus of Maxentius employs by far the The amphoras in the circus are visible only when
most amphoras in any one monument,33 I use it to parts of the vault have fallen away, so in areas in which
test the effect that so many amphoras would have the vault is better preserved, the presence of the am-
had on the project. Calculations for a one-meter phoras cannot be verified. In at least one section of
stretch of the vaulting, which could accommodate the monument, the area in which the amphoras are
eight Dressel 23 amphoras,34 reveal that the pots only visible corresponds to a section that can perhaps be re-
reduce the weight of the vault by 1.43 percent. The lated to a single building crew based on the change in
more common means of lightening a vault was to putlog holes where the ends of the wooden scaffold-
use lightweight caementa such as the tufo giallo della ing beams were lodged during construction (Fig. 56).
via Tiberina or scoria/pumice. If the builders at the So, some individual crews at the project may have
76
P1: KNP/KIC
AMPHORAS IN VAULTS
56. Circus of Maxentius (a.d. 306–312). Vault supporting seating. Arrows indicate amphoras. Lines
indicate nonalignment of putlog holes suggesting different building crews.
chosen to use the amphoras while others did not. If the weight is actually counter to the way Romans
so, the use of the amphoras could have had a more seem to have been thinking about their structures.
tangible effect on individual building crews work- The amphoras are usually located at the haunch of
ing within the structure. Moreover, if the amphoras a vault because this is typically the only place thick
are not distributed evenly throughout the monument enough to accommodate them, but this is precisely
but, rather, in sections belonging to individual work the place where the builders were adding weight to
groups, the figure of six thousand to ten thousand direct the internal forces down through the walls. For
given earlier would be an overestimate. The evi- example, in the Pantheon dome only the crown uses
dence for the density of the amphoras gathered by E. lightweight caementa; the haunch employs heavier cae-
Rodrı́guez-Almeida was based on the sections where menta of brick (Fig. 46, p. 62) and is made even heavier
the amphoras were visible, so the density calculations through the use of step-rings (Chapter 7). A structural
for individual stretches that certainly employ the am- analysis of vaults employing amphoras at the haunch
phoras remain valid. actually shows that the effect is slightly detrimen-
The idea that the Roman builders typically used tal to the stability though fundamentally negligible
the amphoras in the haunches of vaults to lighten (Chapter 8).
77
P1: KNP/KIC
57. “Temple of Minerva Medica” (first half of the fourth century a.d.). Sketch of the dome showing
locations of amphoras (dark gray) and pumice (light gray).
Building amphoras into vaults seems to have been because they occur selectively within the building and
originally intended as an efficient means of saving on were combined with other techniques such as lattice
processed materials and manpower while disposing of ribbing and pumice in the upper parts of the dome.39
nonbiodegradable containers, but their careful place- The placement of the amphoras over the windows
ment in two fourth-century structures, the “Temple suggests that the builders were using the amphoras as
of Minerva Medica” and the Tor de’Schiavi, suggests a means of lightening the area above the openings in
that this original intention developed into a more so- order to channel the weight to the solid areas on either
phisticated structural use. During the restoration of side, much in the same way a relieving arch might be
the dome at Minerva Medica, four amphoras were used (Chapter 5). Moreover, built into the vault above
found placed above one of the windows (Fig. 57). The two (and originally probably four) of the window
excavator, P. Caraffa, argued that the placement of the openings were other arch-covered “windows,” which
amphoras on axis above the windows was intentional really acted as relieving arches and which were filled
because the stairway along the extrados of the dome underneath with a mixture of pumice and mortar
had to be moved slightly off axis to accommodate further suggesting that the builders were concerned
them. A. Choisy also noted that he saw an amphora about the weight over the wall openings (Fig. 57,
embedded in the wall of Minerva Medica directly Pl. XI).40
over one of the window openings.38 These exam- The other unusual use of amphoras occurs at
ples at Minerva Medica are particularly informative the Tor de’Schiavi but surprisingly not in its dome.
78
P1: KNP/KIC
AMPHORAS IN VAULTS
58. Octagonal Hall at the “Villa of the Gordians” (early fourth century a.d.). Circular structure in
foreground and upper part of Octagonal Hall are parts of the medieval renovation of the ancient building.
In inset, white arrows indicate amphoras (Dressel 23) and black arrows indicate ribs. “A” marks one
amphora that was filled with mortar.
79
P1: KNP/KIC
Representations of the building during the eighteenth also employed a clever use of amphoras in a struc-
century show the remains of the pronaos vault where turally precarious situation. By contrast, the other
it connected to the rotunda.41 The most informative uses of the amphoras at the haunches of domes at
one is a painting by H. F. van Lint in which the tri- the Mausoleum of Helena, at the Octagonal Hall
angular gabled area where the pronaos vault attached at the “Villa of the Gordians” (Fig. 58), and of the
to the rotunda displays rows of round, orange im- semidome at the “Temple of Venus and Cupid” at
pressions (Pl. XII). Notes of Francesco de’Ficoroni the Sessorian palace (Fig. 131, p. 145) are actually in
from the mid-eighteenth century record that “the conflict with the established principles of dome con-
portico was covered by a vault constructed of ma- struction in which the haunches are made heavier
terial interspersed with empty vases in order to make than the crown in order to provide a surcharge above
it lighter.”42 The use of concrete vaulting for the the walls. As shown from the calculation at both the
roof of a pronaos is very unusual. Typically, gabled Circus of Maxentius and the Mausoleum of Helena,
roofs consisted of wooden beams, rafters, or trusses. the weight reduction was virtually negligible. This
If the Tor de’Schiavi had a concrete vaulted roof sup- would not have been lost on the Roman builders. I
ported by columns, it would certainly have needed would suggest that they chose to use the amphoras as
as much lightening as possible.43 Its triangular shape a cost saving measure in spite of the weight reduction
would have allowed for a higher percentage of space they caused, understanding well that the weight dif-
to be replaced by amphoras than was typical for other ference was minimal, whereas the material savings was
vault forms. J. J. Rasch calculated that 315 amphoras worthwhile.
would have been used, which would have resulted in Medieval builders imitated the Roman use of am-
an approximately 8 percent decrease in the weight of phoras in vaults, as is illustrated clearly at the “Villa
the vault.44 Certainly any possible weight reduction of the Gordians.” In the thirteenth century, the Oc-
would have been desirable in this precarious struc- tagonal Hall was converted into a tower. In the cen-
ture. In his publication of the monument, Rasch ter of the room, a circular pier was added to sup-
has sensibly suggested that some type of metal bars port an annular barrel vault that formed a floor at
may have been used in the entablature to help stabi- about the crown level of the original dome (Fig.
lize the structure, but no evidence has survived for 58).47 The fallen vaults of both the fourth-century
them.45 and the thirteenth-century buildings remain on the
From these findings, I would suggest that originally site and are easily distinguishable because the later
and in a majority of cases the amphoras were used as vaults have caementa of peperino (probably lapis Gabi-
a means of saving material and labor costs46 but that nus from the quarries at Gabii about 13 km fur-
by the fourth century they were occasionally used ther out the Via Praenestina; Map 3, p. 14). The
in more innovative ways that related to the stability broken remains of vaulting reveal that the medieval
of the building. The selective use of the amphoras builders imitated the use of the amphoras by employ-
above the windows at Minerva Medica follows the ing much smaller pots (33.5 cm high × 35.40 cm
logic established by the builders of the Pantheon in dia). This is an example of the medieval builders im-
the sense that they were being used to lighten particu- itating the ancient technique using their contempo-
lar parts of a vault in order to channel the weight away rary type of vessel. (See also Arch of Janus, Appen-
from openings. The pediment of the Tor de’Schiavi dix 1.35.)
80
P1: KNP/KIC
AMPHORAS IN VAULTS
conclusions not unknown on other types of amphoras, but they

never contain the same level of detail or occur so con-
The accumulated knowledge that first led to the use of sistently. The Dressel 20 is best known from the man-
amphoras in vaults can be traced back to their use made mound of broken amphoras at Monte Testaccio,
in land reclamation projects in which the amphoras 85–95 percent of which are Dressel 20 amphoras.50
were buried in the ground as a means of creating Monte Testaccio’s location near the Horrea Galbana,
a more stable surface on which to build (Fig. 53). where state-controlled oil was stored in the late sec-
The use of amphoras for land reclamation was even- ond century a.d., also suggests that much of the oil in
tually translated into concrete construction, first in the Dressel 20 amphoras discarded at Monte Testaccio
walls by the first century a.d. (Fig. 54) and later in belonged to the state (as opposed to private oil, which
vaults by the second century. The earliest examples was presumably also coming into Rome). A majority
of amphoras in vaults occur outside of Rome in the if not all of the Dressel 20 amphoras at Monte Tes-
Hadrianic period, one at the river harbor at Ostia and taccio bore the tituli picti.51 Moreover, the amphoras
the other in the Villa alla Vignaccia outside of Rome. with the tituli picti that have been found at land sites
The location of the amphoras in both the walls and outside of Rome all come from transshipment ports
vaults at the Villa alla Vignaccia suggests that they in Gaul or near military camps, again indicating that
were used wherever they fit regardless of location and the inscriptions relate to state control and distribu-
that they were not used specifically to lighten vaults. tion of oil to the urban populace of Rome and to the
The innovation in this case is not so much a structural military.52
one as is often assumed but, rather, one focused on Monte Testaccio is a unique monument that may
reducing costs and increasing efficiency. provide some insight into the eventual proliferation
Most of the amphoras used in vaults in and around of amphoras in the vaults of fourth-century Rome.
Rome were olive oil containers from Baetica in In spite of the preponderance of Dressel 20 amphoras
Southern Spain, either the Dressel 20 or its successor, at Monte Testaccio, excavations at Ostia reveal that it
the Dressel 23. The Dressel 20 is unique in that it of- is not representative of all the oil coming into Rome.
ten bears painted inscriptions, tituli picti, which record The contemporary deposits excavated at the Terme
information such as the weight of the oil contained, del Nuotatore at Ostia reveal a much higher pro-
the names of people who weighed and documented portion of Tunisian amphoras than found at Monte
the oil, as well as the names of the district where the Testaccio.53 This contrast suggests that the Baetican
oil was bottled. The amount and type of informa- amphoras were destroyed much more systematically
tion included indicates that the oil in the vessels was than the North African ones. Why then were the
under state control from the first century a.d. when Dressel 20 amphoras broken and discarded, whereas
the tituli picti first appeared on Dressel 20 amphoras. the North African amphoras apparently were not,
The oil would have been destined for the annona urbis or at least not so consistently? D. Mattingly points
(distribution to the people of Rome) or the annona out that the thick-walled Dressel 20 amphoras break
militaris (distribution to the army).48 This was cer- into large curved pieces whereas the thin-walled
tainly the case by the mid-second century, as attested African amphoras break more easily into smaller flat-
by inscriptions that relate to the import of Baetican ter pieces that could be used in construction, such as
oil to the praefectus annonae.49 Painted inscriptions are for cocciopesto.54 Furthermore, the long, narrow shape
81
P1: KNP/KIC
of the African amphoras was useful for making tubes probably another influencing factor. With the aban-
that could be used for drainage once the neck and donment of Monte Testaccio as a dumping ground,
base were removed. I would suggest one further pos- there would have been more whole Baetican am-
sibility is that Dressel 20 amphoras were destroyed sys- phoras available than in earlier times (even though
tematically at Monte Testaccio because they had the fewer Baetican containers were being imported than
tituli picti on them. These painted inscriptions acted previously).61
as official control marks, and the destruction of the The largest concentration of examples of amphoras
amphoras could have been to prevent illicit reuse of in vaults (Appendix 2g) occurs under Maxentius, and
the officially marked vessels.55 the phenomenon may well be related to the polit-
The majority of buildings using the amphoras were ical situation at that time. This was a period when
constructed after Monte Testaccio ceased to be used Rome seems to have been become more dependent
as a dumping ground for amphoras in the mid-third on oil from Africa, a trend which began under Sep-
century. The latest datable titulus pictus from Testaccio timius Severus, when evidence shows that his home
dates from between a.d. 261 and 266,56 and this province of Tripolitania began supplying oil to the
is the latest known from any other context. How- city of Rome as thanks for benefits conferred by
ever, Baetican olive oil continued to come into the Severus.62 However, during the period a.d. 308–310,
city. The Dressel 20 amphoras may have continued the usurper Domitius Alexander had taken control of
to be used without the inscriptions,57 and the pres- the oil and wheat producing areas of Africa, which
ence of Dressel 23 amphoras, which eventually su- may well have prevented African oil from coming
perceded the Dressel 20, in Rome during the fourth into Rome. The Chronographer of a.d. 354 noted
and fifth centuries shows that Baetica was still produc- that there was great hunger in Rome at this time and
ing for the capital city, although probably in reduced that Maxentius had to levy taxes, presumably to pur-
quantities.58 The sudden abandonment of tituli picti chase supplies from alternative sources.63 J. T. Peña
and of Monte Testaccio as a dumping ground sug- has suggested that the large number of Spanish oil
gests a major disruption in the state control of oil amphoras found in the vaults of this period may have
coming into the city of Rome during the 260s. The been because of a shift in the supply source after the
most likely explanation is that the supply of Spanish takeover in Africa, with greater pressure on Baetican
oil was cut under Gallienus when one of his generals, producers to make up for the losses.64
Postumus, rebelled and established his Gallic Empire, In terms of cultural acceptability, the reuse of am-
which included Spain.59 phoras in vaults is part of a general trend of reuse
That the majority of the examples of amphoras in late antiquity that is most visible in the phe-
in vaults occur after the disappearance of the tit- nomenon of using spoliated marbles. Even with the
uli picti and the abandonment of Monte Testaccio reorganization of the brick industry under Diocle-
cannot be accidental. We know that state-sponsored tian reused bricks were typically employed in the
oil distributions in Rome continued into the fourth wall facing of major imperial monuments. The an-
century.60 The sudden proliferation of the technique chor blocks for the tie bars at the Baths of Diocletian,
could simply be due to evident need: Building materi- discussed in Chapter 6, also were made from spoli-
als were in shorter supply than earlier, so any method ated marble fragments. This phenomenon of reuse
of material savings was useful. Economic possibility was became problematic when people began spoliating
82
P1: KNP/KIC
AMPHORAS IN VAULTS
59. Santa Maura outside of Rome (late fourth/fifth century a.d.). Remains of apse showing Dressel
23 amphoras used at crown of vault.
public monuments and especially tombs for build- spite of the objections by some authorities, spoliation
ing materials, as represented in the pronouncements and other types of reuse had become an acceptable
against such practices in the Theodosian Code.65 The practice among builders.
reuse of amphoras was much less controversial, as no The most interesting examples of the use of am-
other structures were damaged, and as indicated ear- phoras in vaults are those in which a technique with
lier it fits into a long tradition of reusing amphoras an apparently prosaic beginning was translated into a
for land reclamation projects. It does, however, re- structural innovation, as at the “Temple of Minerva
flect changing attitudes in late antiquity. During the Medica” and the Tor de’Schiavi. The common as-
late first and early second centuries, the use of the am- sumption that the amphoras were used to lighten the
phoras as filler material would not have been seen as vault is probably not what the builders usually had in
acceptable for a major imperial monument. A certain mind, but in these two cases there may be some truth
level of “conspicuous consumption” in construction in it, and if so it represents the evolution of an inno-
was part of the imperial image during the high em- vation from one related to efficiency to one related to
pire and obvious reuse was not part of the mentality. structural considerations. In most situations, the use
The fact that the early examples of the amphoras do of the amphoras for lightening vaults was not an effec-
not appear in imperial monuments is an important tive means of accomplishing this task and could even
one. Only after political and economic circumstances have had a slightly detrimental effect on the stability
changed did the attitude start to shift toward an ac- of the structure depending on where they were placed
ceptance of a reuse policy in official monuments, as (see Chapter 8). In these cases, however, both of
so clearly represented in the Arch of Constantine. In which are structurally innovative buildings, the careful
83
P1: KNP/KIC
60. Plan and section of St. Jerome in Cologne, Germany (fourth century a.d.). Section shows the
location of Dressel 23 amphoras used in the haunch of the semidomes of the apses. Plan based on
Gerkan 1951: Abb. 3–4.
placement of the amphoras suggests a level of sophis- during this period as a means of reusing old material
tication that was not present in earlier examples. and saving on new, as is demonstrated by the ones
The latest datable use of Dressel 23 amphoras in found in the walls (as opposed to the vaults) of San
vaults around Rome is at the Mausoleum of Helena Giorgio (a.d. 1050–1150) at Riofreddo (near Tivoli
(a.d. 326–330), although two other later examples to the east of Rome).68
show that both the technique and the amphora type In the case of the amphoras used in Roman vaults,
continued. One is an unidentified globular type, the building technique seems to have spread with the
probably a Dressel 23, in an Honorian tower of the containers. Baetican oil was distributed to the army,
Aurelian Walls at Porta Asinaria (a.d. 401–403),66 and and Baetican amphoras are often found in concentra-
the other is a fourth-fifth century (?) example in the tions near army outposts, such as along Hadrian’s Wall
semidome of the remains of the Church of Santa and along the Rhine in Germany.69 During World
Maura, now located in the back garden of a hotel (Fig. War II, many of the churches along the Rhine were
59). In this latter example, the Dressel 23 amphoras are damaged by Allied bombing raids, and in one case the
located at the crown of the vault, where they would bomb damage revealed that Dressel 23 amphoras had
have the greatest structural effect. When the tech- been built into the vaults of the mid-fourth-century
nique reappeared in Rome during the medieval pe- Church of St. Jerome in Cologne.70 The Dressel 23
riod at a tower of the Porta Latina (c. a.d. 1157), at amphoras occur in the semidomes of the apses (4.8–
the cloister of San Paolo Fuori le Mura (a.d. 1200– 7.4 m dia) and would have numbered about 125 max-
1250), and at the arched entranceway to San Anas- imum (Fig. 60).71 The presence of the Dressel 23 am-
tasio at Tre Fontane (a.d. 1244), the vessels were no phoras in Cologne can be explained by the fact that
longer transport amphoras but rather household pots, Constantine had established a fort there to protect
most of which were damaged with use or were kiln a bridge over the Rhine, and it was supplied with
clinkers.67 The technique probably also was employed oil from Baetica as were many of the other outposts
84
P1: KNP/KIC
AMPHORAS IN VAULTS
along the Rhine. Remesal Rodrı́guez has argued that them in place, which accords with the archaeological
with the reforms of Diocletian much of the Baetican evidence of wasters and broken crockery, but his ex-
oil was redirected to the armies in the West rather planation that the cracks were placed deliberately to
than toward Rome.72 prevent water from collecting and adding weight,
When the medieval builders later imitated the an- even if it was correct for his own time, is in no way
cient Roman use of amphoras in vaults, they may well applicable to the ancient Roman use of transport am-
have interpreted it as a lightening technique used by phoras. Alberti was writing in the fifteenth century,
the Romans. When Alberti spoke of the recent prac- over one thousand years after the Romans were using
tice, which he had evidently witnessed, of placing the amphoras in vaults in Rome, but his explanation
water vessels in the haunches of the vaults, he was has been taken up and applied to the Roman exam-
relaying the beliefs and motivations of his own time. ples, which were quite different in both context and
He noted that the pots were cracked before putting morphology.
85
P1: JPJ/JPK/KDF
5
VAULTING RIBS
or to reinforce the concrete,3 whereas others have

S tone or brick arches built into roman
concrete vaults are called “ribs,” but unlike the
ribs in Gothic vaulting the Roman ones are usually
assigned them a more active role in the channeling
of loads through the hardened mass of the concrete
flush with the intrados and would not have been vis- even after the curing of the mortar.4 Both groups ac-
ible once the vault was decorated. Vaulting ribs often knowledge that ribs would have played a useful role
have been discussed in works on Roman construc- in controlling the concrete during construction. C. F.
tion, yet they have rarely been studied systematically Giuliani gives a more complex explanation and points
in terms of form, use, and context.1 Many of the most out that ribs were multipurpose elements that must
authoritative scholars on the subject were writing at be evaluated according to the individual context,5
a time when a number of important monuments had and this approach I find most useful. In the follow-
not been properly dated or, in some cases, even exca- ing discussions, I examine specific examples to show
vated. Recent studies on individual monuments such how the use and form of vaulting ribs changed over
as the Colosseum, the Forum of Caesar, Trajan’s Mar- time.
kets, the Baths of Caracalla, and a series of late Roman
domes need to be put into a developmental context
early development of vaulting ribs and
before we can understand why the ribs were used and
relieving arches
how they changed over time.2 The development of
the ribs in vaulting is closely connected with the use Both relieving arches and vaulting ribs are variants of
of relieving arches over openings in walls, so the two cut stone arches and vaults, which date back to the late
techniques are discussed together. sixth century b.c. in Rome.6 Relieving arches differ
In the past, scholars of Roman architecture have from regular arches in that they are filled with walling
disagreed about the role that ribs and relieving arches material directly underneath their intrados. They of-
play within the fabric of the concrete structure. Some ten occur over an opening, which may or may not
have asserted that once the concrete cured, the re- be visible, such as a window, door, or drain. Both the
lieving arch or rib became part of the hardened mass vaulting rib and the relieving arch were used to direct
and no longer acted independently to divert loads loads away from openings. The beginnings of this idea
86
P1: JPJ/JPK/KDF
VAULTING RIBS
of tuff blocks to support the front cella wall,8 and

in the Augustan reconstruction of the same temple,
the concrete podium was encased in a tuff wall with
travertine blocks inserted at the points underneath
the columns.9 Once travertine became a common
building material in Rome during the first century
b.c., it often was used in place of tuff blocks at points
of concentrated loads, such as for the keystones and
impost blocks at the Forum of Caesar, the Forum of
Augustus, and the Theater of Marcellus (Fig. 63). Ro-
man builders of the Republic developed a hierarchy
of materials going from travertine (the strongest) to
concrete (the weakest). This idea that different types
61. Acropolis gate in polygonal wall at Alatri, Italy (fourth/third

century b.c.). Outlined stones are arranged in voussoir fashion to
form relieving arch over lintel of doorway.
can be seen in the fourth–third-century b.c. polyg-

onal walls at Alatri, where one of the doorways in
the acropolis walls has its lintel covered with a rudi-
mentary relieving arch of polygonal stones (Fig. 61).
Another more developed example occurs over the
arched opening of the Porta Rosa at Velia (Fig. 62),
which probably dates from the third century b.c.7
The idea of using arches to control how a structure
supports its load is related to the idea of reinforcing the
parts of a building that support the greatest loads with
the most durable materials. For example, at the Tem-
ple of Castor in the Forum Romanum (117 b.c.), the
concrete foundations were “reinforced” with a wall 62. Porta Rosa in city walls at Velia, Italy (third century b.c.[?])
with stone relieving arch over gate opening.
87
P1: JPJ/JPK/KDF
the Sanctuary of Hercules Victor and the Theater of

Marcellus the builders were dealing with situations in
which the plan of one level did not coincide with
the one below it. At Tivoli the issue was solved with
the locally available travertine from nearby quarries,
whereas in Rome it was solved with the new tech-
nique of using brick or tiles for walls.
the vaulting ribs at the colosseum

In Rome itself, the earliest datable examples of vault-
ing ribs occur at the Colosseum, and they demon-
strate that vaulting ribs could be used for different
purposes within the same structure from an early
date. The ribs take three different forms (Fig. 67):
(1) arches of travertine voussoirs, (2) arches of radi-
ally laid bipedales acting as voussoirs, and (3) arches of
radially laid bipedales separated at the two edges by par-
tial bipedales so as to create compartments filled with
mortar and caementa (ladder rib). Appearing later, by
the early third century a.d., is a fourth type of rib,
63. Theater of Marcellus (17 b.c.). Inner wall of ambulatory con- consisting of a series of ladder ribs built next to each
structed of tuff blocks with inserts of travertine for imposts and
keystones of arches. other (lattice rib).
The travertine ribs at the Colosseum are located in
the vaults of the arena substructures and are, therefore,
of materials serve special purposes within the struc- the first type to have been built. They are not found
ture is the basis for the vaulting rib as an independent higher up in the building.13 The superstructure was
element. constructed on concrete foundations into which
One of the earliest uses of vaulting ribs occurs at were built service rooms along the main longitudinal
the Sanctuary of Hercules Victor at Tivoli where axis. Because four of the radial walls at ground level
travertine ribs were systematically used throughout were each built directly onto the crown of one of
the complex to reinforce the vaults (10-m span) where the substructure vaults, the builders provided the
walls ran above (Fig. 64).10 The earliest examples travertine ribs as reinforcement (Fig. 68). The situa-
of the use of relieving arches to protect vaults oc- tion is similar to the one at the Sanctuary of Hercules
cur in the Augustan period and are constructed of Victor at Tivoli discussed earlier except that the
bricks or tiles.11 A systematic use occurs at the The- Colosseum walls run along the crown of the vault
ater of Marcellus (17 b.c.) in the brick wall of the parallel to its axis instead of across it perpendicular to
“Passaggio dei Cavalieri,” where the relieving arches the axis. To account for this, the travertine ribs were
“jump” over the vaults below (Figs. 65–66).12 In both set into the vault (3.3–3.9-m span) so that they occur
88
P1: JPJ/JPK/KDF
VAULTING RIBS
64. Sanctuary of Hercules Victor at Tivoli (mid-first century b.c.) showing location of travertine ribs
underneath wall at upper level.
65. Theater of Marcellus (17 b.c.). Relieving arches in “Passaggio dei Cavalieri” occur over vaults at
lower level.
89
P1: JPJ/JPK/KDF
66. Section through the Theater of Marcellus (17 b.c.).
under the ends of the brick wall, which has a reliev- bipedalis ribs occur in the level-2 vaults of both the
ing arch built into it so that the ends align with the inner and outer ambulatories. Their location always
travertine ribs below. Both the travertine ribs and the corresponds to a set of double stairways that were
relieving arches have earlier precedents, but the idea built directly above at level 3 (Fig. 70).14 Their care-
of combining them into a network was a new inno- ful placement underneath the stairs implies that the
vation. The combination of the two techniques also builders intended them to have a structural role in
shows clearly that they were conceived as structural supporting the staircase either during or after con-
devices intended to transfer loads along predefined struction.
routes within the structure. The third type of rib, the ladder rib, appears to have
Like the travertine ribs, the ribs of bipedales in the had no obvious load-bearing role. These ribs always
vaults of the superstructure of the Colosseum also had occur in the vaults supported by the radial walls un-
a structural purpose. The building was surrounded by der the cavea (Fig. 69). They are located consistently
two concentric ambulatories that ran along the outer in the same place where they connect the inner of
edge of the building at levels 1 and 2 (Fig. 69). The the two travertine piers built into the radial walls of
67. Four types of vaulting ribs at the Colosseum.
90
P1: JPJ/JPK/KDF
VAULTING RIBS
along the intrados of arches over door and window

openings, but it did not appear again in the surviving
remains of vaulting in Rome until the mid-second
century. It seems to have been used to aid in the con-
struction of this one very complex situation and then
temporarily abandoned.
The idea of the independent vaulting rib devel-
oped in cut stone construction and was then translated
into concrete construction using brick. The Roman
builders clearly used the travertine and bipedalis ribs
as elements of reinforcement and thought of them
as channeling loads through predetermined routes.
The use of the ladder ribs as constructional aids may
have been a secondary phenomenon, but it was one
with an early history that developed alongside the
more structural uses of the element. Both the The-
68. Colosseum (a.d. 70–80). Drawing showing combination of
travertine ribs and brick relieving arches. ater of Marcellus, which has the earliest examples of
relieving arches over vaults, and the Colosseum are
highly complex building types, and the builders used
tuff blocks. In each case, the ladder ribs spring from a degree of ingenuity in dealing with issues of struc-
travertine blocks built into the haunch of the vault.15 ture and construction rarely encountered in earlier
Some seem to serve the practical function of support- buildings.
ing a landing or providing a transition between vaults
of different inclinations. Others do not support any
other structure above and were probably used for a
the use of solid brick (bipedalis) ribbing
constructional rather than a structural purpose. These
ladder ribs are constructed differently from the ones In later Flavian architecture, the solid bipedalis ribbing
that appear in later architecture. The “rungs” are two continued to be used, but it had more varied pur-
bipedales wide separated by arches of broken bipedales poses. Examples of ribs at the Domitianic Vestibule
at the outer ends of each bipedalis (Fig. 71). In this leading up to the Palatine are noteworthy because un-
case, each side of the rib could have been built inde- like the Colosseum examples these bipedalis ribs have
pendently. Later ladder ribs from the third century are no obvious dead load supported above (Fig. 72). The
usually only one bipedalis wide. The coordination of Domitianic Vestibule was built up against the earlier
the construction of the stair vaults would have been Horrea Agrippiana and consisted of a large hall (31 ×
a very complex endeavor, as wooden centerings with 21 m) that led into a smaller open area giving access
complex forms were required at different levels. The to a long ramp up to the Domus Tiberiana on the
ribs may have been used to form convenient stopping Palatine. A series of two-story, wedge-shaped rooms
places so that the vault could have been built in stages. ran between the large hall and the horrea building and
This type of ladder construction was sometimes used provided support for the very tall wall of the Vestibule.
91
P1: JPJ/JPK/KDF
69. Colosseum. Section and partial plan showing location of ribs. Shaded areas on plan are reconstructed
after the fire of a.d. 217.
92
P1: JPJ/JPK/KDF
VAULTING RIBS
70. Colosseum. Section through ambulatories and mezzanine corridor showing change in ribbing
between original and reconstructed parts at bay 34. Ribs (both original and reconstructed) always occur
underneath double stairs at level 3.
71. Colosseum. (a.d. 70–80) Detail of Flavian ladder rib at bay 37. Note that it is two bipedales wide
(outlined on photo) rather than one bipedalis wide as in third–fourth-century ladder ribs.
93
P1: JPJ/JPK/KDF
72. Domitianic Vestibule (a.d. 81–96). Axon showing bipedalis ribs in vaults between the Vestibule and
the Horrea Agrippiana.
Only the largest of the wedge-shaped rooms contains bipedales. The lower courtyard of the complex was
vaulting ribs. It also is the only one with vaults run- built up against the slopes of the Palatine and formed
ning parallel to the large hall; the others, which run the substructures for the level built on the crest of
perpendicular to it, have no ribs. Because a vault is the hill (Fig. 73). The vaults with the ribs occur in
most effective as a stabilizing element when its axis rooms that formed the platform for the upper level.16
is oriented perpendicular to a wall (see Chapter 7), They cannot properly be called “ribs,” as they cover
the addition of the three ribs in the upper vault may the entire intrados of the vaults, but the construction
have been prompted by the desire to stiffen it and is the same. The location of these brick vaults sug-
to increase its effectiveness as a type of buttress for gests a purpose related more to the construction pro-
the extraordinarily high south wall of the large hall. cess that to long-term structural considerations. One
Whether the large hall was intended to be vaulted is advantage of building vaults of radially laid bricks is
unclear, but if so it was never completed. that they require less mortar than a concrete vault,
At the Domus Augustana, another of Domitian’s thereby shortening the curing process. Because the
building projects on the Palatine, some of the vaults bricks acted as voussoirs, the vaults could have taken
of the substructures are built entirely of radially laid loads at an earlier stage in the building process, thereby
94
P1: JPJ/JPK/KDF
VAULTING RIBS
than to reinforce the vault against external loads.18

These ribs are significant, as they show that the same
type of experimentation undertaken in the Domi-
tianic work in Rome also was occurring outside the
city, albeit in the context of imperial building, and
one can surmise that the builders were ones who had
worked on imperial projects in Rome. The substitu-
tion of roof tiles for the bipedales used in the city is
particularly noteworthy. Presumably this was because
of the difficulty of transporting the bipedales, typically
manufactured in the Tiber valley north of Rome, to
the Alban Hills to the south, which had no water
73. Domus Augustana (a.d. 81–96). Plan of lower level showing

location of bipedalis ribbing in vaults.
reducing the construction time and allowing work at

the upper level to progress sooner.
The development of brick ribs took place largely
in the city where the bricks were easily accessible via
the Tiber, but a notable example of ribbing also oc-
curs outside of Rome in the cryptoporticus (7.45-m
span) of the Villa of Domitian at Castelgandolfo in
the Alban hills (Fig. 74). The cryptoporticus served
as a buttressing element for the terracing of the hill-
side site. It is preserved for 120 m, and it originally
ran for another 200 m. One preserved section of the
vault is decorated with coffers while the rest of the
vault was plain. The ribs, which employ broken roof
tiles, occur at regular intervals between windows in
the plain section of the vault (Figs. 74–75).17 They are
unlikely to have supported structures above and prob-
ably relate to the construction process of this very long
vault, which would have been built in sections. The
coffered portion of the vault is reinforced with ribs
every fourth coffer, but here the ribbing only occurs
in the upper section (Fig. 76), again suggesting that
they were intended to aid in the construction rather 74. Villa of Domitian at Castelgandolfo (a.d. 81–96). Sketch of
a portion of the cryptoporticus.
95
P1: JPJ/JPK/KDF
structural value. This example is unique, but it is a

precursor to the small upright bricks often found in
vaults with brick linings along the intrados (visible in
Fig. 19, p. 31). In this case, the bipedales could
have formed containment barriers that allowed the
builders to build the barrel vaulted sections first so
that the cross vault and the semidomes at each end
could be added later.
The most elaborate use of brick ribbing from the
Trajanic period occurs in the Trajanic latrine added
to the Forum of Caesar. It was built above a series
of preexisting parallel walls of tuff and consisted of
a semielliptical outer wall of brick with travertine
corbels supporting the seating, which was covered
by a colonnaded portico (Fig. 78). The floor of the
structure was raised on suspensurae to provide room
underneath for the necessary drainage, which could
not be buried as was typical for a latrine at ground
level. To account for the incongruity in plan of the
original parallel walls and the new semielliptical wall
and colonnade above, the builders rebuilt the vaults
of the lower rooms so that they contained a series
75. Villa of Domitian at Castelgandolfo (a.d. 81–96). View of the of bipedalis ribs that were carefully placed to coincide
cryptoporticus showing ribbing and filled “windows” in haunch. with the ends of relieving arches of sesquipedales built
into the curving wall above (Figs. 78–79).20 Further
transport; however, some bipedales were used in the precautions were taken by carving each of the traver-
Domitianic work at the villa. tine column supports in the form of a trapezoidal
By the Trajanic period, brick ribs were commonly impost block to accept shallow relieving arches of
employed in Rome, sometimes used in a structural sesquipedales built into the low wall under the colon-
manner to support a wall above and other times used nade. The result was a complex structural network of
apparently to aid in the construction process in some ribs and relieving arches directing the loads of the la-
way.19 An unusual type of ribbing occurs in the trine to the tuff supporting walls of the lower rooms.21
remains at Trajan’s Baths (section E on Pl. VI) where The care taken to coordinate the ribs and relieving
a series of bipedales are set vertically in the mortar arches suggests that the builders intended for the load
(Fig. 77). They occur in a narrow section of barrel of the structure to be channeled through predeter-
vaulting that separated a central cross vault from the mined routes toward the tuff walls below.
semidomes over apses at either end of the room. These The combination of relieving arches and brick
were probably used to control the concrete during ribs at the Trajanic latrine foreshadowed the struc-
construction and would have had little permanent tural system of the Pantheon, which was begun about
96
P1: JPJ/JPK/KDF
VAULTING RIBS
76. Villa of Domitian at Castelgandolfo (a.d. 81–96). View of coffered section of the cryptoporticus.
Arrows indicate partial ribbing at crown.
six years later. The Pantheon is best known for its in Roman architecture. The rotunda wall is 6 m thick,
massive concrete dome, but the walls that support the but it is pierced with voids so that structurally it acts
dome and transfer its weight to the ground are among more like eight large piers than a solid wall as it appears
the most sophisticated yet elegantly simple structures from the exterior. The key to the structural integrity
of the rotunda is the series of vaulting ribs built into
the rotunda wall (Fig. 80). This is where the differ-
ence between vaulting ribs and relieving arches be-
comes rather ambiguous. The relieving arches visible
in the wall in Figure 81 are actually the ends of vaults
built of radially laid bipedales that extend all the way
through the wall in most cases (Fig. 46, p. 62). There
are two systems of arches at work in the rotunda wall:
the major arches (11.8-m span), which span between
the eight piers and cover the niches visible on the in-
terior, and the minor arches (5.35-m span), which
are contained within the hollow piers themselves.
On the interior walls are a series of smaller reliev-
ing arches supported on travertine impost blocks that
are intended to transfer the loads away from the archi-
traves and onto to the columns of the niches (Fig. 80).
77. Baths of Trajan (a.d. 104–109). Detail of construction of vault Like the system at the Trajanic latrine, the one at the
using upright bipedales at section E.
97
P1: JPJ/JPK/KDF
inspired by him. The conception of the Pantheon

took a remarkable vision on the part of both patron
and architect, but the techniques that made it possible
had been in the making for some forty years.
After the Hadrianic period, the use of solid bipedalis
ribs became less common, and the last major exam-
ple occurs at the Baths of Caracalla, where they were
used in conjunction with the newly developed lattice
ribbing, which is discussed further later. The solid
bipedalis ribbing at the Baths of Caracalla occurs in
three main areas: (1) the vaults connecting the piers
of the caldarium at two levels; (2) the vaults connect-
ing the piers of the tepidarium/frigidarium/natatio;
and (3) the inner parts of the vaults covering the side
rooms of the apodyteria (rooms 3a, b, c, d) (Pl. XIII).
In all cases they consist of two concentric rings of
bipedales. DeLaine, in her study of the construction
of the Baths of Caracalla, suggests that the choice to
build the vaults using the radially laid bipedales was
related to their role of connecting piers that support
78. Trajanic latrine at the Forum of Caesar (a.d. 106–113). Draw-
ing showing relationship between ribs in vault and relieving arches a larger vault at a higher lever, such as the frigidarium
in wall. cross vault and the caldarium dome. She reasons that
this technique would have reduced the curing time, as
Pantheon was carefully designed to control the loads the amount of mortar used in between the bipedales is
placed on the supporting structure by the massive much less than that used for typical concrete vaults. In
dome. this case, having the vaults gain their strength quickly
The similarity in conception between the Trajanic would have been a great advantage when the vaults
latrine and the Pantheon can be seen in the combined were erected at a higher level.24 This idea may be one
use of relieving arches and vaulting ribs and in the that was inherited from the Pantheon builders, who
trapezoidal impost blocks that channeled loads from also used the solid brick vaulting to connect the piers
column to column (Figs. 78 and 80). The latrine was supporting the dome there.
part of the renovation for Trajan’s Forum, which was
one of the projects of Apollodorus of Damascus,22
ladder and lattice ribs in barrel vaults
who also has been suggested as a potential candidate
for the architect of the Pantheon.23 There is no di- Although early examples of ladder ribs appear in the
rect evidence to associate Apollodorus with either Flavian parts of the Colosseum, this type of rib, along
the latrine or the Pantheon, but given the imitative with the related lattice rib, did not become common
nature of Roman design one can legitimately think until over a century later. A. Choisy proposed that
of a common mentality at work, either his or one the lattice and ladder ribs were used as a framework
98
P1: JPJ/JPK/KDF
VAULTING RIBS
79. Trajanic latrine at the Forum of Caesar (a.d. 106–113). View showing travertine impost blocks
that supported columns of portico and provided springing point for relieving arches over vaults below.
Dashed lines indicate original configuration.
intended to take part of the load off the vault so that projecting ribs in a long barrel vault (4.7-m span) of a
a lighter centering could be used.25 G. Cozzo coun- cryptoporticus (A on plan in Fig. 82 and Fig. 83). The
tered Choisy by arguing that ribs were not load bear- ribs correspond to buttressing piers along the outer
ing or statically necessary but, rather, acted to com- wall placed about every 4.5 m.29 They are unusual in
bat settling and to distribute the weight more evenly that the outer bricks do not seem to have gone into
along the impost.26 C. F. Giuliani has presented the the core of the vault, and the intrados of the ribs is
modified view that this type of cellular ribbing would crossed only occasionally by a bipedalis (the rung of
have created joints at the surfaces of each compart- the ladder). Given such construction, the ribs did not
ment like those between adjacent voussoirs and could actually form voussoirlike compartments. This barrel
have been intended to offer both structural and con- vault is one of the few in the complex that was not
structional advantages according to the situation.27 built with a brick lining along the intrados, and these
After the initial appearance of ladder ribs at the unusual “psuedo ladder ribs” were probably used to
Colosseum,28 two forms of ladder ribbing reappeared divide this long vault into a shorter sections that could
in barrel vaults in the mid-second century outside of be built one at a time (much like the bipedalis ribs
Rome at the Villa di Sette Bassi on the Via Latina. of the cryptoporticus at the Villa of Domitian). The
The villa was built in three phases, and the ribbing second type of ladder rib at Sette Bassi is similar in
occurs in the substructures of a bath building from the form and context to the ones used at the Colosseum
third phase (a.d. 140–150). The first type is anomalous a century earlier. They occur at the juncture between
and does not reappear again. It consists of a series of barrel vaults connected to the lunettes of a cross vault
99
P1: JPJ/JPK/KDF
80. Pantheon (a.d. 118–128). Drawing showing the system of ribs built into the rotunda wall. “M”
indicates the major system of ribs connecting the piers, and “m” indicates the minor system within the
piers.
forming the substructure of the bath (B on plan in Severan substructures of the Severan Baths on the
Fig. 82 and Fig. 84). As at the Colosseum, the cross Palatine (Figs. 85–86) and in the Baths of Caracalla
pieces consists of two bricks, which would have al- (Pl. XIII). DeLaine has shown that in at least one
lowed them to be built in two halves. case at the Baths of Caracalla, between rooms 9–
Examples of ladder and lattice ribbing appear 10w, a clear break occurs between the rib and the
within Rome in the Severan period when they were adjacent mortar, and she argues that the ribbing was
often used to separate adjacent vaults and to divide used to divide the vaults so that they could have
long vaults into sections. The ladder ribs are always been built in stages and to ensure the rigidity of
a single bipedalis wide, unlike the examples discussed the end supports while the vault was being laid.30
earlier. Ladder and lattice ribs occur throughout the This explanation would coincide with the evidence
100
P1: JPJ/JPK/KDF
VAULTING RIBS
centering (Fig. 87). These partial bipedalis ribs always

occur together in pairs as if they were conceived as
ladder ribs, but the two sides are not connected by
a cross piece in the preserved parts. The corbels and
the ribs are clearly related, as both occur together at
unequal spacing (c. 2 m and 2.8 m) in order to avoid
being located over the openings in the wall below. In
each case, the arch of one side of the “ladder” aligns
with the corbel. This is a unique combination of two
types of evidence (ribs and corbels) that implies that
in some cases the ladder ribs were probably connected
with the placement of the centering frames. In this
case, the unaligned halves of the “ladders” were al-
ways in the section of the widest spacing, suggesting
that they were intended to regulate the load between
the wider spaced centering frames by acting as a type
of relieving arch over the formwork boards. A com-
parison with the three-arch lattice ribs in the barrel
vaults at the Severan Baths on the Palatine (Figs. 85–
86) shows that similar spacing was used (i.e., between
1.8–2.5 m center to center), which suggests that the
function there may have been the same.
A particularly informative example of lattice rib-
bing occurs at the Colosseum in the vaulting that
was rebuilt under Alexander Severus after the build-
81. Pantheon (a.d. 118–128). Exterior view showing the ends of ing was damaged by a fire in a.d. 217. When the
bipedalis ribbing, which appears as relieving arches on rotunda outer ambulatories on the north side of the building
wall.
were rebuilt, lattice ribbing was used to replace the
from the unusual ribs in the cryptoporticus at Sette solid bipedalis ribbing discussed earlier. A clear ex-
Bassi. ample occurs in the outer ambulatory vault, where
Another explanation for some examples of rib- the reconstructed lattice rib has been built right up
bing is that they related to the centering structure, to the remaining parts of the original bipedalis rib
and evidence for this can be seen in a vault at (Fig. 88). The lattice ribs perhaps allowed the builders
the Nymphaeum Alexandri (“Trophies of Marius”), to reduce the number of bipedales, which must have
which was a fountainhead built by Alexander Severus been more costly than the mortar and rubble used
and is now located in the Piazza Vittorio Emmanuele. to fill the interstices. The lattice ribs were perform-
The ribs, partially preserved in a barrel vault at the ing the same function as the original bipedalis ribs,
front of the monument, are related to a series of but they represent a later and more developed form
corbels that would have been used to support the of rib.
101
P1: JPJ/JPK/KDF
82. Villa di Sette Bassi (a.d. 140–150). Vaulting scheme of substructure of bath complex showing locations
of ribs (after Lupu 1937: fig. 31).
A comparison of the reconstructed ribbing

at the Colosseum with that at the Baths of Caracalla
is particularly revealing. The choice to use solid
brick ribbing together with ladder and lattice type
ribbing at the Baths of Caracalla suggests that at this
time the different types of ribbing were each seen to
provide unique advantages, whereas at the somewhat
later Colosseum reconstruction the lattice ribbing
was used to replace bipedalis ribbing implying that
the builders considered them equivalent. The two
situations represent different attitudes toward ribbing,
but they may well reflect different economic climates
as well. The reconstruction of the Colosseum was
not planned but, rather, was a response to a disaster,
and it was at a time when the brick industry seems
to have undergone a contraction after the murder of
Caracalla.
After a hiatus of major building projects during
the mid-third century, the Baths of Diocletian (a.d.
298–306) represent a rebirth of large-scale imperial
construction in Rome and with it came the fully
developed approach to lattice ribbing. Once again a
comparison with the ribbing at the Baths of Caracalla
83. Villa di Sette Bassi (a.d. 140–150). Detail of rudimentary lad- is useful. All ribs used in any form of vault at the
der rib in cryptoporticus. Marked A on plan in Fig. 82.
102
P1: JPJ/JPK/KDF
VAULTING RIBS
84. Villa di Sette Bassi (a.d. 140–150). Detail of ladder rib dividing barrel vault from cross vault in
substructures of bath complex. Marked B on Fig. 82. Bipedalis “rungs” are outlined.
Baths of Diocletian are ladder or lattice ribs (Pl. XIII). of attitude in the intervening years. I see the ini-
The few barrel vaults used in the building cover the tial change occurring first in the reconstruction of
niches located to either side of the frigidarium and the Colosseum vaults because of economic pressures.
were built entirely of lattice ribbing. The same vaults When large-scale building began again under Dio-
at the Baths of Caracalla are among the ones sin- cletian, the new attitude continued. The lattice rib-
gled out for bipedalis ribbing, suggesting a change bing continued to be the most economically feasible,
85. Vaulting scheme for substructures of the Severan Baths (a.d. 193–211) and later the Baths of Maxentius
(a.d. 306–312) on southeast corner of the Palatine showing locations of vaulting ribs.
103
P1: JPJ/JPK/KDF
86. Severan Baths on the Palatine (a.d. 193–211). Shaded zones indicate three-arch lattice ribs in barrel
vault. Remains of bessales of brick lining also can be seen.
but it also provided the impetus for new ways of At the Baths of Maxentius, which were added to
thinking and was in ways more advanced and cer- the Severan Baths on the Palatine, the builders con-
tainly more efficient.31 The new confident use of structed all the barrel vaults of the substructures us-
lattice ribbing is demonstrated clearly in the work of ing continuous lattice ribbing that covered the en-
Maxentius. tire intrados of the vaults (Figs. 85 and 89). It was
87. Nymphaeum Alexandri “Trophies of Marius” (a.d. 222–235) in Piazza Vittorio Emmanuele. Upper
arrows indicate location of ribs. Lower arrows indicate corbels for centering.
104
P1: JPJ/JPK/KDF
VAULTING RIBS
Trajanic latrine in the Forum of Caesar where ribs and

relieving arches were used to compensate for the dif-
ferent wall locations at each level. Rather than creat-
ing an intricate network of ribs and relieving arches,
however, the Maxentian builders used the continu-
ous lattice ribbing so that they no longer had to plan
exactly where to put the ribs, and the vault was re-
inforced with voussoirlike compartments. The new
method shows a trust in the material that was perhaps
lacking in the Trajanic/Hadrianic period as well as a
greater concern for efficiency of construction. Laying
the lattice ribbing would have been somewhat more
time-consuming than laying caementa, but the pre-
planning for placement of the single ribs was avoided,
and the vault was less susceptible to random cracking,
88. Colosseum. Detail of rib in outer ambulatory. Left half and
bottom portion is original (a.d. 70–80) whereas right half of lat- as the bricks forced any cracks to occur radially like
tice construction is part of reconstruction after the fire of a.d. 217. joints between voussoirs. Moreover, the lattice rib-
bing in later examples seems to have been used as a
evidently intended to reinforce the vaults against the means of organizing the construction of the vaults.
load of the walls of the bath building above, which At the Arco di Malborghetto, the lattice ribbing has
were not congruent with those of the substructures. horizontal courses of bipedales that mark discontinu-
In this sense, the situation is similar to that of the ities in the arches of the ribbing. The discontinuous
89. Baths of Maxentius on the Palatine (a.d. 306–312). Detail of lattice ribbing in barrel vault of the
substructures.
105
P1: JPJ/JPK/KDF
that the two layers of ribbing for each vault were com-
pleted before the fill at the haunches was added. How-
ever, the walling underneath the entablature blocks
for the engaged columns had to be built up so that the
blocks could be set (Fig. 90). The space in between
the walling and the ribbing then seems to have been
filled in later. This sequence of events would have had
the advantage of creating a type of ribbed shell over
the centering that could have hardened before filling
in the spandrels.33 Even with such precautions, how-
ever, the builders seem to have encountered problems
during the laying of the vaults, as can be seen in Fig-
ures 26–27 (pp. 36–37).
ladder and lattice ribs in cross vaults

Ribs along the groins of cross vaults only appear in
the mid-second century, about a century after cross
vaults began to be built regularly in Rome. Hadrian’s
Villa in particular is famous for its variety of vault
90. Basilica of Maxentius (a.d. 306–315). Detail of nave façade
forms, many of which are groined and none of which
showing the spandrel between the central and easternmost barrel
vaults. Arrows indicate vertical construction joints. employed ribs of any sort along the groin. Clearly,
they were not absolutely necessary. Unlike the ribs
in barrel vaults, which had a definable purpose from
pattern of the ribbing suggests that it was used to their first use, the ribs along groins were responding
divide the vault into sections during construction.32 to different developmental influences.
The most visible of Maxentius’s projects, his basil- The earliest example occurs in a cross vaulted sub-
ica along Via Sacra, also employed continuous lattice structure vault (7.60-m span) of the phase-3 bath
ribbing in its surviving barrel vaults. These vaults did complex at the Villa di Sette Bassi (Figs. 82 and 91).
not support other structures above, but they are the The ribbing consists of single-bipedalis ladder ribs
largest freestanding concrete barrel vaults known from along each of the groins. It is somewhat different
the Roman world at 24.5 m. The ribbing consists of from the ribbing that was typically used later, which
two concentric rings of lattice ribbing, which appear more often consisted of three-part lattice ribs with
as the face of arches on the nave façade. C. M. Amici the bricks of the central arch ring cut to a point to fit
has recently pointed to a previously unrecognized fea- into the groin and form a sharp edge (Fig. 92). G. T.
ture that may help explain how these vaults were built. Rivoira argued that the Sette Bassi rib projected from
She notes that there are construction joints in the the intrados like a medieval rib, noting that he had
brick facing of the spandrels between two of the barrel seen some plaster remaining on the vaults during the
vaults. The construction joints in this location suggest nineteenth century when it was better preserved.34
106
P1: JPJ/JPK/KDF
VAULTING RIBS
been used to allow a lighter centering structure to be

used.
Within Rome, the use of ribbing along the groins
of vaults appears during the Severan period at the
Baths of Caracalla and then is used systematically from
the time of Diocletian. Most of the third- and fourth-
century examples consist of lattice ribs made with
three arches of smaller bricks connected with cross-
pieces of whole bipedales so that the middle arch of
voussoirs runs along the groin of the vault (Fig. 92).
Choisy suggested that the ribbing along the groins
was related to an attempt to lighten the centering
structure, and, given the evidence for the connec-
tion between ribbing and centering frames in bar-
rel vaults,36 this seems a reasonable proposal. I would
elaborate on Choisy’s original suggestion by adding
that the adoption of the diagonal ribs also may reflect
a change in the construction of the centering struc-
tures for which we have no direct evidence. By the
second century, there was a move away from fine join-
ery in shipbuilding,37 and the same phenomenon may
91. Villa di Sette Bassi (a.d. 140–150). Detail of ladder rib in cross
vault (C on plan on Fig. 82) of substructure of bath complex.
However, given the uneven nature of the edges of

the rib, this interpretation seems unlikely.35 All of the
other vaults adjacent to it employed brick linings, and
there are loose bessales on the ground under the vault.
Perhaps Rivoira saw the remains of the layer of lime
that was typically applied to aid in the adhesion of the
bricks and mistook it for a final layer of plaster. The
slight projection of the rib could then be explained
as the lip necessary to contain the edge of the brick
lining. The cross vault is the largest one in the com-
plex and the only one with the ribs along the groins.
Given the use of the brick linings, which would have
reduced the amount of wood necessary for the form-
work, perhaps the ribbing in this case could have 92. Three-arch lattice ribs along groin of cross vaults of Maxen-
tian substructures (a.d. 306–312) on southeast corner of Palatine.
107
P1: JPJ/JPK/KDF
93. Baths of Agrippa (third/fourth century a.d.). View of remains of dome showing pattern of lattice
ribbing.
have occurred in the construction industry in an ef- domes between the third and sixth centuries that he
fort to increase efficiency. Resolutions from the Theo- examined, fourteen had some form of ribbing and
dosian Code suggest that by the fourth century there none could be proven not to have had ribbing.40
was some difficulty in finding both skilled craftsmen Given the extensive use of ribbing in the barrel vaults
and materials,38 both of which could have resulted of the Severan structures on the Palatine, the use of
in changes that allowed for simpler and more effi- ribbing in domes likewise was probably a Severan
cient wooden structures to be used. If so, the added phenomenon, but there is no securely datable ex-
reinforcement along the groins may have been seen ample preserved. The rebuilt dome of the Baths of
as an extra precaution. By the early fifth century, Agrippa in the Campus Martius is often dated to the
builders managed to dispense with wooden center- reign of Alexander Severus (a.d. 222–235), although
ing completely by using pitched brick vaulting for this dating is based largely on a stylistic assessment of
the cross vaults in some towers of the Aurelian Walls its construction techniques, as there is no other doc-
built under Honorius (a.d. 401–403).39 umentation of a reconstruction under this emperor.
The only brick stamp from the area dates to the time
of Maxentius.41 The ribbing would fit well in either
ribbing in domes
period, but if it does date from the earlier period it
Ribbing in domes is a common characteristic of would be the earliest of the ribbed domes. It is only
fourth-century buildings. J. J. Rasch points out that partially preserved and is now built into the structures
of the twenty-one preserved or partially preserved along via dell’Arco della Ciambella (Figs. 93–94).
108
P1: JPJ/JPK/KDF
VAULTING RIBS
94. Plans of domes showing restored rib patterns. Shaded areas indicate preserved parts of building.
109
P1: JPJ/JPK/KDF
One of the most striking aspects of the ribbing (Fig. 42, p. 57)). The bipedalis courses divided the
found in the late Roman domes is the apparent move domes into vertical layers, and the ribs divided them
away from being conceived as a load transfer system in- into wedges so that the whole structure was orga-
tended to direct load to particular points of the sup- nized into convenient constructional units. The cor-
port structure to a load distribution system intended respondence of the formwork to the marked levels
to distribute loads more evenly throughout the vault in these domes indicates that lattice ribbing played
itself. This tendency has been seen already in the dis- a large part in the regularization of the construction
cussion of lattice ribbing in barrel vaults, such as at sequence of domes by creating both horizontal and
the Baths of Maxentius on the Palatine, but it be- vertical divisions so that the vaults could be built in
comes clearer in the examination of domes. For ex- stages.
ample, the ribs often occur directly over openings in If the ribbing and horizontal courses of bipedales
the wall below (Minerva Medica, the “Planetarium” were intended to divide the work into manageable
at the Baths of Diocletian, and probably the Baths of sections, one question that arises is, “How much
Agrippa), which is certainly not where one would work does one section represent?” A calculation of
want to direct the load (Fig. 94). Likewise, some ex- the man days required to lay the caementa between
amples of ribbing do not even continue down to the the ribs and courses of bipedales shows that at both
impost, suggesting that it was not conceived as trans- Minerva Medica and the Mausoleum of Helena the
ferring the load to a particular point on the wall (Mau- section could be laid by two workers in less than a
soleum of Helena, Octagonal Hall at the “Villa of the day.43 If the ribs were built separately and just ahead
Gordians,” and Tor de’Schiavi). Of these examples, of each new layer, then the bipedalis covering could
the load distribution system is best exemplified at the be used by the rib builders as a work platform for the
Gordian Octagon, which has a continuous mantle of next vertical section of ribs. In this way, the dome
lattice ribbing along the intrados, thereby completely could rise alternating with ribs and infill so that each
disregarding the location of wall openings or piers in level of infill provided the platform on which the
the wall below. rib builders could stand to access the inner parts of
J. J. Rasch’s photogrammetrical study of the in- the vault. This also could explain why the formwork
trados of the domes has yielded information on matches the bipedalis courses: The carpenters who ap-
the relationship between the ribbing and the form- plied the formwork to the centering could use the
work. For domes built with radial formwork (Min- bipedalis layers as platforms on which to stand to add
erva Medica, Mausoleum of Helena), the ends of the the next layer of formwork to the centering structure.
boards aligned with the horizontal courses of bipedales, This use of lattice ribbing in domes probably devel-
which suggests that the construction was methodi- oped out of the ribbing that first appeared along the
cally organized around the principle of building the groins of vaults and then was adopted and expanded
dome up layer by layer in correspondence to sec- for circular and semicircular vaults. In the process of
tions of formwork.42 Some of the examples of rib- the transformation, however, the purpose underwent
bing are also integrated with other features such as a subtle change from transferring loads along the groin
the horizontal courses of bipedales that form the cov- to specific points at the corners to distributing the
ering of step-rings (see Chapter 7) on the extrados loads and regularizing the construction process. The
(Minerva Medica (Fig. 95), Mausoleum of Helena use of the lattice ribbing in domes appears to have
110
P1: JPJ/JPK/KDF
VAULTING RIBS
construction, as seen at the Sanctuary of Hercules

Victor at Tivoli (Fig. 64, p. 89), but the technique is
most commonly employed with brick. The develop-
ment of the ribs, not surprisingly, therefore, parallels
the development of the brick industry, as was the case
with the brick linings (Chapter 2). The ribs that use
the most bricks occur when the brick industry was
expanding, whereas those that use fewer whole bricks
were developed once the brick industry began to de-
cline. The ribs, unlike the linings, were structural el-
ements and were more important for the long-term
stability of the building, so when large-scale building
returned after the mid-third century, the ribs reap-
peared in their new form (lattice ribbing), whereas
the linings were completely abandoned.
The change in the form of the ribs from solid brick
to lattice was accompanied by a change in concep-
tion. The Baths of Caracalla offer a unique view into
this transition, as it is the only surviving building to
employ both types together in the same phase of con-
struction. The solid bipedalis ribbing was reserved for
vaults that underwent higher load bearing conditions,
whereas the lattice ribbing was used more for con-
structional purposes. Shortly thereafter, however, in
the reconstructed parts of the Colosseum, the lat-
tice ribbing was used as a replacement for original
95. Section of “Temple of Minerva Medica” (first half of the solid brick ribbing. By the fourth century, bipedalis
fourth century a.d.). The numbers along the intrados indicate ribs were no longer used at all, and the lattice ribs
the length of the formwork boards in RF.
were not typically used to channel loads to particular
points anymore but, rather, to provide an overall sys-
been intended to aid in this process and to stiffen the tem of load distribution within the vault and a means
concrete as it was curing. of stiffening for large vaults. For domes in particular,
the ribbing also seems to have had a role in the orga-
nization of the construction process by dividing the
conclusions
vault into horizontal and vertical units that could be
The ribbing in vaults is a multifunctional element, completed in a day.
the intention and purpose of which changes both The change in the conception of how the ribs func-
from context to context and from time period to tioned was first affected by the switch from travertine
time period. The original idea comes from cut stone to brick in the first century and then later by the
111
P1: JPJ/JPK/KDF
economic fluctuations regulating the production of under the tetrarchy seems to have been largely in the
the brick. The solid brick rib had developed out of the hands of the state. In this context, the efficient use
travertine rib, as seen at the Colosseum, and was con- of the material was the impetus that eventually re-
ceived in the tradition of using a hierarchy of mate- sulted in the ribbing being used in a different man-
rials whereby the areas of greatest load concentration ner from those original travertine ribs, which had
within a structure were reinforced with materials of first been used at Tivoli near the travertine quarries
higher strength and durability. The development of themselves.
the lattice rib marked a change away from this idea. The development of vaulting ribs provides a useful
By the fourth century, the lattice ribbing was not model for the application of the four criteria used
used to reinforce sections of vault that were per- in Chapter 1 to define an innovation. The origi-
ceived as bearing greater loads but, rather, to dis- nal travertine ribs were a result of the accumulated
tribute the weight of the vault more evenly through- knowledge of centuries of cut stone arch construction
out the structure or to reinforce the entire vault. This applied to the evident need for noncongruent plans in
was not possible in the original material of travertine the late Republican sanctuary at Tivoli. The economic
due to the time-consuming necessity of carving each incentive lay in the existence of the nearby travertine
block. With the substitution of brick for travertine, quarries. The change to the use of bricks for ribbing
new ways of building were possible without adding was paralleled by the growth of the brick industry,
greatly to the labor costs, but the change in con- which offered economic incentive. Eventually the devel-
cept took time to develop. The innovation of lattice opment of the lattice ribbing represented further ad-
ribbing also was given some impetus by fluctuations vances in accumulated knowledge, and the reduction in
in production and in the economy, which encour- the number of whole bipedales used provided the eco-
aged the development of a method employing fewer nomic incentive for the change once brick production
whole bricks. By the fourth century, the brick indus- declined. Moreover, the shift in the purpose of the
try was no longer an area that provided the potential ribbing led to greater efficiency in the construction
for social mobility at many levels, as it had been ear- of domes, which in the fourth century became in-
lier. The incentive to produce massive numbers of creasingly desirable for tombs and later for churches
bricks, as was apparently the case under Trajan and (see further in Chapter 9), thereby reflecting social
Hadrian, was by then gone, and brick production acceptability.
112
P1: NDZ
6
METAL CLAMPS AND TIE BARS
O ne of the hidden but nevertheless crucial

elements in the development of the architec-
tural vocabulary of imperial Rome was the use of
earliest datable use of clamps in Rome are the iron
pi clamps used to attach the travertine facing of the
Metellan rebuilding of the Temple of Castor in the
metal fittings, such as clamps, dowels, and tie bars. Forum just after 117 b.c.4 Other examples occur spo-
These elements became particularly important when radically during the late Republic such as in the round
marble was introduced into Rome as a major build- temple in the Forum Boarium (c. 100 b.c.), where
ing material. The growth of the marble trade during iron pi clamps were used to attach the Pentelic marble
the imperial period created an environment where, facing of the cella to the travertine backer blocks.5 In
on the one hand, Greek classicism provided mod- the first century b.c., wooden dovetail clamps were
els of trabeated structures that had developed in the also being used, usually to connect blocks of softer
marble-rich areas of the Aegean and, on the other, volcanic stone, such as the peperino (lapis Gabinus)
concrete provided the potential of creating new types blocks at the Tabularium (78–65 b.c.) and later at the
of interior spaces. The marriage of these two was ulti- Forum of Augustus, where the remains of the oak
mately made possible by the hidden metal fittings that clamps have been found.6 The wooden dovetail
allowed the concrete vaults to be securely attached to clamps were typically used in foundations or walls
the marble support structure. rather than in entablature blocks where stability was
Metal clamps and other experimental uses of metal a consideration. None of these early examples of
in architectural contexts had appeared in Greece by clamps was intended to resist great tensile stresses
the fifth century b.c.,1 but they only appeared in but rather to secure the blocks together or to attach
Rome at the end of the second century b.c. when revetment.7 M. Blake noted that the dovetail clamps
travertine and imported Greek marbles began to be were not used after the time of Augustus,8 but in
used there.2 The Romans had access to the Etruscan fact they were adopted extensively (in bronze or
iron resources from Elba and Populonia from the mid- iron) to connect the marble entablature blocks of
third century b.c.,3 so the late development of the use trabeated structures throughout the imperial period.
of iron for clamps seems to have more to do with the By the Augustan period, the use of metal clamps
supply of stone than with the supply of metal. The and dowels especially for connecting marble blocks
113
P1: NDZ
96. Forum of Trajan (a.d. 106–113). Architrave block with remains of iron tie bar set in lead (inset
upper left).
was fairly common, and this provided the means Even the cuttings can be open to interpretation since
to experiment with larger and more daring uses of the original position of fallen blocks is often debated.
vaulting in combination with stone supports, but the For standing structures, the task is sometimes even
transition came slowly at first. The most challenging more difficult, because the evidence is still embed-
structures were the arcades and colonnades of ded in the stone itself. In some cases, the presence of
porticos. Traditionally these types of structures were the metal clamps or dowels can be detected from the
covered with a wooden roof, which was lighter than robber holes from which they were taken, although
concrete vaulting and bore straight down onto its we do not always know what the robbers found in
support structure without exerting any horizontal their holes because they often chipped away the evi-
thrust. Once the builders decided to cover them with dence. Sometimes, however, the parts of the original
concrete vaults, they had to deal with the possibility cuttings are still visible.
that lateral thrusts could develop. When Roman builders began to construct arcaded
Metal fittings of any sort rarely survive, usually hav- and colonnaded porticos supporting concrete vaults,
ing been removed to use for other purposes, but they they often had to provide some means of counter-
often can be identified by the cuttings on the stones ing the horizontal thrusts that could develop both
to which they were anchored. With luck, traces of during and after construction. One strategy was the
rust can be used to identify the material in the case one adopted at the Theater of Marcellus where the
of iron or a greenish patina in the case of bronze. builders chose to cover the second level ambulatory
114
P1: NDZ
with a series of side-by-side barrel vaults so that they

each balanced each other and no horizontal thrust
could develop along the outer wall (Fig. 65, p. 89).
This was not a particularly elegant solution and was
not repeated in other theater/amphitheater buildings
in Rome, although it did occur outside of Rome in
the amphitheaters at Nimes and Arles, which were
evidently inspired by the Theater of Marcellus. Two
other solutions involved the use of metal. One was to
connect the adjacent blocks of the entablature sup-
porting the vault with metal clamps, and the other
was to use transverse tie bars to connect the entab-
lature blocks on either side of the vault. Only one
block survives with the remains of an iron tie bar set
in lead (Fig. 96), but the cuttings in other blocks attest 97. Horrea Agrippiana (33–12 b.c.). Reconstructed section
to their existence. showing proposed location of tie bar (after H. Bauer).
portico. Two of the Type A blocks with the cuttings

the earliest use of tie bars
have cornices abutting the engaged columns at either
The earliest evidence for tie bars occurs in the Augus- side and continuing on the two lateral faces. The top
tan period at two buildings: the Horrea Agrippiana side of one of the blocks is worked smooth along the
(33–12 b.c.) and the portico (of Gaius and Lucius [?]) front two thirds and has a projecting, roughly worked
along the front of the Basilica Aemilia (14 b.c.–a.d. section along the back third (Fig. 98). The cutting
4). The superstructures of both buildings are fragmen- for the tie bar occurs in the roughly worked section.
tary, but H. Bauer, who was the first to suggest the The rear face of the block is flat. Bauer noted that the
existence of the tie bars, has proposed reconstructions cutting could not have been used to attach another
for both buildings. As the evidence for these tie bars block because traces of plaster on the rear face indi-
is based on reconstructions from blocks no longer in cate that it was visible. The location of the cutting
situ, I review the basic logic by which Bauer recon- in relation to the cornice suggests that the bar was
structed both conditions. visible (Fig. 97). The Type B blocks, which Bauer
Bauer shows the Horrea Agrippiana as a series of interpreted as anchor blocks in the dividing walls be-
barrel vaulted rooms facing onto an arcaded corridor tween the barrel vaulted rooms, also have L-shaped
surrounding the four sides of an open courtyard at two cuttings. The upper surface and two sides of each
levels (Fig. 97). Based on the L-shaped cuttings found block are rough, indicating that they were probably
in three of the travertine cornice blocks (Type A) be- embedded in the concrete of the vault. The front face
longing to the engaged pilasters of the arcade and of one block has anathyrosis and therefore must have
in two travertine blocks (Type B) probably from the been placed next to another one.9 Because this face of
walls of the barrel vaulted rooms, he reconstructed the block was obviously not visible, Bauer showed it
the corridors as having exposed tie bars at the base located behind the façade block in his reconstruction
of the springings of the vault, as in a Renaissance (Fig. 97).
115
P1: NDZ
Because the cutting is off-center on the

block, Bauer restored a similar cutting
on the missing portion and showed the
portico as having double bars above each
of the piers. Both blocks have two con-
tinuous longitudinal grooves (7.5 × 7.5
cm) cut along the upper surface for metal
bands that apparently connected each to
its adjacent voussoir blocks.
The reconstruction of the portico is
based on very few remaining blocks and
is open to some debate. The cuttings for
the metal bands, which occur on two dif-
ferent blocks, are more definitive than the
single dovetail cutting for the tie bar. In
this case, the builders may have been using
98. Horrea Agrippiana (33–12 b.c.). Type A cornice block of arcade pier with a combination of methods to ensure the
L-shaped cutting for tie bar on top surface.
stability of the structure. The dovetail cut-
Bauer’s reconstruction of the tie bars at the portico ting for the tie bar at frieze level would have resulted
in front of the Basilica Aemilia was likewise based on in an exposed bar like those proposed for the Hor-
cuttings in surviving blocks of the entablature, only rea Agrippiana, but the Romans evidently found the
two of which survive. He showed the structure as a exposed bars unsatisfactory because they were rarely
two-story arcaded portico covered by cross vaults off repeated in later buildings. The exposed tie bars low
of which opened two levels of barrel vaulted shops in the vault makes sense to modern reasoning because
(Fig. 99). Both of the surviving blocks are trapezoidal it is where the horizontal thrusts are greatest, but in
marble blocks that belonged to a Doric frieze; one later examples, the Roman builders went to some ef-
has the remains of a metope displaying a bucranium fort to hide the metal bars at the crown of the vault.
and the other has part of a metope with a patera. The The Augustan evidence shows that the builders were
shape of the blocks was clearly formed to accept adja- confronting the issue of how to deal with the lateral
cent voussoirs of a lintel arch that protected the round thrusts of vaults, but the final resolution took a dif-
arches of the portico, so they can both be located ferent form from these first experimental examples.
directly above piers. He assigned the patera block to
the first level frieze and the bucranium block to the
tie bars in the imperial thermae
second level frieze.10 The bucranium block is the only
one of the two that has a tie bar cutting (Figs. 100– The definitive evidence for the use of tie bars in
11
101), but it is different from any other surviving standing structures comes from the imperial thermae.
example in that it is a dovetail cutting rather than an L- The earliest example can be seen at the Baths of Trajan
shaped cutting. The back of the block is not preserved where the anchor block (travertine or marble [?]) used
to indicate how the bar would have exited the block. to house the end of the metal tie is still embedded in
116
P1: NDZ
99. Portico fronting the Basilica Aemilia (14 b.c.–a.d. 4). Reconstruction showing proposed location
of tie bar (after H. Bauer). Elevation (right) shows preserved bucranium block in restored location.
100. Portico fronting the Basilica Aemilia (14 b.c.–a.d. 4). Measured sketch of bucranium block with
images of block superimposed.
117
P1: NDZ
101. Portico fronting the Basilica Aemilia (14 b.c.–a.d. 4). View of bucranium block. “A” indicates
dovetail cutting for tie bar. “B” indicates cuttings for dowels. “C” indicates longitudinal cuttings for
metal bands.
the wall (Fig. 102).12 Renaissance drawings of the plan agree on the use of the tie bars there. The areas
of the Baths show that the wall with the blocks formed of the Forum for which tie bars have been pro-
the south boundary of the east palaestra (Pl. XIII).13 posed include the aisles of the Basilica Ulpia and
The wall retains the traces of the barrel vault that once the porticos surrounding the Column of Trajan.
covered the colonnaded portico (c. 6-m span) of the The Basilica Ulpia was the first of the major basil-
palaestra. One anchor block remains and the loca- icas in Rome built with colonnaded aisles sup-
tion of another is marked by the hole that contained porting concrete vaults, fragments of which remain
it. Similar holes are preserved in both palaestrae at lying on the ground. The ground level aisle vaults
the Baths of Caracalla (Fig. 103),14 and the anchor were supported on over one hundred columns
blocks themselves are preserved in both palaestrae at of Mons Claudianus gray granite, which was one
the Baths of Diocletian (Fig. 104). The latter blocks of the more precious colored stones used by the
were clearly made from spolia, as some have the re- Romans. The use of so many granite columns sup-
mains of moldings visible.15 In all cases, the blocks porting concrete vaults was one of the largest and
occurred just below the crown of the vault so that most spectacular examples of the combination of a
the bars were concealed in the crown (Fig. 105). trabeated stone support with concrete vaulting. One
question is whether this innovative design relied on
the use of metal tie bars to ensure stability.
tie bars at the forum of trajan
C. M. Amici was the first to propose the existence
The reconstruction of the Forum of Trajan is a of tie bars in the Forum complex based on the cut-
notoriously controversial topic, and not all scholars tings in three marble blocks found on-site (Fig. 106),
118
P1: NDZ
Fig. 106) are cornice blocks belonging to the portico

surrounding the Column of Trajan, and the third (C)
is a cornice block belonging to the Basilica Ulpia
(Table 4, p. 123). All scholars agree on the general
locations of these blocks; the precise locations are
somewhat more controversial. A fourth block from
the Forum of Trajan has the remains of tie bar still
embedded in it (Fig. 96), but it is more difficult to
locate and is not included in the following analysis
(see further in Appendix 1.14).
Amici and Packer each restored blocks A and B,
both of which have L-shaped cuttings for bars exiting
their rear faces, to different locations in the Column
Portico (Fig. 107). Because the rear face of block A
was left roughly carved with a point chisel, Amici
restored it above the architrave/frieze blocks of the
colonnade so that the rough portion is embedded in
the concrete of the vault and the proposed tie bar runs
concealed in the crown of the barrel vaults.17 Packer,
by contrast, restored blocks A and B as embedded
in the east façade wall of the West Library and sug-
gested that the L-shaped cuttings were for pi clamps;18
however, in this proposal there are no adjacent blocks
to which A and B could have been clamped. Both
of these blocks also have cuttings for dowels in the
top surface indicating that there was a block above,
which is not accounted for in Packer’s reconstruc-
tion. Of these two proposals, Amici’s fits best with
the evidence from the blocks. The remains of the
anchor blocks from the identical situation at the
102. Baths of Trajan (a.d. 104–109). View of section K showing palaestra at Trajan’s Baths lend credence to Amici’s
holes for anchor blocks for tie bars of palaestra portico. Inset shows reconstruction.
detail of hole with remaining block that has cutting for tie bar.
Block C is a cornice block that Packer and Amici
agree belongs to the first order of the south façade
but J. Packer in a more recent publication challenges of the Basilica Ulpia, but they disagree on the exact
her interpretation, and other scholars working on the placement because of their different reconstructions
complex have expressed doubts on the use of the tie of the façade wall (Fig. 108).19 Amici restored the
bars.16 Given the controversy, a reexamination of the south façade as a solid wall with openings for the three
evidence is warranted. Two of the blocks (A and B in porches based on structural considerations,20 whereas
119
P1: NDZ
103. West palaestra of Baths of Caracalla (a.d. 212–216). Arrows indicate robber holes where anchor
blocks for tie bars have been removed.
104. East palaestra of Baths of Diocletian (a.d. 298–306). Arrows indicate anchor blocks for tie bars.
120
P1: NDZ
105. Generic reconstruction of palaestra portico from an imperial

bath showing location of tie bar and anchor blocks.
Packer restored it as an open colonnade based on re-

mains of a Renaissance drawing of the Severan Marble
Plan that represents the south façade with dots instead
of lines.21 Block C has a cutting for one bar extend-
ing out its rear face (Fig. 106), which is not preserved,
but another block that probably belongs to the same
cornice has a roughly carved rear face suggesting that
it was embedded in a vault.22
The aisles of the Basilica Ulpia were certainly
vaulted because a section of the fallen vault from the 106. Forum of Trajan (a.d. 106–113). Blocks A–C showing top
north aisle is preserved, but the exact form of the surfaces with cuttings.
vault has been disputed. Amici noted that some of
these chunks of vaulting retain the original surfaces
121
P1: NDZ
107. Column Portico at Forum of Trajan (a.d. 106–113). Comparison of reconstruction proposals by
C. M. Amici and J. Packer showing locations of blocks A/B.
108. Basilica Ulpia (a.d. 106–112). Comparison of reconstruction proposals by C. M. Amici and
J. Packer showing locations of block C and fallen vaulting.
122
P1: NDZ
table 4. Blocks from Trajan’s Forum with cuttings for tie bars
Block Spacing of
TF Amici Packer type/size Cuttings cuttings
Block Inv. # 1982 1997 (m) (cm) (m)
A 2545 figs. 137–140 cat. 157 Cornice tie bars: ∼1.5

3.14 (l) 43 (l) × 5 (w) × 4 (d)
42 (l) × 6 (w) × 5 (d)
B 2846 figs. 141–144 cat. 157A Cornice – ∼1.0
1.53 (w-top)
0.83 (w-bot)
0.75 (h)
C 2706 figs. 50–52 cat. 167D Cornice tie bars: –
0.98 (h) 4.8 (w) × 4 (d) cm
dovetail clamp:
12/7 (w) × 6 (d)
of three sides: the top, the side, and the intrados, restores the vaults with a higher, more rounded form
which allow it to be accurately placed. In her restora- of vault in order to allow room for the reconstruc-
tion, the proposed tie bar projects through the vault tion of the Dacian frieze along the parapet. In this
just above its crown and was never visible.23 Packer reconstruction, the intrados side of the vault is not
109. Basilica Ulpia (a.d. 106–112). Plan showing author’s proposed location for tie bars and location of
fallen vaulting.
123
P1: NDZ
110. Hall of the Doric Pilasters at Hadrian’s Villa (a.d. 125–133). Reconstruction of lintel construction.
considered to represent the finished surface, and the An argument against the existence of the tie bars
tie bars are eliminated because they would project is that there is no evidence for them in the largest
awkwardly through the middle of the vaulted space.24 remaining section of fallen vaulting, which stretches
He concludes that the cutting noted by Amici was for approximately 5 m.25 This section of vaulting,
for a pi clamp rather than a tie bar; however, this, however, belongs to the north aisle and would have
too, runs into the problem that there is no ad- been supported on one side by the brick wall of the
jacent block to which to clamp it. Since Packer’s West Library (Fig. 109). As the only block from the
publication, some of the fallen vaulting has been Basilica with a cutting belongs to the south façade,
raised so that the underside is now more easily vis- there is no direct evidence that the tie bars were nec-
ible revealing that it does have formwork imprints, essarily used throughout the building. Packer rightly
which indicates that it formed the finished surface noted that the weight of the superstructure above
of the vault (see Appendix 2a.6 for formwork in- some sections of the vaulted colonnade would have
formation). The existence of the formwork imprints been enough to counter the lateral thrust that could
on the remaining vaulting and the cutting on block develop in the vaults and that the three porches along
C make Amici’s proposal for the lower vault and the south façade would have provided buttressing for
tie bar combination the more reasonable reconstruc- those sections. Moreover, all the blocks were con-
tion of the vaults, but Packer’s reconstruction of the nected with two iron dovetail clamps that could
open south façade is supported by the fact that all have helped resist some degree of outward thrust
the surviving architrave blocks of this façade have (Fig. 106a). So, the most vulnerable sections of the
carved soffits indicating that they spanned between building, as noted by both Packer and Amici, were
columns. the three-bay sections between the porches. These
124
P1: NDZ
111. Hall of the Doric Pilasters at Hadrian’s Villa (a.d. 125–133). View of reconstructed portico. Inset
shows marble soffit panel (foreground) and travertine impost block (background). Arrows indicate
cuttings for iron bars.
sections had neither the buttressing effect of the B) from the Column Portico at Trajan’s Forum both
porches nor the downward pressure of the upper sto- have cuttings for two tie bars spaced approximately
ries. If Packer is correct in reconstructing the south 1.5 m and 1.0 m, respectively, compared to the 3.7-m
façade as an open colonnade, these sections would spacing of the columns. The bars could have occurred
have been even more likely to have been reinforced. in pairs near every column or they could have been
I would propose that the tie bars were used but only more widely spaced. The anchor blocks remaining in
in these locations as indicated on Figure 109. This situ at the bath buildings provide more evidence for
would explain both the existence of the tie bar cut- the spacing of the bars over long distances. The aver-
ting and the lack of tie bar evidence in the fallen age spacing is around 3 m at all three of the imperial
vaults from north aisle. Likewise, the use of the bars thermae, but it does vary. At the Baths of Caracalla,
for the Column Portico, which in Amici’s recon- the robber holes are spaced between 3.0 and 4.8 m
struction supports only a terrace, would have been apart (Pl. XIII), and in some places at the Baths of
required because of the lack of downward pressure Diocletian the spacing could have been as great as
from any other structures above. These are issues ex- 10 m (Pl. XIII). Whether the bars were aligned with
plored further using techniques of structural analysis the palaestra columns is difficult to know, because the
in Chapter 8. precise column placement is unknown in all cases. A
restoration of equally spaced columns suggests that the
spacing of tie bars
bars and columns were not necessarily in alignment,
The spacing of the tie bars can sometimes be recon- although they tended to be placed at roughly the same
structed. Their location at the crown of the vault interval depending on other features of the wall such
eliminated any necessary correspondence with the as down drains, which may have required them to
column placement. The two cornice blocks (A and be shifted.
125
P1: NDZ
the use of tie bars with lightweight the Stadium Garden, the Hall of the Doric Pilasters
caementa (Fig. 111), and the Nymphaeum directly northwest
of it.26 Unlike the examples of metal tie bars exam-
Examples of tie bars from the second century and
ined earlier, the system of metal bars at Hadrian’s Villa
later are always used in conjunction with colonnaded
was used specifically to create architraves of less pre-
porticos and are always combined with lightweight
cious materials, brick and travertine, which were then
caementa. As the tie bars would resist any horizon-
revetted with thin marble panels. The ultimate inten-
tal thrust that developed, the lightweight caementa
tion, however, was to create the appearance of a mar-
must have been intended to reduce the weight on
ble trabeated support structure for concrete vaulting.
the architrave of the colonnades rather than to reduce
The iron architrave bars were anchored in holes
the thrusts. This is supported by the fact that at the
in the top surface of the impost blocks and then ran
Basilica Ulpia the architrave and cornice are carved
down grooves carved into the oblique side surfaces,
from a single block, thus creating a taller “beam” to
under the flat brick arches, and back up grooves in the
span between the columns. The earliest example of
next block. (Fig. 110). Variations in form occur so that
the bars, at the Baths of Trajan, uses caementa of tufo
some blocks fit onto a corner column or fit up against
giallo della via Tiberina alone, but all other examples
a sidewall. Impost blocks found at the Stadium Garden
employ either scoria or pumice. The Basilica Ulpia
have the remains in the holes and along the grooves
vaults contain a mix of tufo giallo and Vesuvian scoria,
of the lead that was used to fix the bars in place.27
whereas the later palaestra vaults at both the Baths of
The lodging holes in the top of the blocks at the
Caracalla and the Baths of Diocletian are built with
Hall of the Doric Pilasters would have accommodated
scoria and pumice, respectively, in the crown with-
bars 3–4 cm square,28 which coincides with the iron
out the admixture of tufo giallo. The decision to com-
dowels remaining in the complex. The builders were
bine the scoria with tufo giallo may represent an effort
probably receiving stock bar sizes that could then be
to create a more resilient concrete than using sco-
cut on site into dowels or bent into the architrave bars
ria alone. The broken remains of the fallen vaults of
as needed. There were no doubt blacksmiths on the
the Basilica Ulpia demonstrate that the scoria created
building site for such work and for tempering and
planes of weakness in the concrete. The two later bath
sharpening carving tools.
buildings used the heavier tufo lionato at the haunch
All of the examples of the impost blocks are trape-
and the lighter scoria or pumice at the crown.
zoidal in form and raised on a rectangular base to
create a slot between the top of the column capital
for a marble soffit slab to fit (Fig. 110). The soffit
iron architrave bars and impost blocks
panels used to cover the bottom of brick lintel arches
at hadrian’s villa
are still lying on the ground at the Hall of the Doric
At Hadrian’s Villa, a different method of construc- Pilasters. Some have roughly carved depressions in the
tion was used for architraves of colonnades that sup- upper surface that align with the metal bars indicating
ported concrete vaulting. Travertine impost blocks that they were put into place before the bars were
connected with iron bars were combined with brick added (Fig. 111). Ideally, the bars provided a barrier
lintels (Fig. 110, p. 124). This combination was used in between the soffit panel and the brick arch so that
colonnades at the Maritime Theater, the Serapeum, no pressure was placed on the relatively thin marble
126
P1: NDZ
protected the marble soffit panels from pressure.32 If

the bars had not been used, either the brick voussoirs
would have had to be constructed directly against the
marble slab, thereby using it as formwork, or else the
marble slab would have had to have been inserted
after the arch was already built.
Fixing the bars with lead both in the top holes and
along the grooves would have been crucial, as the un-
fixed bars by themselves would not have sufficiently
stabilized the structure during construction. Once the
entablature was completed, the iron bands would have
served to resist any lateral thrusts from the vaults by
binding the entablature together. If the builders were
only concerned about the stability after construction,
they could have easily added the bars after the brick
112. Santa Costanza (mid-fourth century a.d.) showing arcuated
brick lintels supported on spoliated columns and capitals.
arches were complete and simply connected the tops
of the two impost blocks. The advantage of running
the bands below the arches is that they served to sta-
panels (c. 15 cm thick spanning 1.8 m), as can be seen bilize the row of columns while the arches and vaults
by the small lip left at the bottom of the grooves in were being built. These bars are an excellent example
most impost blocks (Fig. 110). Each soffit panel also of the ingenuity of the Roman builders in coming
has the pry holes along the edges for positioning the up with a multifunctional element that solved many
vertical architrave/frieze panels once the brick arches problems at once.
were in place.29
The construction method was chosen to allow for
conclusions
entablatures of brick and travertine revetted in mar-
ble panels to be substituted for those of solid blocks of The use of tie bars grew out of the tradition of using
marble.30 The travertine quarries were located within metal fittings for cut stone construction. This tradi-
three kilometers of the villa (Map 3, p. 14), and tion came late to Rome and was evidently related to
brick stamps reveal that a majority of the bricks came the exploitation of the travertine quarries at Tivoli in
from the Tiber valley in the area between the conflu- the late second century b.c. and to the importation
ence with the Aniene and with the Farfa.31 Imported of Greek marble by the early first century b.c. By the
blocks of marble, on the other hand, had to be trans- Augustan period, the builders were experimenting
ported up the Aniene from Rome. This technique of with the use of iron in marble structures, but it was
using the iron bars with travertine blocks for porticos by no means the rule as it became later. As long as tuff
is one that was particularly suited to Hadrian’s Villa remained a major building material, wooden dovetail
given its location near both the travertine quarries and clamps, such as those used throughout the Forum of
clay beds. It would have provided a permanent cen- Augustus, were employed whenever possible. These
tering for the brick lintel arches and at the same time clamps were only used to join adjacent wall blocks
127
P1: NDZ
and were not intended to undergo any great tensile end, into the Forum proper, and onto the upper parts
stresses, unlike the later use of iron dovetail clamps of the Column of Trajan. The other Trajanic exam-
in entablature blocks that were intended to stabilize ple of vaults supported on columns comes from the
colonnades. palaestrae at Trajan’s Baths. We do not have evidence
The technique of using tie bars was prompted for how these terraces were used, but the comparable
by the desire to combine stone support structures, ones at the Baths of Caracalla were covered with fig-
particularly colonnades, with concrete vaulting. The urative mosaic indicating that they would have been
substitution of the typical wooden roof structure used by the patrons (as opposed to being a service
with concrete vaults meant that the colonnades had area), probably as a gallery to watch the exercising tak-
to be stabilized both during the construction while ing place below. The combining of colonnades with
the concrete was being laid in a semifluid state and concrete vaults was the result of both aesthetic and
afterward when cracks could easily occur and result in functional considerations, but it also was a manifesta-
lateral thrusts. The early attempts at using tie bars in tion of the desire to display power through the con-
the Augustan period appear to have been made on ar- quering of foreign territories and the technological
caded structures, which are more stable than the later prowess necessary to exploit what they had to offer.
trabeated colonnades. By the early second century, The combined use of the metal architrave bars and
however, the marble trade was fully developed and impost blocks at Hadrian’s Villa represents a clever
columns were fashionable. The use of colored stone device that served to stabilize the structure and to
columns in imperial structures had steadily grown provide a way of constructing a post and lintel sys-
during the first century, and this was paralleled by in- tem that appeared to be of marble. Given the location
creasing state control over the quarries as can be seen of the villa outside of Rome and up the Aniene, this
in the quarry inscriptions.33 The Basilica Ulpia rep- method of construction offered a much more efficient
resents the most prodigious use of Mons Claudianus means of achieving the desired effect without the
gray granite columns ever assembled. This was a mate- added transportation costs of hauling very large pieces
rial with particular prestige value, and it was only used of solid marble upstream to the building site. The de-
in Rome for imperially sponsored state projects, as has gree to which this technique was used elsewhere is
been shown by D. P. A. Peacock in his work on the unclear, but a few impost blocks with cuttings found
Mons Claudianus quarries.34 Each column weighed around Rome and Ostia suggest that it may have been
over thirty tonnes and had to be carted over 100 km more common that typically thought.35
of the Eastern desert of Egypt before reaching river The earliest examples of the tie bars at the Horrea
transport along the Nile, thus the use of so many of Agrippiana and at the portico of the Basilica Aemilia
them was a statement of power. appear to have been exposed, but in later examples
If the intention was to display the columns, why the builders went to some effort to hide them at
then the concrete vaults if a wooden roof structure the crown of the vault. The one exception is the
over the aisles would work just as well? The most ob- anomalous block from the Forum of Trajan shown
vious advantage of the vaulted aisles is that they pro- in Figure 96, which cannot be precisely located but
vide flat usable terraces above. At the Basilica Ulpia, which clearly had an exposed tie bar given its loca-
the vaults created upper floor levels that gave views tion within the entablature. The goal of the Roman
into the nave area, possibly into the exedrae at either designers in the developed form was to hide the tie
128
P1: NDZ
bars from view presumably to prevent them from in- Byzantine use of metal tie bars and wooden tie beams
truding on the spatial qualities provided by the curv- is much more well known that the earlier Roman
ing ceilings of the vaults. The transition between the practice because they were exposed, but it is probably
aesthetic values displayed in the architecture of the one adopted from the well-established, albeit hidden,
first and second centuries in Rome and those of later technique developed by Roman builders.39 The move
times and other places can be seen during the fourth toward using more exposed ties represents a very dif-
century. At this time in Rome, a new aesthetic de- ferent aesthetic sense from those of second-century
veloped in which colonnades with flat lintels were Rome, but it also probably represents a change in
often replaced by arcuated lintels.36 Tie bars could social acceptability that came with the different social
no longer be concealed in the architrave as they had values of a culture that did not have the material
been at Hadrian’s Villa. The new design motif can resources that the imperial capital had once com-
be seen in early Christian buildings in Rome, such manded. Seneca aptly sums up the “superficial” values
as Santa Costanza (Fig. 112, p. 127) and Santa Sabina. that he sees developing under Nero when he notes:
Most often these later colonnades were created with
brick arches supported on columns taken from earlier Children are delighted by the smooth and variegated
structures. The new aesthetic probably developed, at pebbles they pick up on the beach while we take
least in part, as a means of avoiding having to use delight in tall columns of spotted marble brought
either from the Egyptian sands [granite from Mons
whole architraves of marble,37 which if taken from Claudianus or Aswan] or from African deserts [giallo
other buildings had to be made to fit together, as was antico from Chemtou] for a colonnaded portico or
done at San Lorenzo Fuori le Mura. a dining-hall large enough to contain a city crowd;
By the sixth century, both exposed and concealed we admire walls veneered with a thin layer of mar-
iron tie bars were used in Byzantine architecture in ble, although we know what defects it conceals. We
cover our eyes, and when we have overlaid our ceil-
the east. The method also was adapted to be used with ings with gold, what else is it but a lie that delights us
wooden tie beams, often decorated, which would so? For we know that beneath all this gilding there
have been cheaper and more easily available.38 The lurks some unsightly wood.40
129
P1: KIC/KNP
7
VAULT BEHAVIOR AND BUTTRESSING
R oman builders had an intuitive

understanding of vault behavior and struc-
tural form as seen through their use of lightweight
was an aesthetic principle rather than a structural one,
but when Roman architects and engineers were faced
with determining the form of an arch or the appro-
caementa, vaulting ribs, and iron tie bars. They had no priate wall thickness for a given structure, they would
means of quantifying and calculating vault thrusts, have likely resorted to the same type of proportional
but they had developed ways of controlling behavior system that was used to ensure symmetria. These rules
through long experience with the problems that need only have provided general guidelines, because
could occur. Because most readers do not have the main objective was to determine functional values
the same benefits of firsthand experience of vault rather than aesthetic ones.
construction, I present some basic principles of vault The desire to identify specific geometrical rela-
behavior and examine how the Roman builders tionships or arithmetical formulae used by Roman
and designers developed techniques to control it. In designers can result in overly complex and mislead-
Chapter 8, I then give an overview of the historical ing proposals. Wilson Jones points to the problems
development of the modern understanding of vault that can occur when numbers and geometry be-
behavior and methods of analysis that have been come a primary focus of study and provides an exam-
developed to study it. ple of two different approaches to the design of the
The way Roman architects and builders ap- Pantheon (Fig. 113).2 His own approach is very prac-
proached design would have influenced how they tical. He sees the design starting from a 150-RF circle
determined the appropriate size and form for their that defined the centerline of the exedrae columns.
vaulted structures. M. Wilson Jones has recently ex- Once the column diameters were determined, the
amined the design methods used by Roman archi- builders determined the location of the interior wall,
tects and has pointed out that they typically used then measured out 20 RF to determine the wall
rules based on numerical proportions and/or geo- thickness, which results in an interior wall diame-
metrical relationships, both of which were principles ter of 147.5 RF and an overall exterior diameter of
used by Vitruvius for attaining symmetria, or math- 187.5 RF. In terms of proportional relationships, the
ematical harmony.1 Vitruvius’s concept of symmetria resulting ratio of exterior diameter to column center
130
P1: KIC/KNP
standpoint, rigid relationships based on theoretical

principles are very difficult to translate accurately
onto the building site and are therefore less useful
to the builder, whereas flexible proportional relation-
ships can provide the practical tools for both designer
and builder.
Determining rules of thumb used in concrete con-
struction is complicated by the fact that the builders in
imperial Rome were working largely within a mod-
ular system based on standard brick sizes, so the wall
and vault thicknesses were often determined by the
brick sizes. For example, arches were usually either
one or two bipedales thick, and walls were typically
built in units of whole or half Roman feet, such as
2 RF, 2 1/2 RF, 3 RF, and so on, so that whole or
half bipedales could be used. Any rule of thumb, there-
fore, was not intended to provide the exact measure
but rather a minimum or maximum that could then
be fit into the modular system.5 The result of such
a method is that the final product never conforms
113. Diagram of two proposed design schemes used at the
Pantheon. exactly to any absolute rule that can be precisely de-
termined in hindsight; nevertheless, a careful exami-
diameter is 187.5:150 = 5:4 = 1.25.3 This process is a nation of the structural factors that could have affected
flexible one that could be used on both the drawing the builders’ decisions helps provide some insight into
board and the building site. In opposition to this ap- the design procedures.
proach is a more abstract and theoretical one proposed The stability of a vaulted structure depends on two
by F. Esposito and A. Michetti in which the starting main issues: whether the vault itself can support its
point is seen as the golden section, or pi (= 1.618). own weight without developing cracks that could
In this scheme, the critical design element is the ratio cause it to collapse in on itself and whether the abut-
of exterior diameter (187.5) to interior wall diameter ments are capable of withstanding horizontal thrusts
(147.5) because it is equivalent to the square root of pi from the vault so that they do not spread out and tip
√
(i.e., 187.5:147.5 = 1.618 = 1.27).4 Wilson Jones over (Fig. 114). The first issue is usually not a crit-
points out that both proposals fit the building with ical factor for Roman vaults, which were typically
a reasonable amount of accuracy (1.25 vs. 1.27), but solidly built and in no danger of collapsing. The sec-
neither can be proven without other similar exam- ond issue tends to be the more critical one, which
ples for comparison. As a general principle, I agree is demonstrated by the outwardly tilting walls and
with the more practical and flexible approach, be- the addition of buttresses to many Roman concrete
cause it seems to fit better with the variability that one vaulted structures. In what follows, I examine the
sees in Roman monuments. From a purely practical structural behavior of basic forms of vaults used by the
131
P1: KIC/KNP
embrasure) the less the horizontal thrust. Barrel vaults

used by the Romans usually have an arc of embrasure
somewhat less than 180◦ . For example, the numerous
barrel vaults at Trajan’s Markets range from 104◦ to
170◦ , with many in the range of 150◦ .
The relationship between the thickness of the arch
and the free span has been proposed as a likely rule of
thumb used for arch design in bridges. C. O’Connor
suggested that Roman bridge builders used a ratio
for arch thickness to free span of 1:10 with the un-
derstanding that it could go as far as 1:20.6 The
largest arch in the Roman world is the Augustan
bridge called the Pont St. Martin outside of Aosta
with a free span of 35.6 m, a ratio of arch thick-
ness to free span of about 1:34, and an arc of em-
114. Diagram showing the difference between an arched structure
that fails because of spreading abutments and one that fails because
brasure of 144◦ .7 O’Connor shows in his structural
of arch collapse after hinges develop. analysis of the bridge that an arc of embrasure less
than 180◦ allowed the builders to use safely a thin-
Romans: the barrel vault, the cross vault, the dome, ner arch.8 The three factors thus far discussed are in-
and the semidome. Each type acts on its support terrelated and could be manipulated for different ef-
structure in a different way, and an understanding of fects. He further shows that the concrete infill above
the differences allows for a more precise structural the haunch of the arch provided an added degree of
“reading” of a building and assessment of why the safety. The Roman builders of the Augustan period
builders chose the materials and techniques they did. did not have the tools available to calculate so accu-
rately, but they had developed an intuitive feel for
structural form and understood that variables could
arches and barrel vaults
be adjusted within a certain range to create the desired
The structural behavior of the arch or barrel vault effect.
(which is essentially an extruded arch) is dependent The builders in imperial Rome were working with
on four variables: the arc of embrasure, the thickness of the different parameters than the bridge builders. First,
arch, the free span, and the abutment thickness (Fig. 115). their vaults were supported on walls or, as seen in
The arc of embrasure of an arch (or vault) is the angle Chapter 6, on columns rather than on the fixed abut-
created by the two lines extending from the center ments of riverbanks or hills. Second, by the mid-first
point of the defining arc to the springing point of century they were working with units of predeter-
each side of the arch. It could range from 0◦ for a flat mined brick sizes for the arch thickness, so a compar-
arch to 180◦ for a full semicircular arch. The shallower ison of the ratio of arch thickness to span is not going
the arch (i.e., the lower the angle of embrasure) the to result in consistent ratios as the span increases. The
greater the horizontal thrust on the abutment, and the 1:10 rule may well have provided a limiting factor
more rounded the arch (i.e., the greater the angle of (Table 5), but the builder would have then considered
132
P1: KIC/KNP
At Trajan’s Markets, the wall thicknesses at the lowest

level are 3–4 RF and decrease to 2.5 and 2 RF in
the higher levels. The ratio between wall thickness
and free span in all the buildings is in a range of 1:4.8
to 1:10.9. Only rare examples go over a 1:10 ratio,
and those are surrounded by other vaulted rooms that
provided buttressing, which is an alternative means of
countering the horizontal thrusts of a vault (see later).
The 1:10 ratio proposed for the relationship of arch
thickness to free span may have also been a limiting
factor for the relationship of wall thickness to free
span, but the builders then took other factors, such
as standard wall thickness units and location within
the structure, into consideration. They seem to have
established a safe range of spans in which they could
115. Diagram showing four factors that affect the stability of an
work, and they followed general standardized rules
arched structure.
for determining wall thicknesses.
From the earliest days of concrete vaulting, a
whether he was in the range of a one-bipedalis or a method of providing buttressing was to juxtapose ad-
two-bipedalis arch depending on the arc of embrasure jacent barrel vaulted rooms so that any lateral thrust
of the vault. Moreover, the ratio of vault thickness was countered by that of the next vault. The stretch
to free span focuses on the stability of the arch itself, of thirty bays of warehousing shown on the Severan
that is, its ability to sustain its load without falling in Marble Plan and traditionally attributed to the
on itself, which is appropriate for bridges with fixed Porticus Aemilia employed this principle. Of course,
abutments but less relevant for Roman buildings with the series must come to an end, and unfortunately
freestanding abutments. The more important
factor, therefore, may have been the relation- table 5. Proportions of barrel vaults
ship between free span and abutment thickness. Floor Wall thickness m Free Span m Ratio
The wall thickness (= abutment thickness) Monument level (WT) (FS) FS/WT
for Roman buildings was fairly standardized Domus Aurea 1 0.75 (2.5 RF) 3.9–5.3 5.1–7.1
for average sized rooms with a free span of 0.90 (3 RF) 5.2–9.9 5.8–10.9
4–8 m, with walls becoming thinner on the Trajan’s 1 0.90 (3 RF) 4.3–4.7 4.8–5.2
upper floors, as recommended by Vitruvius. 9 Markets – Aula 1.18 (4 RF) 4.2–5.7 3.5–4.8
Table 5 gives the typical wall thicknesses (in 2 0.75 (2.5 RF) 4.5–6.5 6.0–8.7
whole or half RF) and spans for barrel vaults 3 0.60 (2 RF) 4.3–5.7 7.2–9.5
at various monuments in Rome. The mini- Baths of 1 1.45 (5 RF) 8.0 5.5
mum thickness for a wall supporting a con- Caracalla 2.06 (7 RF) 17.5 8.8
crete barrel vault was 2 RF (i.e., the size of the Basilica of 1 2.95 (10 RF) 24.4 (interior) 8.3
Maxentius 4.45 (15 RF) 24.4 (exterior) 5.5
largest standard Roman brick, the bipedalis).
133
P1: KIC/KNP
the Marble Plan does not preserve either end of the

building to reveal how this was handled. Normally
either the end walls were thickened, as recommended
by Vitruvius,10 or else a series of projecting buttresses
were added.
Another method was used for large spans whereby
a central barrel vault (or later a cross vault) was but-
tressed by a series of smaller barrel vaulted rooms
running perpendicular to it. This was the method
used at the first-century b.c. market buildings at
both Ferentino (4.3-m span) and Tivoli (5.3-m span)
(Fig. 116), although in the latter example semidomes
were substituted for some of the barrel vaults. The
use of perpendicular vaults as buttressing can be
seen at a larger scale in the Nymphaeum Suite at
the Domus Aurea. The largest barrel vaulted room
(13.75-m span), which opened onto the west court,
was buttressed by the perpendicular walls of the three
rooms on either side of the impost walls (Fig. 116).
The way that this vault was constructed is un-
usual in that it was added after the barrel vaulted
rooms to either side had already been constructed.
Exactly how long after remains a contentious is-
sue, but, regardless of the answer, this large barrel
vault was solidly buttressed both during and after
construction.11
cross vaults
The behavior of the cross vault differs from that
of the barrel vault: It is a load concentration sys-
tem that directs both vertical load and the hori-
zontal thrusts to the corner supports. The structural
behavior of the cross vault is more complex than
that of the barrel vault, and some controversy ex-
ists amongst engineers regarding the best way to un-
derstand it.12 For our purposes, however, most im-
116. Drawing showing configurations of vaults used as buttress- portant is that the cross vault produces horizontal
ing elements in two Republican monuments and two imperial
monuments.
thrusts in two directions on its abutments so that
134
P1: KIC/KNP
juxtaposition of vaults and by methods of external

buttressing.
As with barrel vaults, cross vaults can be arranged
so that the horizontal thrusts are partially canceled by
those of the neighboring cross vault. The best exam-
ples occur in the imperial thermae where the hori-
zontal thrusts were resolved by using cross vaults in
the corner rooms to provide some balancing thrusts
in both directions. The corners were then thickened
to provide more substantial piers (Pl. XIII). Because
the horizontal thrusts of cross vaults are concentrated
at particular points rather than spread continuously
as in barrel vaults, the ratio of wall thickness to free
span along the diagonal tends to be somewhat more
conservative than the 1:10 ratio proposed for barrel
117. Diagram showing the lines of compressive force in a cross
vault. Based on Alexander et al. 1977: fig. 8. vaults. A comparison of cross vaults located on ex-
terior walls show that it usually remains under 1:8
the resultant line of thrust acts along the diagonal (Table 6).
(Fig. 117). Because the loads are concentrated at the One of the greatest advantages provided by cross
corners rather than distributed along the length of a vaults was the ability to let more light into a space than
wall, the horizontal thrusts at the corners are typi- was possible with a barrel vault. The cross vaults of the
cally greater than those of a similar sized barrel vault. central hall of the Aula at Trajan’s Markets demon-
The Romans were able to use the structural behav- strate how this was accomplished using a principle
ior of the cross vault to their advantage by judicious similar to the example mentioned earlier in the
table 6. Proportions of cross vaults on exterior walls
Wall thickness Free span Wall thickness Free span

orthogonal orthogonal WTO/ diagonal diagonal WTD/
Monument (WTO) m (FSO) m FSO (WTD) m (FSD) m FSD
Trajan’s Markets 0.75 3.62 4.8 0.92 6.92 7.5

Baths of 1.80 11.0 6.1 6.0 23.0 3.8
Caracalla (Rm 6),
2.50 16.0 6.4 2.5 16.0 6.4
(Rm 19)
Baths of 3.8 14.5 3.8 5.0 18.5 3.7
Diocletian
Baths of 3.8 8.0 2.1
Maxentius–
Substructures
135
P1: KIC/KNP
the rooms, which would have provided buttressing for

the vaults, were detached from it. The solution was to
create arches that transferred any lateral thrust from
the vault over the galleries to the dividing walls of
the rooms (Figs. 118–119). The cross vaults originally
sprang from corbels cantilevered from the travertine
piers along the galleries, but the projecting parts of
the corbels and the lower parts of the vaults were cut
away when the space was built into a monastery. The
support of the vaults on these piers created a precar-
ious situation in which the buttressing was a critical
element in the success of the whole.13 An indica-
tion of the lateral thrusts that were contained by the
118. Aula at Trajan’s Markets (a.d. 106–113). The metal sculpture

in the center is part of a temporary exhibit (1992).
Nymphaeum Suite at the Domus Aurea, whereby the

dividing walls of barrel vaulted rooms were used as
buttressing elements. An added degree of complex-
ity in this structure, however, resulted in a unique
configuration and the introduction of the buttressing
arch. The central hall of the Aula was covered by a
series of six cross vaults (8.5-m span), which were
flanked on either side by six barrel vaulted rooms
(Fig. 116). In order to access the rooms to either
side, open roofed galleries were added at the sec-
ond level. By effectively separating the barrel vaulted
rooms from the cross vault, the designers were able
to allow for light to enter through the lunettes of the
119. Aula at Trajan’s Markets (a.d. 106–113). View down
cross vaults, but at the same time the dividing walls of
gallery overlooking the central hall with the buttressing arches
overhead.
136
P1: KIC/KNP
120. Baths of Caracalla (a.d. 212–216). View of frigidarium of with central cross vault and side buttress
indicated in dashed lines.
buttressing arches can be traced in the outward tilt Aula at Trajan’s Markets, there were no vaulted rooms
(c. 15 cm) of the west wall of the Aula overlooking to which arches could span, so the builders devised
via Biberatica. This is probably due to the plasticity of large freestanding buttresses that were placed along
the concrete, which changed form, or creeped, over the exterior of the vaults between the lunettes. Such
time because of the constant pressure exerted from buttresses are still visible along the exterior of the
the vaults of the central hall. cross vaulted frigidaria (c. 21-m span) of the Baths of
The cross vaults in the Aula at Trajan’s Markets Caracalla (Fig. 120) and the Baths of Diocletian.
were likely inspired by those used in the frigidaria Given the similarity of these later bath structures to
of the imperial thermae, most obviously those at the what is known of the plan of the Baths of Trajan from
contemporary project of Trajan’s Baths. The frigi- Renaissance drawings and the Severan Marble Plan,
darium, or cold room, of an imperial bath building the Trajanic bath complex almost certainly provided
was typically located in the center of the structure, the model for the later structures.14
leaving the edges free for other rooms where light The buttressing scheme of the cross vaulted nave
and solar heat were more important. Nevertheless, of the Basilica of Maxentius combined aspects of
light also was a critical element in the experience of both the Aula at Trajan’s Markets and the frigidaria
the frigidarium, and the designers responded by ele- of the bath buildings: The central nave (c. 30-m
vating the triple-bay cross vaults above the terraces span) was flanked by barrel vaulted rooms, and free-
of the surrounding structures to create clerestory standing buttresses to either side directed any lateral
lighting through the lunettes of the vault. Unlike the thrusts down onto the supporting walls of the barrel
137
P1: KIC/KNP
121. Basilica of Maxentius (a.d. 306–315). View showing remaining buttressing arches for the central
cross vault. Fototeca Unione c/o American Academy in Rome, neg. #5022.
vaults (Fig. 121). The Basilica of Maxentius contains domes and semidomes
both the largest cross vaults (30-m span) and barrel
vaults (24.5-m span) from any ancient structure in The dome exhibits its own unique behavior because
Rome. It was massive in conception and exploited of its double curvature. A dome of cut stone con-
many of the developments that had occurred in pre- struction is essentially like a series of self-supporting
vious centuries including the use of pumice and lattice horizontal rings stacked one on top of the other. If
ribbing in its vaults. In spite of the precautions taken, the dome were sliced vertically, the converging joints
the builders ran into problems when working on such of the voussoirs would allow each slice to stand on its
a massive scale. The deformation at the crown of the own, and if it were sliced horizontally the converg-
barrel vaults has already been mentioned in Chapter 2 ing joints would form horizontal rings in compres-
(Fig. 26, p. 36). Another structural problem was en- sion, each of which could support itself (Fig. 123a).
countered on the west end of the building where the Hence, the cut stone dome is like a three-dimensional
lateral thrusts of the barrel vaults had to be contained arch where each successive ring acts as a keystone to
by adding buttressing arches that spanned the road lock the blocks into place (as long as the support-
running between the Basilica and the Forum of Peace ing wall does not spread). Roman domes of concrete,
(Fig. 122). The original level of the terrain dropped however, are usually built in horizontal layers of un-
substantially at the west end and consequently the shaped caementa laid in an abundance of mortar, so
west wall was much taller than the east wall and re- there are no converging voussoir edges to perform
quired more substantial buttressing.15 the same function as in the cut stone dome.
138
P1: KIC/KNP
of Minerva Medica,” and the Mausoleum of Helena

(Fig. 47, p. 69) have vertical cracks in their lower zones
indicating that the concrete was not able to resist the
tensile stresses that developed in the haunches. Some
repairs to the cracks in the Pantheon dome can be
dated by brick stamps and reveal that the cracks occ-
urred soon after construction,18 so the builders clearly
would have been aware of the phenomenon and
by this period would not have assumed monolithic
122. Basilica of Maxentius (a.d. 306–315). Plan and section show-

ing location of buttressing arch on west side of building.
Unlike a simple barrel vault, a dome also has stresses

occurring in both directions of curvature: meridional
stresses (along the longitudinal lines) and circumfer-
ential hoop stresses (along the latitude lines). Struc-
tural analyses of domes show that both the meridional
and hoop stresses are in compression at the crown but
that tensile hoop stresses develop in the haunches.
In a hemispherical dome, the point of change from
compression to tension occurs at about 52◦ from the
crown of the vault with the tension increasing towards
the base (Fig. 123b).16
Roman concrete vaults have sometimes been at-
tributed monolithic properties because of the use of
pozzolanic mortar,17 but most large Roman domes
and semidomes, including the exedrae at the Baths 123. A: Diagram of construction of dome built of stone vous-
of Trajan (Fig. 4, p. 8), the Pantheon, the “Temple soirs. B: Diagram showing stress patterns in an uncracked concrete
dome.
139
P1: KIC/KNP
table 7. Proportions of freestanding domes
Wall Free Int

thickness span Dia Ext Dia1
Monument (WT) m (FS) m WT/FS RF RF
Pantheon 6.0 43.3 1:7.2 150 188

Tor de’Schiavi 2.6 13.2 1:5.1 46 ∼64
Gordian 1.2 11.4 1:9.7
Octagon
Minerva 2.6 24.8 1:9.5
Medica
Mausoleum 2.4 20.2 1:8.4 68 94
of Helena
1
Diameters taken from Wilson Jones 2000: Table 4.1.
A semidome behaves somewhat differently than a

dome. Because it is not a full circle in plan, it does not
form the horizontal rings that a dome does and there-
fore does not benefit from the structural advantages
offered by the closed circle. The structural deforma-
tion of a semidome through creep or cracking results
in a spreading at the edges accompanied by a drop
in the crown in the crown (Fig. 124b). This pattern
of deformation has been documented in the large
124. A: Deformation pattern of cracked dome. B: Deformation semidomes at the Hagia Sophia.19 So, the semidome
pattern of cracked semidome.
in many ways acts like a cracked dome, but it does
not have the balancing thrusts from the wedges on
properties for their large spanned domes. Once a the opposite side that a full circular dome would have,
dome develops cracks, it results in a series of wedge- which often resulted in the forward tilt at the crown.
shaped arches propped up against each other at the This is one reason that semidomes were rarely built as
crown (Fig. 124a). As long as the abutments do freestanding structures. More commonly, they were
not give way, the dome with radial cracks will rebuilt as appendages to other structures that could pro-
main stable. The critical issue is how the horizon- vide support at the crown. When there was no other
tal thrust at the impost is channeled down through structure built up to the face of the semidome, as in
the structure; therefore, the most important relation- the massive semidomes of the exedrae that were fre-
ship in this context is that of the dome span to the quently built into the enclosure walls of the imperial
wall thickness, which for the surviving freestanding thermae, the edges of the room at either side had to
domes in Rome is in a range from about 1:4 to 1:10 be substantially buttressed to prevent the ends of the
(Table 7). dome from spreading.
140
P1: KIC/KNP
Pantheon dome (Fig. 125). Two main explanations

have been proposed for their purpose. One is that
they were intended to act structurally by increasing
the load on the haunch in order to reduce the hor-
izontal thrust of the vault by countering it with ad-
ditional vertical load. Another is that they were in-
tended to act as devices to facilitate the construction
by allowing the workers to build in vertical incre-
ments rather than to have to shape the steep lower
portions of the dome.20 An examination of the de-
velopment of domes and semidomes suggests that the
first explanation (structural) was the original intention
and that the second explanation (constructional) was
an advantage only exploited later.
The earliest surviving example of a step-ring in
125. Pantheon (a.d. 118–128). View of stair and step-rings on the
Rome occurs on a semidome at Trajan’s Markets
dome.
(Fig. 126), which is particularly instructive because
A method for regulating the outward thrusts of part of the extrados has now been cut away revealing
both domes and semidomes was the use of a series of a section through part of the vault. The step-ring was
step-rings built above the haunches of the extrados. clearly added after the dome was completed, as the
The most famous example is on the exterior of the ring itself is built up against the layer of cocciopesto
126. Trajan’s Markets (a.d. 106–113). View of extrados of large semidome with remains of step-ring.
141
P1: KIC/KNP
provides some insight into the development of the

step-rings.
The dome of the “Temple of Mercury” at Baiae is
the earliest known Roman concrete dome (21.6-m
dia),22 dating probably a century before the semidome
at Trajan’s Markets. The cylindrical rotunda consists
of two concentric walls. The inner side of the outer
wall was plastered before the inner wall and dome
were added suggesting that the dome was not origi-
nally part of the plan. The dome is unusual for such
an early one in that it is pierced with an oculus at the
crown as well as four windows along the haunch on
the diagonal axes. The back (west) side of the dome
is stabilized along the outer haunch by a stair that
runs between it and the hillside and provides access
to the extrados (Fig. 127). The front (east) side of the
1.60-m thick rotunda wall projects up above the sur-
rounding structures such that the haunch of the vault
is buttressed by only a single stairway giving access to
the ledge that leads around to the rear stairway. The
builders evidently decided that additional buttressing
walls were needed and built them after the dome was
127. “Temple of Mercury” at Baiae (late first century b.c.). Detail completed. The awkward placement of these is deter-
of stairs built onto extrados of the dome. mined by the location of the walls between the rooms
below (Figs. 128–129).23 This sequence of the con-
that covers the extrados. The step-ring also covers struction of the Mercury dome suggests that the entire
the stairs that were built into the extrados. The coc- enterprise was an experimental one with decisions
ciopesto covers the stairs in such a way that they re- being made as construction progressed. The effec-
mained useable implying that the dome was standing tiveness of the added buttressing is examined further
for some time before the step-ring was added. How- in Chapter 8.
ever, Trajanic brick stamps in the facing of the step- Another dome that employed a similar type of but-
ring suggest that it was added not long after the vault’s tressing wall but in a more systematic manner occurs
completion.21 This early example of a step-ring added in the octagonal room of the Domus Aurea (Fig. 130).
to a vault indicates that the device was not originally The vaults around the octagonal room were com-
intended to aid in the construction but rather had a bined in a way to create a very clever series of well-
structural purpose. A comparison with methods used lit rooms. It was one of the most inventive uses of
at two earlier domes, the “Temple of Mercury” at vaulting yet created by the Romans and one that
Baiae and the octagonal room at the Domus Aurea, ushered in a new way of thinking about light and
142
P1: KIC/KNP
128. “Temple of Mercury” at Baiae (late first century b.c.). View of dome showing buttressing walls
added to rotunda wall.
space. It also created new structural issues to be re- Once domed and semidomed structures were built
solved. Like the Mercury dome, the octagonal dome freestanding, the builders had to contend not only
at the Domus Aurea was built within other vaulted with stabilizing the vault but also with stabilizing the
structures, the walls of which provided buttressing abutments. The more efficiently the loads were chan-
for the support structure. The most innovative aspect neled down through the structure the less massive
of the design was the way that light was brought in the abutments had to be. At the Mercury dome and
above the haunches of the octagonal vault by means the Domus Aurea octagon, the buttressing walls on
of clerestory windows. As result of the configuration, the extrados of the dome were used to accommo-
the dome had to be quite thin if there was to be date windows in the haunches, which was possible
enough space at its haunches for the clerestory light- because of the support from the surrounding struc-
ing into the adjacent rooms. The structural resolu- tures but also which precluded the use of continuous
tion was a more elegant form of the one employed at step-rings. A freestanding dome such as the Pantheon
Baiae. The octagonal dome was buttressed with a se- was more precarious, and the continuous step-rings
ries of eight triangular piers, each constructed above offered the advantage of spreading the counterweight
one corner of the vault so that the clerestory win- evenly around circumference so that the loads were
dows could fit between them (Fig. 130). This solu- distributed more efficiently throughout the struc-
tion goes a step beyond the one used at the Mercury ture rather than to particular points. Structural anal-
dome and, in a sense, is a precursor to the buttress- yses presented in Chapter 8 show that the use of
ing arches used at the Aula at Trajan’s Markets in the step-rings was one of the most effective means
terms of inventing structural solutions to problems of of controlling the lateral thrusts of the Pantheon
lighting. dome.
143
P1: KIC/KNP
Medica. At the Pantheon, a series of ribs and reliev-

ing arches were built into the thickness of the wall
to channel loads down to eight large piers (Fig. 80,
p. 100). Any lateral thrust was countered by the
6-m thickness of the walls, which has a ratio of 1:7.2
with the span of the dome. Minerva Medica, by con-
trast, is a much less ponderous structure with a higher
ratio of dome span to wall thickness of 1:9.2. The
builders compensated for the wall thickness by adding
external semicircular apses that projected out from the
decagonal abutments (Frontispiece). Semidomes tend
to make good buttresses due to the natural tendency
of the crown semidome to push forward, as noted
earlier. Even within these apses, however, the builders
demonstrate a daring attitude by replacing large por-
tions of the walls with columned arcades. The result
would have been a very light and airy feeling since
the design was stripped to the bare minimum in terms
129. “Temple of Mercury” at Baiae (late first century b.c.). Plan of structural support (Frontispiece). Movements ei-
of dome showing wall locations at lower level (shaded). Buttressing
ther within the structure (creep) or possibly in the
walls (“B”) of dome are positioned so that they align with walls
below. foundations caused enough concern that some time
By the fourth century, the step-rings had further

developed into constructional aids as well as structural
devices. This can be seen at the “Temple of Minerva
Medica,” where the bipedales that cover the top of
each step-ring extend all the way through to the in-
trados and divide the dome into separate layers that
corresponded with the ends of the formwork boards
(Fig. 95, p. 111).24 The buttressing elements that
began as later additions at both the Mercury dome
and the semidome at Trajan’s Markets eventually de-
veloped into a feature that solved both structural issues
and at the same time became integrally connected
with the overall process of construction as seen at
Minerva Medica.
The way in which builders approached issues of
stability over time is demonstrated by a comparison 130. Octagonal room of Domus Aurea (a.d. 64–68). Drawing
of the abutments of the Pantheon to those of Minerva showing clerestory windows and buttressing walls at corners of
octagonal dome.
144
P1: KIC/KNP
ran into problems during construction and added

buttresses that caused a change in the design. The sup-
porting wall of the semidome is quite thin (1.45 m)
compared to its span (17.25-m dia), giving a span/wall
131. “Temple of Venus and Cupid” at Sessorian palace (a.d. 312–

330). Plan and elevation redrawn from Colli 1996: fig. 7 and
Rivoira 1925: fig. 178.
later the arcades in the apses were filled in and large

buttresses were added to the exterior of the structure.
Radial cracks are visible in the remains of the walls and
dome, and once these developed, the lateral thrusts
on the abutments would have increased. This is one
of the clearest cases demonstrating that large Roman
concrete vaults did not display monolithic qualities,
and in spite of the efforts on the part of the builders
to ensure the stability of the structure by using lattice
ribbing in the dome, pumice in the crown, amphoras
over windows, and apses as buttressing elements, they
had to succumb in the end and add the large buttress-
ing elements that marred the elegant exterior of the
original design.
The desire on the part of the fourth-century de-
signers to reduce the support structure to a bare min-
imum also can be seen at the “Temple of Venus and
Cupid” in the present-day garden of the Museo della
Fanteria near Santa Croce in Gerusalemme. This ap-
sidal structure once formed one end of a basilica hall at
132. “Temple of Venus and Cupid” at the Sessorian palace
the Sessorian Palace (Fig. 131). The builders evidently (a.d. 312–330). View of buttress added to the apsidal structure.
Note the use of travertine blocks to anchor buttress to wall.
145
P1: KIC/KNP
thickness ratio of 1:11.9. Moreover, the wall was procedure, but for structural issues they are difficult
originally pierced by five large windows (3.5-m to establish in hindsight because of modifications that
wide) leaving only narrow (1.5-m) stretches of wall could occur to the structure during construction,
between them as supports. The sidewalls were built whereas numerical relationships provide an idea of
very thick (3.3-m) to resist the lateral thrust at the the range in which the builders were working. For
edges of the semidome, but the stretches of wall be- rules of thumb in determining wall thickness, numer-
tween the windows were evidently problematic.25 As ical relationships would also have been less cumber-
at Minerva Medica, buttresses were added later, but some to use on-site, as they do not require a compass
these display a curious detail. The buttress walls were and straightedge. Moreover, the builders in Rome
attached using travertine “ties” to bond them to the were working within a modular system provided by
apsidal wall (Fig. 132). The addition of the buttresses standard brick sizes that would have overridden both
required the closing of the windows with infill (no complex geometrical constructions and precise nu-
longer in place), which also was anchored by means merical relationships. The ratios of wall thickness to
of travertine ties.26 Presumably the travertine ties were vault span listed in Tables 5–7 rarely go above 1:10,
inserted into the wall later as a means of establishing which suggests that the builders were using this ratio
anchors, but a more detailed study of the monument as a rule of thumb to determine the limits within
is necessary before coming to definitive interpreta- which they could safely work. Cross vaults reveal
tions of the construction history of the structure. a somewhat more conservative limit of 1:8 along
The builders of the fourth century had developed the diagonal due to the concentration of loads at
an understanding and control of their materials so the corners.
that they could produce lighter structures than had A technique used by builders from an early time
ever been attempted previously, but the apse of the to control vault thrusts was to balance thrusts of vari-
“Temple of Venus and Cupid” may be an example ous vault forms. The technique was used with barrel
where the limits were pushed a bit too far. Worth vaults from an early period as seen at the Republican
noting, however, is that both the “Temple of Minerva sanctuaries at Palestrina, Tivoli, and Terracina. The
Medica” and the “Temple of Venus and Cupid” were next step was to turn the barrel vaults so that they ran
built on private imperial property where structural perpendicular to the lateral thrust being countered, as
experimentation may have been more acceptable than seen in the barrel vaulted room of the Nymphaeum
in a structure designed strictly for public use. Suite at the Domus Aurea (Fig. 116, p. 134). This
technique was used to provide buttressing for large
spanned vaults. An extension of the principle can be
conclusions
seen in the rotunda wall of the Pantheon, which con-
We have no direct evidence for the rules of thumb sists of a series of eight barrel vaults each running in
used by Roman builders, but they most likely con- line with the thrust from the dome. By the second
sisted of geometrical and numerical ratios. This century, cross vaults were often used in buildings at
chapter focused largely on identifying possible nu- freestanding corners to distribute the thrust obliquely
merical rather than geometrical relationships that as in the imperial bath complexes. The principle of
could have served as limiting factors. Geometrical combining vaults to act as buttressing elements al-
constructions were certainly used as part of the design lowed for the creation of the light structure of the
146
P1: KIC/KNP
“Temple of Minerva Medica.” Unlike at the Pan- the third century. The Pantheon must be seen as the
theon where the vaults were hidden within the walls, apogee of Roman dome construction, but Minerva
at Minerva Medica the apsidal structures themselves Medica represents an attempt to combine many of the
were used as a design element. preceding techniques into the most efficient package
The more innovative examples of buttressing oc- possible. In the end, it was not entirely successful, but
curred when complex juxtapositions required new it shows the extent to which the builders could use
solutions to allow lighting into the spaces. The oc- the accumulated knowledge to pare down the support
tagonal room of the Domus Aurea was stabilized by structure to the ideal minimum.
triangular buttressing walls combined with clerestory Another phenomenon that must have had an effect
windows to allow light into the adjacent rooms. Like- on the Roman desire for more light in their structures
wise, at the Aula at Trajan’s Markets, the first use of was the introduction of window glass. Seneca, writing
the buttressing arch was a response to a particularly in the mid-first century, notes the recent innovation:
difficult situation in which lighting was crucial to the
success of the space. The imperial thermae, which We know that certain devices have appeared only
within living memory – like the use of windows that
were such an important aspect of the imperial so- let in light through transparent tiles, and the baths
ciopolitical agenda, were dependent on the external with raised floors and hollow tubes set into the wall
buttressing arch to allow for the clerestory windows to diffuse heat and maintain an even temperature at
of the frigidaria cross vaults. the lowest as well as the highest levels.27
The three largest surviving domes represent dis-
Indeed, much window glass was found at Pompeii,
tinct phases in Roman dome development. The Mer-
and a workshop producing cast window glass in the
cury dome was not well planned, and the addition of
first century has been excavated at Sentium in the
the buttressing walls suggests that the construction
Marche region of Italy.28 Although difficult to imag-
of large spanned domes was still in an experimental
ine today, the change this created in the experience
phase. The transformation from the Mercury dome
of one’s environment must have been something akin
to the Pantheon dome represents the ultimate move
to the change that occurred with the introduction
from an embedded dome to a freestanding dome. As
of electricity. With the advent of window glass the
with the Mercury dome, however, the experiential
expectations regarding spatial quality increased, thus
aspect of the structure was focused internally. The
producing the social acceptability that provided the im-
transformation from the Pantheon, where the dome
petus for using vault forms such as the cross vault that
was supported on massively thick walls, to Minerva
allow for more light and with it the buttressing to
Medica, where it was supported on a much lighter
keep it standing.
and articulated structure, represents a move toward a
Seneca often uses light in his invectives against lux-
focus on the exterior as well as the interior of the
ury, and he pairs it with bath buildings (as with win-
domed space. Once again, the innovations in but-
dow glass above, albeit in a different context) when
tressing created structures that allowed for much more
he advocates a return to the old-style baths like those
light than was possible earlier. The changes are also
of Scipio Africanus:
representative of the confidence and understanding
of the material that had developed since the second Nowadays, however, people speak of a bath as fit
century a.d. in spite of the intervening turmoil of only for moths unless it has been designed to catch
147
P1: KIC/KNP
the sun through enormous windows all day long, himself with the two gods connected with the sun,
unless a person can acquire a tan at the same time as Sol and Apollo. E. Champlin has recently emphasized
he is having a wash, unless he has views over coun- the role of these gods in Nero’s propaganda,32 and the
tryside and sea. . . . Some people these days con-
demn Scipio as a boor because he did not let daylight connections were not lost on Seneca. In response to
into his caldarium through wide panes of glass, be- Nero’s excesses, he obliquely criticizes Nero’s associ-
cause he did not stew in strong sunlight, daydream- ation with sun god by quoting lines from Ovid:
ing until he was perfectly cooked.29
People seem to think that the immortal gods cannot
Although Seneca is literally talking about private give any better gift than wealth – or even possess
baths, he may well have had in mind Nero’s Baths anything better: “The lofty palace of Sol was set
on the Campus Martius from which came Martial’s with pillars tall and bright with glittering gold,” or
quip “What was worse than Nero? What is better than they describe the chariot of the Sun: “The axle was
Nero’s Baths?”30 The Baths of Nero are difficult to of gold, golden chariot the pole, of gold the rims
of the wheels, and silver the radiating spokes.” And
discuss from an archaeological point of view, as they finally they would praise an epoch as the best, they
were extensively rebuilt by Alexander Severus. These call it the “Golden Age.”33
later baths had the traditional triple cross vaulted frigi-
darium, as indicated by Palladio’s drawings of the Nero used the imagery of light and gold in the Domus
remains, and so they presumably employed buttress- Aurea, although there it did eventually come to lose its
ing arches similar to those on the Baths of Caracalla. luster. Of course, gold only has a visual impact when
Whether Nero’s original bath building employed the it sparkles in the presence of light, so the control of
same technique is difficult to say, but, given the inno- light ultimately conferred the power to disseminate
vative solutions of buttressing in the octagonal dome the message. The prosaic constructional element of
at Domus Aurea, one can imagine that Nero’s Baths the buttress is usually defined as a structure to support
were well lit.31 or reinforce another structure, but during the first
The ability to control and manipulate light through century a.d. it came to be an element that allowed
architecture would have resonated with Nero, who the Romans to control and manipulate light through
both had an active interest in design and associated the medium of concrete.
148
P1: JPK/JPJ
8
STRUCTURAL ANALYSIS: HISTORY
AND CASE STUDIES
S tructural analysis is often viewed by the

nonengineer as a mysterious and magical process
leading to the Truth. Attempts by engineers to de-
fundamental differences provides some basis for evalu-
ating the results. After discussing the historical devel-
opment of arch analysis and the modern approaches
mystify the process have not always been successful, to it, I present a series of case studies intended to
which has led many nonengineers interested in an- demonstrate various ways in which structural analysis
cient structures to the conclusion that if one can sim- can be useful to the archaeologist examining ancient
ply find an amiable engineer competent with a com- buildings.
puter all structural questions can be readily answered.
In fact, much of the basic understanding of arch and
historical development
vault behavior was developed long before computers
became available. In any case, the computer provides The fundamental concepts for understanding and
answers only as accurate as the information entered. predicting arch behavior were only developed in the
Moreover, different approaches can be applied to Renaissance. The most basic one is the concept of
structural analysis, and there is some debate regarding a force, which is an influence that produces an accel-
which approach yields the most useful information. eration of a body or mass. The most common force
My goal in this chapter is to make yet another attempt that we experience every day is gravity because it is
to demystify the basics of structural analysis and to constantly pushing us toward the center of the earth.
present the two major approaches most often applied A force can be represented graphically by a vector,
to the analysis of historical structures, finite element which is an arrow pointed in the direction that the
modeling and thrust line analysis. Each approach force is applied and scaled in length so that it is pro-
makes different assumptions about the structure and portional to the magnitude of the force. For example,
asks different questions; therefore, the answers are a vector representing the force that a human body
not always comparable. The choice of approach de- (Fig. 133) applies to whatever supports it is deter-
pends in part on the question one is trying to an- mined by the weight, or mass, of the body multiplied
swer. For the nonengineer interested in the structural by the standard gravitational force acting on it (which
behavior of historical buildings, understanding the is 9.81 m/sec/sec), and the resulting force is expressed
149
P1: JPK/JPJ
The concepts of forces, vectors, and stresses were

necessary before arch behavior could be explained
scientifically. The Dutch mathematician Simon
Stevin (1548–1620) was the first to represent forces
as vectors.1 Sir Isaac Newton (1642–1727) later pro-
vided a critical framework for working with forces
with his three laws of motion. The Second Law
states that a force equals mass times acceleration,
which is why we had to multiply the weight of
the human (mass) on the beam in the earlier ex-
ample by 9.8 m/sec/sec (acceleration) to convert it
into Sir Isaac’s namesake, the Newton. The Third
Law, which states that for every action there is
an equal and opposite reaction, is fundamental for
establishing the state of equilibrium for structures
133. Diagram demonstrating the concept of the vector.
and, most importantly for this discussion, for arch
behavior.
in units of Newtons (N). So, if the man in Figure 133 Another important concept for understanding arch
weighs 80 kg, he exerts a force of 785 N onto the behavior is moment, which is the tendency of a force
beam. The location of the force vector is deter- to produce movement, or rotation, around a point. A
mined by the center of gravity of the body, which moment is calculated by multiplying the magnitude
in the case of the man would be through his navel. of the applied force times its distance from a fulcrum
The beam then exerts a force equal to the man’s point, or hinge, about which the rotation occurs. The
weight (785 N) plus its own weight (100 kg × distance from the fulcrum to the force is called the
9.81 m/sec/sec = 981 N) onto the two triangular moment arm. A simple example of these concepts
supports for total of 883 N on each one. If we assume applied to the human body can be seen in Figure 134.
that each of the triangular supports in Figure 133 sits A woman holds a bottle in her right hand with her
on top of a column, then each of the columns is un- arm partially extended in front of her. The weight
dergoing stress (compressive), which is defined as the of the bottle applies a force (through gravitational
amount of force applied divided by the area that resists pull) to the end of her arm. The fact that her arm is
the force. So, the stress in the column can be calcu- extended creates a moment arm extending from her
lated by dividing the force exerted by the man and hand to her shoulder joint, which acts as the fulcrum
beam by the area of top surface of the column. For of rotation. If the woman in Figure 134 were to extend
comparison, a column with a 10-cm diameter (= area her arm straight so that distance (d) is greater, her
of 79 cm2 ) resisting a force of 883 N would undergo arm would become tired quicker because the moment
a compressive stress four times greater than a column would increase even though the force remains the
with a 20-cm diameter (= area of 314 cm2 ) resist- same. Stress is what she feels in her muscles as her arm
ing the same force and about 16 times greater than a becomes tired holding the bottle. In an inert material
column with a 40-cm diameter (= area of 1,257 cm2 ). like stone or brick, the stress is resisted by the ways in
150
P1: JPK/JPJ
STRUCTURAL ANALYSIS: HISTORY AND CASE STUDIES
In Figure 133, the beam on which the man stands

bends; when he steps off the beam it will return to
its original form. The fact that it bends (rather than
breaks) is because of the bonds between the atoms in
the material and represents the phenomenon called
elasticity, which is described in terms of stiffness. All
materials have some degree of elasticity but some ma-
terials are much stiffer than others. Stiffness is differ-
ent from strength: Concrete is very stiff and strong in
compression but it is very weak in tension, whereas
wood is less stiff and less strong in compression but
much stronger in tension.3 Because concrete and
stone are stiffer than wood, they will develop cracks
more easily when subjected to tensile stresses.
The places where cracks develop in an arch before
134. Diagram demonstrating the concept of moment.
it finally collapses are predictable. During the eigh-
teenth century, the failure patterns of arches were
which the atoms are bound to each other within the tested empirically by constructing arches and then
material. loading them until they collapsed and noting where
The concept of moment relates to the behavior of the hinges formed.4 These tests showed that the loca-
arches because an arch will tend to fail when a section tions of the hinges were consistent and were directly
of it moves around a fulcrum point or hinge, which
typically forms at a crack or a joint between two vous-
soirs. As long as an arch has only three hinges it will be
stable, as it results in two partial arches propped against
each other. The problems occur once the arch devel-
ops four or more hinges (Fig. 135), at which point
collapse occurs, not because of deficiencies in the
strength of the material but because of the changing
form, or deformation, of the arch. An analogy be-
tween the woman holding the bottle and the three-
hinged arch can be seen in the lower right corner of
Figure 134. As early as the sixteenth century, Leonardo
da Vinci (1452–1519) in his notebooks had drawn di-
agrams showing that he understood the concept of
forces and their moments and their relationship to
arch behavior.2
The point at which cracks develop is determined
both by the strength of the material and by its stiffness. 135. Diagram of stable three-hinge arch (top) and of collapsing
four-hinge arch (bottom).
151
P1: JPK/JPJ
related to a principle published in 1679 by Robert

Hooke in the form of the Latin anagram:
abcccddeeeeefggiiiiiiiillmmmmnnnnnooprrssstttttt
uuuuuuvx
that unscrambles into:
Ut pendet continuum flexile, sic stabit contigum

inversum rigidum
that in turn translates:
As hangs the flexible line, so but inverted will stand

the rigid arch.5
In other words, if one takes a flexible line and hangs

from it weights equivalent to the weights of the sec-
tions of a comparable standing arch, the line will take
the form of a curve, which, in turn, if made rigid and
flipped upright, represents the ideal shape of the arch
capable of carrying those loads in compression.6 The
shape can vary somewhat depending on the length of
the line and the distance apart of the supports, but as
long as the curve can fit within the thickness of the 136. Diagram showing the concept of Hooke’s hanging line.
original arch, then the arch will stand and no ten-
sion will develop. The curve formed by this hanging line of thrust must pass through these hinges, and a
line represents the line of thrust of the arch (Fig. 136). single thrust line can be isolated. The arch then be-
The line of thrust is defined as the internal line of comes statically determinate.
force vectors within an arch due to a particular set In the late seventeenth century, a method for deter-
of applied loads (in this case, the weight of the arch mining lines of thrust in an arch called the funicular
itself). The empirical tests of arch failure showed that polygon method was developed. It involved a tedious
the hinges form at the locations where the line of mathematical process, but in 1866 the German en-
thrust becomes tangent to the intrados or extrados gineer Karl Culmann presented an easier graphical
of the arch. These points of tangency are where ten- means of using this method to establish the thrust
sile stresses develop and cracks occur. When four or lines of arches.7 The funicular polygon method is
more cracks, or hinges, develop, the arch will collapse based on the principle that a stable arch must ad-
(Fig. 135). The line of thrust in an uncracked arch (one here to Newton’s Third Law. So, the force vectors
with no hinges) can take a variety of different paths that make up the line of thrust of an arch must form a
that will fit within the arch. Because there is no single closed polygon if they are to be in equilibrium. This
path that can be determined, the uncracked arch is means that the horizontal and vertical forces are bal-
called statically indeterminate. Once cracks develop, the anced. In Figure 137, each of the forces (F1, F2, etc.)
152
P1: JPK/JPJ
The latter part of the nineteenth century was also

a time of growing interest in analytical ways of study-
ing ancient structures. In 1873, A. Choisy, a French
engineer, wrote his important work, L’art de bâtir
chez les romaines, thereby promoting interest in con-
struction and the analysis of Roman structures. In
1875, A. Leger published Les travaux publics, les mines
et mètallurgie aux temps des Romains, in which he fo-
cused on strengths of ancient materials and on anal-
yses of proportional and geometrical systems of arch
design. By 1904, G. Giovannoni published a thrust
line analysis for the “Temple of Minerva Medica” us-
ing the funicular polygon method,8 and G. B. Milani
followed with a similar analysis of the Pantheon in
1923.9
Interest in the structural analysis of historical ma-
sonry structures has increased greatly over the past
half century, particularly since World War II, when
so many historical structures in Europe were dam-
aged. As a result, engineers have become more in-
137. Example of funicular polygon diagram used to determine
volved in the analysis and preservation of historical
line of thrust in a concrete barrel vault. monuments and have brought with them new ways
of looking at the structures. They have often modi-
fied previously held notions regarding the structural
represents the weight of a section of a concrete vault. behavior of ancient buildings, which had been de-
When these vectors are put together end to end as veloped by archaeologists and architectural historians
in the central diagram and then combined with the who did not necessarily have backgrounds in engi-
oblique vectors making up the thrust line (A, B, etc.), neering (see Chapter 1). During the past decades, a
all the triangles formed must close at point O if the debate has developed regarding the pros and cons of
arch is to be in equilibrium. A detailed explanation is two different approaches: elastic theory versus plas-
provided in Appendix 4. This graphic method of plot- tic theory.10 Elastic theory was developed during the
ting lines of thrust provided a relatively easy means of nineteenth century and plastic theory is a more re-
calculating the stability of arches that relied more on cent twentieth-century development. Each approach
drawing than on heavy number crunching. Moreover, is based on different assumptions, discussed further
it gives a visual representation of the potential thrust later, and a central issue in the debate is which ap-
lines acting within the structure. The method, which proach provides the most useful and accurate infor-
became an accepted method of structural analysis dur- mation for accessing the stability of masonry struc-
ing the nineteenth century, has come to be known as tures. Both offer advantages and disadvantages, and
“graphic statics.” the critical issue in applying them is in asking the
153
P1: JPK/JPJ
appropriate questions and setting up the models to the known yield strengths of the material, the struc-
answer them. Before choosing one over the other, the ture should be stable. The main criticism of using this
student of ancient structures must first understand the method to analyze historical structures is that it was
benefits and limitations of each. conceived as a method for designing new buildings
In elastic theory, materials are assumed to have the in which the criteria were controlled by the designer.
ability to undergo a certain amount of deformation For historical structures, the models are often depen-
when a load is applied and to spring back to the orig- dent on information that is not always knowable given
inal form once the load is removed. This is the con- the heterogeneous nature and degradation of the ma-
cept of elasticity discussed earlier. Each material has terials, the sometimes unclear constructional history,
a different degree of stiffness. The elastic limit, or and the discontinuities between materials introduced
yield strength, of an individual material is measured in by building joints;11 therefore, there is always the dan-
terms of the maximum stress it can undergo before it ger of assuming values that do not represent the actual
can no longer return to its original form. The yield conditions. Even in new construction, this can be an
strength of iron is very high; it can undergo great de- issue. For example, the authors of one recent struc-
formation and spring back to its original form. Wet tural design handbook provide the cautionary tale of
clay, on the other hand, has very low yield strength; an offshore oil platform that collapsed in the North
it can be pulled or pushed into various shapes with- Sea in 1991 after having undergone rigorous stress
out springing back to its original form. At a certain analysis using sophisticated finite element software;
point, however, both iron and clay will actually come small errors in the assumptions made for the model
apart, and this point is called the ultimate strength of resulted in the platform’s failure at a critical location.12
the material. Plastic theory, also called limit analysis, is based on
One of the most common methods of analyzing a different set of assumptions regarding the materials.
historical structures using elastic theory is finite el- It was first used to analyze historical masonry struc-
ement modeling, which is based on 3D computer tures in the late 1960s in the work of J. Heyman.
modeling. A model of the structure is created by di- The basic assumptions for a masonry structure are
viding it into a finite number of elements for which (1) that the compressive yield strength of the materi-
the stresses that develop at their boundaries can be als (stone, brick, and mortar) are so much greater than
quantified. This type of analysis is called stress anal- the compressive stresses that could ever develop in the
ysis. The basic assumptions in finite element model- structure, the materials can effectively be assumed to
ing are: (1) that each material has a quantifiable yield be infinitely strong in compression; (2) that the ma-
strength (expressed in terms of stress) and (2) that de- terials have no tensile strength; and (3) that there is
pending on the strength of its materials the structure no slippage between material elements (e.g., between
can deform to a moderate degree without failing. The the voussoirs of an arch). The goal of this method is
goal of this method is to compute the location and to examine the relationship between form and mass
magnitude of internal stresses within various mate- to determine the point at which the structure will
rial elements resulting from loads on the structure so collapse (i.e., to find the “limit”) and to demonstrate
that the stress levels can be compared to the known whether there is a possible state of equilibrium for a
yield strength of the materials employed. If the com- structure of a given form and load distribution. This is
puted stresses within the structure remain well within sometimes called a static equilibrium model. Because
154
P1: JPK/JPJ
the strength of the material is assumed to be irrelevant, the effects of different parameters on a given structure
the results do not have to be compared to the tested (relative). In the latter, some of the pitfalls of finite
strength of the material, which in the case of concrete element modeling for determining absolute stability
is notoriously difficult to establish. The limit state of can be avoided by setting up ideal models and then
an arch is determined by finding the thrust line that testing the effects of changes. In this way, all vari-
forms a three-hinge arch and then tracing it down ables are responding to the same set of conditions
through the abutments, or support structure. As long so that the relative effects can be quantified and com-
as the line of thrust remains within the structure, it pared. For example, one can create a single model and
will be stable. This method tends to produce conser- then substitute various weights of materials to get an
vative results because the materials do, in fact, display idea of how they affect the stability of the structure.
some resistance to tension, but it eliminates some of The strength of those materials is irrelevant as long
the problems relating to material properties and dis- as they are reasonably strong in compression, as most
continuities encountered in finite element modeling. rocks are.
In sum, plastic theory asks questions regarding the In the following case studies, I provide examples of
appropriate form and mass distribution of a structure both absolute and relative applications of plastic the-
whereas elastic theory asks questions regarding the ory using the graphic method of funicular polygon
strength of the materials (expressed in terms of stress) thrust line analysis. I have chosen to apply plastic the-
used in a particular form of structure. ory, or limit analysis, using thrust lines for a number
In any structural analysis of a building, regardless of of reasons. The first is that it has the advantage of be-
whether it is based on elastic or plastic theory, certain ing based on a concept that is close to the way that the
assumptions are made to simplify the process. Under- Romans thought about their buildings. Romans were
standing these assumptions is important in evaluating clearly thinking in terms of forces acting in particu-
the results of an analysis, which are only as accurate as lar ways on a structure. For example, we know from
the model. One has to be clear on the question being Vitruvius’s advice on how to build an arched struc-
asked before evaluating the success of any particular ture, he was thinking of the arches as “pushing out”
model, as it can never reproduce all the conditions (extruderent).13 Finite element modeling is based on
of the actual building. For those not trained in the testing stresses, which is a far more abstract concept
details of structural analysis, it is easy simply to ac- that the Romans would not have grasped. The second
cept the results of any type of structural analysis as the reason is that thrust line analysis using graphic meth-
truth, but one should be aware that computed results ods is one that a mathematically literate nonengineer
inevitably reflect a certain amount of subjectivity on can apply to simple structures. A step-by-step expla-
the part of the person setting up the analysis. nation of the process is supplied in Appendix 4. In
Another point worth emphasizing is that either many ways, finite element modeling is actually more
type of structural analysis, elastic or plastic, can be flexible in terms of the questions it can ask about a
used in two different ways: absolute versus relative. It structure, but it is not a method that can be readily
can be used to predict the stability of a structure given employed by a nonengineer for even the simplest of
a particular set of circumstances (absolute). This is tra- problems. Moreover, the thrust line analysis provides
ditionally the goal of structural analysis for historical a visual link between the geometry of the vault and
preservationists. It can, however, also be used to test the form of its thrust line, which should provide the
155
P1: JPK/JPJ
reader with a better understanding of the behavior of vault, the solid wall counters it by distributing the
the structures.14 I discovered that the process of per- weight more evenly and is, in fact, the more stable
forming the various analyses provided a “feel” for the of the two conditions. The third condition tested is a
changes in the behavior of the structure as the pa- combination of the two proposals in which a shallow
rameters were manipulated. After performing a few vault as in Amici’s proposal is placed on a colonnaded
of these analyses, I could come much closer to pre- façade as in Packer’s proposal. The result is roughly the
dicting the result before I started. This type of “feel” same as Packer’s original proposal, so the more critical
for structural behavior is precisely what the Roman element to the stability of the structure is the nature
builders developed over time through their manipu- of the support structure rather than the form of the
lation of form and mass on the building site. vault. Nevertheless, as noted in Chapter 6, the mar-
ble entablature elements from the south façade lend
support to Packer’s reconstruction of a colonnade. In
basilica ulpia
all cases, the three sections where the porches on the
The reconstruction of the Basilica Ulpia, which is no façade were located would have been stable, as they
longer standing, has been the focus of much contro- acted as buttressing elements.
versy during the past two decades. As discussed in These analyses, which test the ultimate stability of
Chapter 6, the use of metal ties for the aisle vaults has three hypothetical reconstructions, suggest that the
been proposed but not universally accepted, so I take sections of the south façade wall of the Basilica Ulpia
this as a test case and calculate the thrust line for the between the porches would have required some sort
reconstructions. I use the two most common recon- of resistance to the lateral thrusts to retain stability.
structions for the Basilica Ulpia discussed in Chapter One possibility is that the iron dovetail clamps that
6: J. Packer’s with an angle of embrasure of 180◦ sup- connected the blocks of the entablature were suffi-
ported on a colonnaded façade and C. M. Amici’s ciently strong in tension to resist the outward pressure
with an angle of embrasure of about 152◦ supported once all the pieces were in place and the system was
on a solid wall. I add a third configuration that com- closed. Testing this hypothesis is beyond the parame-
bines elements of both proposals to see how it affects ters set up in these simple models and is an example of
the results (Fig. 138). how stress analysis could be advantageously applied to
A calculation of the thrust line for a 5.27-m stretch determine the levels of stress in the iron clamps. The
of vault (i.e., the intercolumniation distance) shows other possibility is, of course, that tie bars were used
that the thrust line on the innermost colonnade fac- in these areas. I have argued in Chapter 6 that the evi-
ing the nave is directed downward by the weight of dence from the cuttings in the blocks suggests that tie
the roof structure. The central colonnade is balanced bars were used at some points along the south façade.
by the thrusts from the vaults to either side. The south
façade wall is the one that runs into problems. The
“temple of mercury” at baiae
thrust line for neither Packer’s nor Amici’s proposal
remains entirely within the outer support, although In Chapter 7 we saw how the builders of the “Temple
the thrust line on Amici’s does cross within the base of of Mercury” at Baiae set up buttressing walls along
the wall. Even though Amici’s shallower vault gener- the east side of the dome where the rotunda wall
ates more horizontal thrust than Packer’s semicircular projected up above the surrounding structures
156
P1: JPK/JPJ
138. Thrust lines for three proposals of the reconstruction of the south façade of the Basilica Ulpia.
(Fig. 129, p. 144). A question that arises is whether ble. A number of factors could have prompted the
they were necessary. As the lower part of the struc- builders to add the buttresses. Rakob’s study of the
ture is partially built into the hillside and partially shape of the dome shows that it underwent defor-
surrounded by other rooms, the only critical point is mations, and these could have caused outward move-
where the dome projects above the terrace. A plot ment in the external wall, particularly due to creep
of the thrust line of the dome shows that it falls well in the concrete before it had completely cured. The
within the rotunda wall at the joint with terrace level domed hall of the “Temple of Mercury” was actu-
(Fig. 139). Indeed, the construction joints between ally part of a bath building that took advantage of the
the buttresses and the outer rotunda wall suggest that natural hot springs created by the volatile volcanic ac-
the dome existed for some time before the buttresses tivity in the Campi Flegrei on the Bay of Naples. The
were added, although the similarity in construction same volcanic activity that created its raison d’être also
technique suggests that the interval was not long. In was the source of periodic seismic events that could
any case, the thrust line analysis confirms that the have caused movements in the foundations, which are
dome and its abutment wall would have been sta- now under the water level and no longer accessible.
157
P1: JPK/JPJ
provide an overall picture of the factors acting on the

structure.
pantheon
The best known of all Roman domes is the Pantheon,
and as such it has always been of special interest to en-
gineers, a number of whom have conducted various
types of structural analyses on it (Milani 1923, Thode
1975, Mark and Hutchinson 1986).15 In what follows,
I add to this collection by applying a thrust line anal-
139. “Temple of Mercury” at Baiae (late first century b.c.). ysis to the building as a means of testing the relative
Thrust line through dome. effects that various techniques discussed in previous
chapters had (or would have had) on the structure in
Whether the builders were responding to particular a state of minimum horizontal thrust. The approach
circumstances or were just reassuring themselves by I take differs from the two other authors who applied
adding the buttresses is not known, but their addition thrust line analysis, Milani and Thode, in two main
does suggest a level of insecurity that is not seen in ways. First, my intention is to test the relative effects
later large domes. that various techniques had on the structure rather
Structural analysis could provide even further in- than to determine absolute stability. I do this by cre-
sight into the behavior of this early dome. Finite el- ating different models incorporating various parame-
ement modeling could provide a useful complement ters to test how they affect the thrust line. Second, by
to both the photogrammetric study of the deforma- taking all the models to their limit state of minimum
tions of the dome and the thrust line analysis above. horizontal thrust, that is, to the point where hinges
The Mercury dome has four windows built into its develop (which neither Milani nor Thode did),16 the
haunch. Whenever an opening is introduced into the results provide parallel and meaningful comparisons.
fabric of a structure, there are concentrations of tensile The techniques explored include the use of the step-
stress, particularly at corners (which is why we often rings on the extrados and the use of lightweight cae-
find cracks around the doorframes in our homes). For menta in the dome. I also test a hypothetical situation
the calculations of the thrust line in Figure 139, I took in which I add a series of amphoras to the haunch of
a wedge of the dome between the windows so that the vault.
they are not a factor in the calculations. A stress analy- In order to compare the results of the different
sis using finite element modeling could provide addi- models, some method of quantification must be used.
tional information on the patterns of tensile stress that The method can be fairly arbitrary as long as it is
develop in the dome because of the openings. This consistent between the models. I have chosen to use
information could then be compared to the patterns a formula proposed by W. J. Rankine in the mid-
of deformation to see if any obvious correlations ex- nineteenth century for determining a safety factor
ist. Each method of analysis provides a different type for new construction.17 The Rankine factor is easily
of information, which can ultimately be compiled to determined by dividing the width (b) of the abutment
158
P1: JPK/JPJ
meridional cracking. Model P1 is the full model in-

corporating both step-rings and lightweight caementa.
The thrust line is calculated so that it represents
the minimum horizontal thrust; it is tangent to the
140. Definition of Rankine safety factor.
by twice the distance (x) of the thrust line from the

centerline of the abutment (Fig. 140).18 I chose this
formula over other possible ones in part because it
is the basis for the “middle third rule” that is often
cited in studies of vaulted construction.19 The rule
was originally developed in the context of using thrust
line analysis based on elastic theory as rule of thumb
for keeping the stresses to an acceptable level. A fac-
tor of 1 indicates the collapse point of the abutment,
whereas a factor of 3 or above was considered safe.
Important to note is that I am not using it to deter-
mine absolute levels of safety but rather as a numerical
basis for comparing the relative effects that different
techniques have on the stability of a structure.
All the Pantheon models tested are based on the as-
sumption that the primary relieving arches/ribs join-
ing the eight piers of the rotunda are effective in trans-
ferring the load of the dome to the piers; therefore,
each model assumes that one pier is carrying a 45◦
wedge of the dome, which is one eighth of the entire
dome (Fig. 141).20 Also inherent in this assumption is
that no hoop tensions develop in the dome, which in
fact replicates the actual state of the vault due to the 141. Pantheon (a.d. 118–128). Thrust line through 45◦ section
of dome under different conditions (P1-P5).
159
P1: JPK/JPJ
extrados at the crown and to the intrados at the builders were wise only to use the lightweight mate-
haunch. The thrust line is contained within the pier rial at the crown. Model P4 was created by reversing
and exits in the middle quarter at ground level for a the parameters and eliminating the lightweight ma-
Rankine factor of 4.23 (Fig. 141, bottom). This con- terial at the crown to create an overall heavier dome.
firms that a single pier would be able to contain the This produced a more significant change. The result-
thrust of one eighth of the dome transferred to it by ing thrust line moves outside of the middle quarter,
the system of relieving arches/ribs built into the wall. thereby reducing the Rankine factor by 12 percent to
The analysis does not confirm whether or not the rib 3.6. Therefore, the careful use of heavier caementa in
actually does this, but it does show that the building the haunch and lighter only at the crown had a sig-
would be stable if it did. nificant effect on the structure, but ultimately not so
The Pantheon does not employ amphoras in its significant as to make the difference between failure
haunch, but in Model P2, I examine what effect the and stability.
addition of 127 Dressel 20 amphoras to the haunch of Model P5, which is P1 with the step-rings re-
the vault would have had on the structure. The result- moved, reveals that the step-rings have a much greater
ing thrust line was barely affected, but the amphoras effect on the thrust line than the lightweight caementa.
did serve to reduce the Rankine factor by 1 percent When the step-rings are removed, the Rankine factor
from 4.23 to 4.20, so the net result is a negative one. is reduced by 37 percent to 2.6. When the builders
The addition of the amphoras to the haunch of the chose to add the step-rings, they were reducing the
vault actually increased slightly the horizontal thrusts. “pushing out” effect described by Vitruvius by adding
The haunch is where the weight needs to increase to weight to push down. They were not thinking in
get a “pushing down” effect rather than to decrease. terms of vectors or thrust lines, but conceptually they
In fact, adding them would have had a virtually neg- were thinking in a similar way: The “pushing out”
ligible effect on the line of thrust. These results re- could be countered by “pushing down” in the form
inforce the idea that the builders did not typically of adding extra weight or surcharge.
intend for the amphoras to act as structural elements A comparison of these results with those of the
but, rather, were using them as a means of saving on stress analysis using finite element modeling per-
material and labor costs. formed by Mark and Hutchinson is useful in setting
To test the effectiveness of the lightweight caementa out the advantages and the limitations of each ap-
in the crown of the vault, I manipulate the weight of proach. Mark and Hutchinson used a series of sim-
the materials in Models P3 and P4. Model P3 was cre- plified finite element models of the domed structure,
ated by taking P1 and substituting the brick and tuff each with a different set of assumptions geared toward
caementa used in the haunches with the lightweight evaluating the effectiveness of the use of graded cae-
caementa used in the crown to create an overall lighter menta and step-rings. They then compared the lev-
dome. This resulted in a 3 percent decrease in the els of tension present in the various models, which
Rankine factor to 4.1. This result shows that the were divided into two groups. One group used an
overall lightness of the dome is not the critical is- uncracked dome so that tensile hoop stresses could
sue, because the lighter dome would have been less develop, and another group used a cracked dome
stable. The critical point is that the caementa be graded that simulates the actual state of the dome. They
in the appropriate places to control the thrust. So the discovered that the use of graded caementa was an
160
P1: JPK/JPJ
effective method of reducing hoop tension in the with different points of reference. The finite element
uncracked dome. An advantage of using elastic the- models give levels of tensile stress within the structure,
ory and finite element modeling is that tensile hoop which must then be compared to the tested resistance
stresses can be tested, which is not possible using plas- of the materials, whereas the thrust line analysis gives
tic theory and thrust line analysis because of the initial a picture of overall stability. The thrust line analysis
assumption of no tension. In the uncracked dome, cannot provide information on the difference in be-
they found that the addition of the step-rings actu- havior between the cracked and uncracked dome, but
ally increased the circumferential hoop tension from the finite element model generates information that
0.5 kg/cm2 to 0.6 kg/cm2 , both of which are within must be compared to material properties, which are
the tensile capacity of good quality Roman concrete, not always known. Each, therefore, has its own ad-
which is estimated in the range of 15–25 kg/cm2 .21 vantages, disadvantages, and pitfalls.
However, when the model was allowed to develop
cracks so that the dome consisted of a series of wedge-
“temple of minerva medica”
shaped arches (as in the thrust line analyses) the re-
sults were reversed. With the cracks, the tension along By the fourth century, the approach to building
the meridians of the wedge-shaped arches more than domes had changed, as is especially evident in the re-
doubled. So, for example, in the model without step- mains of the “Temple of Minerva Medica,” which
rings, the tensile stresses reached 1.3 kg/cm2 , whereas employs brick ribbing, step-rings, lightweight cae-
in the model with the step-rings, only small, local- menta as well as a less massive supporting structure
ized tensile stresses occured at the inner corners of than the Pantheon. To understand better what effect
the rings.22 These results caused them to question these factors had on the structure, I applied a similar
how and why the cracks developed in the first place, type of thrust line analysis to the original phase of
given the low levels of tension. They concluded that the building, which was later reinforced with further
they could have spiked to higher levels as a result of buttressing.
rapid temperature changes, such as occurs when a Two of the major differences between the Minerva
rain hits a hot concrete surface and causes rapid cool- Medica dome and Pantheon dome are the use of brick
ing. Because the cracks occurred during or soon after lattice ribbing and thinner abutment walls at Minerva
construction, which is clear from patches containing Medica, so in creating Model MM1, I chose the point
Hadrianic brick stamps, shrinkage during the cur- in the structure where there was a rib and where
ing of the concrete is also another explanation. The the buttressing was minimal – the decagon corner.
results, therefore, caused the researchers to think be- The thrust line is calculated for a 7.5◦ wedge of the
yond the model to other factors such as process and dome that corresponds with the corner buttresses of
environment and acted as catalyst for developing new the decagon. The results in Model MM1 show that
insight into the factors affecting the structure. the line of thrust just fits within the structure so that
The final results of both the finite element anal- it is tangent to the extrados near the crown and to
ysis and the thrust line analysis support each other the intrados at the haunch (Fig. 142). As the thrust
in that both confirm that the lightweight caementa line continues down through the structure, it remains
and the step-rings increased the stability of the dome, within the thickness of the abutment with a Rankine
but the results are reported in different “languages” factor of 3.6.
161
P1: JPK/JPJ
in the Rankine factor to 5.1. At first glance, these

results suggest that the ribs actually had a deleterious
effect on the stability of the structure; however, the
eighteenth-century drawings of the building, which
show the ribs standing alone in places where the rest of
the vault has fallen, suggest that the ribbed sections of
the dome were structurally more sound than the un-
ribbed sections. The illustration in Figure 143 shows
that the part of the dome that fell corresponds pre-
cisely with the parts at the crown made with pumice
caementa. One explanation for this phenomenon is
that the concrete with caementa of pumice alone can-
not be assumed to have infinite resistance to compres-
sion, which is one of the assumptions of limit analy-
sis. The resistance of a stone is in part a function of
the relationship of weight to volume,23 so the lighter
the material the less resistant it is likely to be. Actual
tests on the strength of the pumice or the pumice
mortar mixture have not been conducted. Another
possible explanation is that the parts made of pumice
were more susceptible to water infiltration because of
the increased vesicularity and the damage that occurs
from the expansion during periods of freezing. This
would have occurred only after the cocciopesto protec-
tive layer had deteriorated over time. In either case,
the brick lattice ribs would have played a critical role
in reinforcing the less resilient parts of the dome.
Nine of the ten sides of Minerva Medica were but-
tressed by apsidal structures, so in Model MM3, I re-
move them to see what effect they had on the struc-
ture. Models MM1–2 were based on an assumption
that the corner piers were supporting only the cor-
responding 7.5◦ wedge of dome while the remaining
142. “Temple of Minerva Medica” (first half of the fourth cen- 28.5◦ comprising one tenth of the decagon would
tury a.d.). Thrust line through dome under different conditions.
have been buttressed by the apsidal structures. In or-
To test the effect that the added weight of the ribs der to test the buttressing effect of the apsidal struc-
had on the thrust line, in Model MM2, I removed the tures, the dome wedge is increased to 36◦ so that
ribs and left the step-rings and lightweight caementa. each of the ten piers supports a full one tenth wedge
Removing the ribs resulted in a 40 percent increase of the dome. In this case, the Rankine factor drops
162
P1: JPK/JPJ
143. “Temple of Minerva Medica.” Drawing by Franz Innocenz Kobell (1780) before the dome fell in
1828. Fototeca Unione c/o American Academy in Rome, neg. #9079F.
to 1.4 in the piers; therefore, the apsidal structures factors, such as the settlement of the foundations,
played an important role in ensuring the stability of gradual leaning of the abutments over time, or earth
the structure by relieving the corner piers of some of tremors, upset the equilibrium that they had so care-
the horizontal thrust. fully created.24
In terms of an absolute application of the analy- The “Temple of Minerva Medica” is a very com-
sis, the results confirm that the original structure was plex structure and still has many secrets to reveal
stable but precariously susceptible to external factors. through further study. Finite element modeling has
The second phase of massive buttressing appears to great potential for providing a better understanding
have been added within at least a couple of decades of the relationship between the parts. The thrust line
of the original construction. Small movements in the analyses presented above are useful in asking questions
foundations could have easily caused problems to de- about particular parts of the building, but they do not
velop. In terms of a relative application, the analy- reveal how the various elements interact. More inves-
sis provides insight into the advantages of using the tigation into the behavior of the apsidal buttressing
brick ribs and raises questions about the potential elements of the first phase, which were quite dar-
problems of using pumice caementa. The builders at ing in the use of colonnaded rather than continuous
Minerva Medica pushed the form and materials to the walling, could provide additional information on the
limits and perhaps discovered that other unforeseen issues that prompted the addition of the buttressing
163
P1: JPK/JPJ
144. “Temple of Diana” at Baiae (late second century a.d.[?]). View of preserved half of dome.
in the second phase. Moreover, the Minerva Medica over the haunch, could play a significant role in the
dome is not a continuous dome but rather has groins stability of the abutments of vaulted structures. How-
at each of the ten corners. These breaks in the form ever, the use of graded caementa alone was not enough.
are places where stress levels tend to change. The work To be effective, it had to be placed judiciously within
of R. Mark and others on stress patterns in the groin the structure so that it related to the curve of the
vaults in medieval cathedrals has demonstrated how vault: light at the crown and heavy over the haunches.
stresses are concentrated in these types of vaults.25 By contrast, the use of amphoras in the haunch of
Stress analyses of models based on the reconstructed vaults actually had a slightly negative effect on the
form of Minerva Medica could help provide a bet- thrust line, pushing it closer to the outer edge of the
ter understanding of this masterpiece of late antique abutment. This is contrary to the common assertion
architecture. that lightening the dome at the haunch would have
reduced the horizontal thrusts on the structure.26
Only lightening at the crown helped reduce the
conclusions
horizontal thrusts. Interestingly, one of the latest
The thrust line analyses presented above provide documented examples of Dressel 23 amphoras occurs
insight into the effectiveness of various vaulting tech- at Santa Maura, where they are added to crown of the
niques discussed in this study. The use of lightweight vault (Fig. 59, p. 83). The addition of ribs that reach
caementa, particularly tufo giallo della via Tiberina, into the crown of a dome could also have a detrimen-
scoria, and pumice in the crown of a vault was an tal effect on the thrust line of the structure by adding
effective means of reducing the horizontal thrust. weight at the crown, but as shown with the dome
The controlled use of caementa graded into different at Minerva Medica, the ribs also had the beneficial
weights, especially when combined with a surcharge effect of reinforcing vaults that employed material of
164
P1: JPK/JPJ
low compressive strength or low durability such as the The buildings that remain today are usually the
Sabatini pumice. The results also show that the use more successful ones, but undoubtedly there were
of this type of limit analysis, which assumes infinite examples that “pushed out” too much and collapsed.
strength in compression, must be used with caution The builders would have learned empirically what
when dealing with structures using lightweight today we can begin to re-create on paper (or on
caementa. a computer). Though the Romans never developed
Carefully devised programs of structural analyses the concept of thrust lines, they appear to have un-
can provide a means of reproducing the behavior that derstood by the third century a.d. that the most
a structure would have displayed under different load- efficient arch or vault was not defined by a circle
ing conditions and with various forms. Any structural but by a parabolic curve as in Hooke’s hanging line.
analysis is going to idealize the building to some de- The remains of the late second-century a.d. dome
gree, as there are always unknowable variables. Stress of the “Temple of Diana” at Baiae (29.8-m dia) was
analysis is more susceptible to misplaced assumptions, built in a form that more closely approximates a
but it also is more flexible in some ways if applied parabolic thrust line (Fig. 144). J. Heyman observes
carefully. The advantage of the thrust line analysis is that “ . . . the problem of design is essentially one of
that it provides a means of reproducing visually what geometry. The calculation of stress is of secondary
the Romans were experiencing on the site as they interest; it is the shape of the structure that governs its
built. Ultimately, both can be of use to the archaeol- stability.”27 The importance of structural form is cer-
ogist, as long as the questions are carefully crafted to tainly a concept of which the Roman builders were
the method of investigation. well aware.
165
P1: JPK
9
T he focus of the preceding chapters has been

on the innovations in the use of materials
and construction techniques involved in the cre-
increased understanding no doubt came from years of
experience with materials. Another, less direct, influ-
ence is that of military technology, which manifested
ation of large and technologically advanced concrete itself largely in the use of timber construction and in
vaulted structures in imperial Rome. The period from metallurgy for making tools and connectors used on
Augustus to Constantine is one in which the Roman the building site.
world underwent great transformation, the nature and The quality of the mortar gradually improved from
causes of which are often the subjects of debate.1 As the second century b.c. to the first century a.d.
a means of generating an overview and putting the as builders gained a better understanding of how
conclusions into context, I employ the four criteria to process the materials for the best effect. At first
for technological innovation described at the end of they depended on pozzolanella, which could be eas-
Chapter 1: (1) accumulated knowledge, (2) evident need, ily quarried in the open air. The gradual shift by the
(3) economic possibility, and (4) cultural/social/political first century b.c. to pozzolana rossa and nera, both of
acceptability.2 In what follows, I use these four criteria which had to be quarried underground, suggests a
to explore some of the most salient issues involved in more selective approach toward materials. Contin-
understanding the changes that affected vaulted con- ued improvement in mortar quality came with bet-
struction during the three and half centuries under ter processing so that the inert soil, which was often
investigation. mixed in with the pozzolana quarried underground,
was removed to create a higher percentage of reac-
tive material in the mortar. The growing awareness of
accumulated knowledge
the importance of distinguishing inert material from
Perceptible changes can be seen in the way builders the pozzolana is reflected in Vitruvius’s advice that
approached vaulted construction during the imperial it must be sharp angled so as to make a noise when
period including the understanding of the properties rubbed between the fingers and that it should be clean
of materials, of centering construction, and of the ef- enough not to leave a stain when pounded in a white
fect of form and mass on vault behavior. Much of the cloth.3 By the end of the first century a.d., the mortar
166
P1: JPK
had a very clean appearance with the granules of poz- techniques were factors that allowed for a more effi-
zolana standing out clearly against the white lime ma- cient design of the wooden centering in late antiquity
trix (Pl. VIII), which suggests that the pozzolana was when materials and labor were scarcer.
being washed and sieved before use. The extreme The choice and placement of caementa underwent
hardness of the mortar in the buildings of the late first changes that indicate a growing understanding of the
and early second century was probably also in part be- relationship between form and mass in controlling
cause of a more developed infrastructure for supplying vault behavior. By the Augustan period the builders
lime so that similar stones were fired together at op- distinguished between different weights of caementa
timum temperatures and times (Chapter 3). Finally, for concrete, and scoria had already been used as
the importance of keeping the mortar wet during the caementa in vaulting. However, the systematic use of
curing process is mentioned by Dio Cassius in the lightweight caementa in vaults only appeared under
third century a.d.4 As neither Vitruvius nor Pliny Vespasian with the construction of the Colosseum,
mentioned this aspect in their earlier writings, Dio’s where caementa of tufo giallo della via Tiberina were used
addition may reflect the increased awareness of the for all the vaults, and small pieces of white pumice
importance of water in the curing process. were added to the mortar mix in the upper-level vaults
As concrete vaulting became larger and more com- (Pl. V). In the second century, the concept was ex-
plex, the carpentry required had to meet the new panded to include the use of caementa of Vesuvian
demands. The centering of cross vaults and segmen- scoria mixed with the tufo giallo as a means of light-
tal domes required more intricate centerings than did ening the vaults. In the Basilica Ulpia, the scoria/tufo
the earlier barrel vaults. Improvements in metallurgy giallo mix was used throughout the vaults, but in the
and tool design also may have had some effect on the Pantheon and the Hadrianeum, it was only used in
efficient production of the boards and timbers used the crown. The decision to concentrate the lighter
to build the centering. F. Rakob’s photogrammetrical materials in the crown of the vault suggests a growing
analysis of the “Temple of Mercury” at Baiae reveals understanding of ways of controlling outward thrust.
anomalies in the form of the dome that suggest the As shown in the thrust line analyses in Chapter 8,
builders had not yet mastered the construction of such the greater the difference between the weight of the
a large centering structure in the Augustan period. In crown (light) and the haunch (heavy), the less out-
comparison, J. J. Rasch’s study of the fourth-century ward thrust there is on the support structure.
domes in Rome reveals a sophisticated coordination The change in the use of amphoras in vaults from
of the wooden centering with the construction of a constructional aid to a structural innovation is an-
the external concrete elements (Chapters 2 and 7). other example of accumulated knowledge. In Chapter 4,
Such coordinated work was probably the result of in- I argue that the use of amphoras in vaults as means of
creased efficiency on the building site through long- controlling vault behavior is a late development. The
term experience. Moreover, the ladder and lattice rib- earliest use of the amphoras in vaults occurred dur-
bing from the third and fourth centuries suggests that ing the Hadrianic period and had little to do with an
the ribbing was related to the construction of the attempt to control thrusts. Their appearance in both
centering in some cases (Chapter 5). Thus, the accu- the walls and the vaults at the Villa della Vignaccia
mulated knowledge of the material properties of wood indicates that they were initially used as space filler
and concrete as well as advances in tools and joinery to save on materials. The technique was probably
167
P1: JPK
adopted from land reclamation projects where the windows, and apsidal vaults below the windows pro-
reuse of amphoras had been common for hundreds viding buttressing so that light could enter much in
of years. The use of the amphoras in the vault of the the same way the clerestory windows of a basilica
“Temple of Minerva Medica” in the fourth century, worked. The weights of the materials used in the
however, displays a much more calculated application dome were distributed in the most efficient man-
of the amphoras in conjunction with pumice. Am- ner and consolidated with ribbing. The thrust line
phoras were placed in the haunches of the vault only analyses in Chapter 8 show that the builders were re-
above the windows in an apparent attempt to direct markably accurate in gauging their limits, although
the weight of the vault away from the openings by cre- changes to the structure, probably due to founda-
ating less dense areas above. This is similar in concept tion settlement, necessitated that the original struc-
to the age-old practice of building relieving arches ture be buttressed with less elegant exterior additions.
over windows, but here the amphoras were used in Nevertheless, the structure demonstrates a remark-
a new manner that reflects a sophisticated attempt at able understanding of the balancing of forms and the
controlling vault behavior. strengths and weaknesses of individual materials that
In addition to manipulating the distribution of mass was the result of more than five centuries of previous
within structures, the Romans eventually managed experience.
to attain a reduction in mass. During the Augustan
period, vaults were supported on thick walls or ar-
evident need
cades, as in the portico at the Basilica Aemilia, but
by the second century a.d., porticos were supported The evident need for larger and more complex con-
on much lighter colonnades as at the Basilica Ulpia crete vaulted structures was created by the demand
and the Baths of Trajan (Chapter 6). The use of iron for larger gathering spaces for an increasing popula-
dowels, clamps, and tie bars, which were easily con- tion in the capital and for new architectural forms that
cealed, allowed for the change in aesthetic. Instead of responded to well-established social institutions.5 The
the ponderous structures prevailing during the first increased scale and complexity of such buildings re-
and second centuries b.c., lighter support structures quired that new techniques be developed in response
consisting of colonnades that made the vault seem to to new types of problems that arose, such as deal-
float above them were often the goal. The “Temple ing with noncongruent plans in multistoried build-
of Minerva Medica” represents the clearest attempt to ings, stabilizing complex combinations of structural
create a light and airy vaulted space by reducing the elements during and after construction, and cover-
support structure to the bare minimum. The builders ing very large spaces. At a more prosaic level, the fires
combined a variety of different methods: lightweight that continually plagued the city first created the need
pumice, buttressing vaults, step-rings, amphoras, and for more fireproof structures than had been provided
vaulting ribs. It is practically a compendium of all by timber roofs, and then later they provided both
the developments that had taken place in the three opportunity and incentive to rebuild older structures
previous centuries. Although clearly inspired by the using increasingly advanced technology.
scale and centrality of the Pantheon, Minerva Medica The need for larger gathering spaces affected the
was stripped down to the least massive support struc- development of vaulting in two ways. On the one
ture possible. The upper walls were pierced with large hand, a structure such as the Theater of Marcellus
168
P1: JPK
or the Colosseum provided space for large groups by helped to achieve it. In the case of the bath build-
combining relatively small vaults in a complex man- ings, another goal, less obviously related to vaulting
ner to create efficient circulation for large numbers technology but nevertheless important, was to pro-
of people while at the same time providing great vide the setting to display the varieties of colored
expanses of seating. The complexity of the inter- stones available from conquered territories, a subject
connections of such vaulted passages required new that was the object of one of Seneca’s conservative
techniques such as relieving arches and vaulting ribs criticisms.7 The demand for monolithic colored stone
(Chapter 5), both to provide support for noncon- columns created the need for some method of stabi-
gruent floor plans and to aid in the construction as lizing colonnades that supported concrete vaults, the
the complex forms were built. On the other hand, result of which was the development of iron tie bars.
there also was a need to provide large covered gather- The most systematic use of the tie bars occurred in the
ing spaces, especially for bath buildings, where con- colonnades surrounding the palaestrae of the imperial
crete vaulting was a critical element. It was a material thermae. The necessity for a concrete vaulted roof in-
more suited to the moist environment than was wood, stead of a simpler wooden roof evidently came from
and it eventually allowed for larger spanned spaces the desire to provide terraces from which to view the
that could accommodate more people. The increase activities in the open palaestrae below, as suggested by
in scale represented by the imperial thermae required the figural mosaics decorating the palaestra terraces at
new vaulting techniques, and not surprisingly the sys- the Baths of Caracalla.
tematic combination of advanced techniques is first Destruction of monuments by fire in ancient
evident in the Baths of Trajan, the earliest of the series Rome was a recurring problem, but it also created
of imperial bath buildings for which physical evidence the evident need for reconstruction, which resulted in
remains. There the builders used imported Vesuvian a number of the most impressive structures discussed
scoria to lighten the crowns of vaults (Chapter 3), tie in this study. Examples include the Basilica Aemilia
bars to stabilize vaulted marble colonnades (Chap- (fire in 14 b.c.), the Domus Aurea (fire in a.d. 64),
ter 6), and buttressing arches to stabilize the elevated the Baths of Trajan (fire in a.d. 104), the Pantheon
cross vaults of the central frigidarium (Chapter 7). (fires in a.d. 80 and a.d. 110), the Colosseum (fire in
The two cultural institutions that had the greatest a.d. 217), the Basilica Julia (fire in a.d. 283), and the
effect on the advances in vaulted technology during Basilica of Maxentius (fire in a.d. 307).8 Fire often
the imperial period were public bathing and pub- had the effect of virtually clearing the slate so that
lic entertainment. Both were institutions that went emperors and their builders had the opportunity to
back to Republican roots, but they were comman- have another go at a major public monument using
deered during the imperial period as a means of pla- more up-to-date techniques. One of the most crit-
cating and instilling a sense of pride in the Roman ical points in the development of the urban fabric
citizenry, as sarcastically noted by Juvenal in his fa- of Rome came after the fire of a.d. 64 under Nero,
mous quip about the Roman people’s interest only when large portions of the city had to be rebuilt.
in “bread and circuses” (panem et circenses).6 Creating Understandably, the fear of fire was uppermost in
large-scale, impressive spaces that provided an effi- the minds of both the emperor and the population,
cient means of dealing with great numbers of people and concrete vaulted construction offered an obvious
was one goal, and developments in concrete vaulting hedge against similar occurrences in the future. We
169
P1: JPK
mostly hear about the rebuilding campaign of Nero of Statilius Taurus. Most important were the build-
himself with his Domus Aurea, but Tacitus also em- ings of Agrippa, including the baths that bore his
phasizes the effort that went into the reconstruction name, the completion of the Saepta Julia, the Aqua
of the areas not taken up by Nero’s new palace.9 There Virgo, and the original Pantheon. So, the Augustan
was a tremendous amount of more mundane building period was one of transition with elements remaining
going on that provided economic potential for work- of the Republican system of taxation and senatorial
ers and redemptores as well as a chance to increase the munificence combined with the massive wealth of
pool of experienced labor, both free and slave. The a single person. For the first time, there was an in-
destructive episodes in the history of the city provided dividual whose wealth was greater than that of the
both patron and builders the opportunity to upgrade state, as implied by Augustus’s comment that four
the architectural character of the main public monu- times he contributed to the aerarium when funds were
ments, and most often this resulted in the use of the needed.10
latest concrete vaulted technology. Under Augustus, the supply of building materials
was consolidated, particularly for metals and decora-
tive stones. High-quality iron from Noricum was be-
economic ability
ing exploited, and the marble quarries at Luni were
Building large and technologically advanced struc- in full production. With the conquest of Egypt, the
tures required funds and material resources, and with Mons Claudianus and Mons Porphyrites quarries,
the advent of the imperial system economic ability in- which would become important sources of granite
creased. During the late Republic, the financing for and porphyry, came under Roman control. During
public works in Rome came from the aerarium, the the first century a.d., the state control and organi-
state treasury, which was filled through taxation and zation of the quarries producing the most coveted
from the generosity of its most prominent citizens. stones increased, as much recent work has shown.11
Under Augustus, this system continued to exist, but The acquisition of marble, particularly for imperial
it was augmented by the vast personal wealth of the use in Rome, became economically feasible due to
princeps himself. Augustus accumulated his wealth by the organization of trade and infrastructure. This in
various means including his inheritance from Julius turn created a preference for colored marble revet-
Caesar, from spoils of war, and from his own patri- ment and columns. The influx of marble into Rome
monium, which would have included the legacies that during the first century a.d. came at time when con-
accrued from his slaves, freedmen, and friends. Other crete was becoming more common, particularly after
less salubrious methods of acquiring funds included the fire of a.d. 64, and the result was a merging of
the confiscation of property of murdered enemies the two materials to create a new aesthetic. By the
and people condemned for treason (bona damnatorum). early second century, one finds colonnades of Mons
Many building projects including the Forum of Claudianus granite supporting vaults employing both
Augustus and the Theater of Marcellus were built iron tie bars and imported scoria from Vesuvius at the
with August’s own money, although he still encour- Basilica Ulpia.
aged other wealthy members of the senatorial class to Nero was the most ambitious builder of the Julio-
sponsor public buildings that would bear their name, Claudian successors of Augustus, but his profligate
such as the Theater of Balbus and the Amphitheater lifestyle and the catastrophic damage caused by the
170
P1: JPK
fire of 64 required that new methods of raising funds the imperial coffers, and his building program in
be devised. The cleaning up and rebuilding of the Rome reflects that success.
city after the fire was expensive as were Nero’s vi- Waging wars as a means of acquiring wealth of-
sions of grandeur for his new city. He expanded on ten was a double-edged sword regarding the financial
established methods of enriching the state such as the health of the state. The upkeep of the army and the
confiscation of bono damnatorum but applied it on an waging of war usually offset any gains from war booty.
unprecedented scale. His most creative initiative was In fact, very few major monuments can be directly
in his manipulation of the monetary system. The sil- linked to the spoils of war, although the Forum of
ver denarius had lost about 1 percent of its value by Augustus and the Colosseum are among them.16 The
slight reductions in weight since the time of Augustus, last and largest of the building projects advertised as
but after 64 Nero imposed a 10 percent weight reduc- funded from the spoils of war was Trajan’s Forum,
tion. Moreover, he was the first emperor to reduce which once contained inscriptions noting that it was
the purity of the silver content (4.5 percent) in order built ex manubiis.17 The Dacian gold and silver mines
to stretch even further the silver supply. Both meth- captured by Trajan were soon under imperial control
ods of debasement were to be used consistently by as attested by an inscription listing one of Trajan’s own
future emperors in need of more funds.12 freedmen as a procurator.18
Vespasian, taking over power after the civil wars In the early third century under the Severans, ma-
of a.d. 69, used his own building program much as jor changes in the structure of the imperial treasuries
Augustus had to help establish legitimacy. However, occurred as well as in the methods of raising funds.
after the excesses of Nero and the civil wars, he in- Like Vespasian, Septimius Severus came to power as
herited an empire in dire financial straits. Vespasian the victor of a civil war and inherited a treasury de-
was known for his shrewd fiscal policies, and to raise pleted by an irresponsible predecessor, Commodus.
money he used many of the same methods as his He resorted to massive confiscations of the proper-
predecessors plus some new ones. He both created ties of his rivals, which may relate to his reorgani-
new taxes and enforced old ones that had fallen into zation of the treasury system to accommodate the
disuse.13 His concern for increasing imperial finances new property.19 Using both Augustus and Vespasian
also can be seen in his reforms to the imperial trea- as models, Severus used building programs in Rome
sury. Since the time of Augustus, there had been the to help establish his legitimacy, and he advertised this
distinction between the traditional state treasury, the on coinage with a seated Roma surrounded by the
aerarium, and the private funds of the emperor, the fis- legend RESTITVTOR VRBIS.20 He succeeded in
cus, which had come to play an ever more important replenishing the treasury and left his sons with a flush
role in state finance.14 Under Vespasian, new fisci ap- budget, but it was not enough for Caracalla, who had
peared, among which were the fiscus Alexandrinus and an ambitious new bath complex planned as well as
fiscus Asiaticus, which contained funds from Egypt and wars to wage. Caracalla instituted one of the most
Asia. A. M. H. Jones argued that these new fisci may profound enlargements of the tax base up until that
represent Vespasian’s attempt to bypass the aerarium point by extending the Roman citizenship through-
by creating new treasuries under his own control that out the empire.21
contained the surplus production from these areas.15 After the death of Alexander Severus, the last of
Vespasian was remarkably successful in replenishing the Severans, the third century saw a great decline
171
P1: JPK
in building in Rome as funds were directed toward material resources required for the construction of
the constant warfare. After a.d. 284, when Diocletian large and technologically advanced building projects
brought stability back to the empire and to Rome it- were considerable and only under the imperial sys-
self, he instituted great changes that must have affected tem were the financial and material resources available
the ways in which public construction in Rome was for such an extended period so that the technology
funded. The regular sources of funds throughout the of vaulted construction had the time to develop as
imperial period had been based on taxation. The con- it did.
fiscations, debasements of coinage, and spoils of war
were simply sporadic additions to the basic method of
social/cultural/political acceptability
taxation. During the third century when the coinage
became severely debased, there was a gradual shift The idea of social/cultural/political acceptability cannot
toward taxation in kind and the requisitioning of la- be entirely separated from evident need: The need
bor. Diocletian formalized the system when he re- for baths and entertainment buildings was ultimately
organized the provinces and the system of taxation. determined by cultural expectations. Interestingly,
Evidence from the Theodosian Code and a letter of for concrete vaulting, the transition from need to
Symmachus indicates that during the fourth century acceptability seems to have occurred largely outside
there was a complex trading of goods and services in of Rome during the Republican period so that by
central Italy in order to supply the capital with lime the time of Augustus when the economic ability was
and wood.22 The increased use of taxation in kind established, the acceptability factor had largely been
and requisitioned labor would have had a significant overcome. Vaulted technology had already been used
effect on the structure of the building industry in for theaters, amphitheaters, and baths as can be seen
and around Rome. The sudden popularity of using from the remains at Pompeii. So, when Pompey man-
amphoras in vaults and the switch to local pumice aged to build the first permanent theater in Rome
for lightweight caementa during the early fourth cen- in 55 b.c., the builders could already draw on estab-
tury are probably related to such changes (Chapters 3 lished techniques. With regard to stylistic and formal
and 4). acceptability, however, the development of vaulted
With regard to the economic ability to finance im- bath buildings was critical. In this context, vaulted
perial building projects, the critical point is that the spaces provided a new and different type of interior
emperor increasingly had ways of acquiring funds environment that was experienced in a social set-
even during difficult economic times, and many of ting. As F. Yegül pointed out in the introduction to
these methods were not ones available to leaders un- his book, Baths and Bathing in Classical Antiquity, the
der the old Republican political system. Some of the bath building provided a context for architectural ex-
methods were adaptations of earlier ones used in both perimentation that lay outside the purely utilitarian
public and private contexts during the Republic, but (terracing or storage) or the purely traditional (tem-
over time and through periods of unrest and dubi- ple or basilica).23 Over time, these forms became an
ous leadership, the attitude of both emperor and cit- accepted style that was then applied to other more
izens toward what was acceptable was modified in traditional building types during the imperial period
relation to the changing demands. The funding and once the resources were consolidated.
172
P1: JPK
Concrete vaulting was a uniquely Roman inven- on concrete vaults at the Domus Tiberiana (Chap-
tion and as such it offered a certain amount of ide- ter 3), perhaps in imitation of the Hanging Gardens
ological potential. Strabo noted that, in contrast to of Babylon. With the romantic notion of an ele-
the Greeks who focused on beauty and the founding vated garden came the very real technological issue of
of cities, the strengths of the Romans were their ad- waterproofing the support structure, as demonstrated
vances in the areas of engineering, such as roads, aque- by surviving descriptions of the construction of the
ducts, and sewers covered with arches.24 Had he been Hanging Gardens at Babylon. The most detailed ac-
writing a century later, he might well have added con- count is given by Diodorus Siculus (mid-first century
crete vaulting to that list. Strabo’s observation about b.c.), who based his description on that of Ctesias of
the difference between the Greeks and the Romans Cnidos (c. 400 b.c.), a Greek physician in the court
comes immediately after he has described the natu- of King Artaxerxes II of Persia:
ral assets of Rome, especially the access to rivers for
commerce. In contrasting the Greeks and Romans in There was also, beside the acropolis, the Hang-
this way, there is the subtext that the Romans went ing Garden. . . . The park extended four plethora
[400 feet] on each side, and since the approach
beyond what nature had to offer with their engineer-
to the garden sloped like a hillside and the sev-
ing works – rivers are made by nature but roads and eral parts of the structure rose from one another
aqueducts are made by man. Both the influence of tier on tier, the appearance of the whole resem-
and the competition with the Greek world were ma- bled that of a theatre. When the ascending ter-
jor forces in the developing mentality of the Romans races had been built, there had been constructed
beneath them galleries which carried the entire
during the late Republic. This was one of the factors
weight of the planted garden. . . . The roofs of the
that inspired the construction of the concrete vaulted galleries were covered over with beams of stone six-
sanctuaries in Roman Italy at Palestrina, Tivoli, and teen foot long, inclusive of the overlap, and four feet
Terracina (Chapter 1). All are examples of man’s con- wide. The roof above these beams had first a layer
quest and shaping of his natural environment, which of reeds laid in great quantities of bitumen, over
this two courses of baked brick bonded by cement,
ultimately translates into a sense of power. The same
and as a third layer a covering of lead, to the end
mentality that inspired these creations no doubt also that the moisture from the soil might not penetrate
lay behind the developing concept during this same beneath.27
period of what eventually became the Seven Wonders
of the World.25 As discussed in Chapter 3, excavations at the Do-
The major wonders of the world, which were char- mus Tiberiana provide some evidence for how the
acterized by both size and grandeur, captured the waterproofing of the vaults was accomplished in
imaginations of the cultures living on the Mediter- Rome by combining raised floors of bricks (like those
ranean, and they often served as models and inspira- used in the hypocausts of bath buildings) and water-
tion for Roman emperors. Nero recreated in Rome proof mortar (cocciopesto) to create a barrier to protect
at least one of the Seven Wonders when he set up a the concrete vault below (Fig. 44, p. 60). Whether or
colossal bronze statue of himself in imitation of the not the technical information recorded by Diodorus
Colossus of Rhodes, an analogy that was not lost on Siculus inspired the Roman builders in their own re-
Martial.26 He also built elevated gardens supported sponse to the challenge is difficult to say. They were,
173
P1: JPK
however, drawing on established bath technology in they did defy nature, and the development of concrete
devising a method. Some hint that the designers and vaulting helped make the job easier.
builders may have been directly influenced by ac- The Seven Wonders represented a collection of
counts of the Hanging Gardens of Babylon, at least monuments for which size was paramount, but the
in conception, may be inferred from C. Krause’s re- focus was largely on exterior grandeur. The Romans
cent work on the metrology of the Neronian part of took the idea and shifted the emphasis to the interior
the Domus Tiberiana, which has revealed that the by creating grand spaces covered by concrete vaults,
substructure also was 400 RF long on the side fac- the curving form of which brought the added benefit
ing the Sacred Way.28 This is the same dimension of celestial analogies. From at least the first century
agreed on by all the ancient authors who described b.c., the idea of vaults, and particularly domes, was
the Hanging Gardens of Babylon, which implies that used to invoke an image of the heavens.32 For exam-
Nero may have at least been attempting to match it ple, Varro describes his aviary as a tholos covered by a
in scale. hemisphere (hemisphaerium) with two stars that circled
Nero’s tutor, advisor, and in-house Stoic, Seneca, it by mechanical means, thereby invoking celestial im-
laments about the ostentation of such displays when agery in his description. Elsewhere, in his discussion
he wonders: of the origin of the word “caelum,” or sky, he gives an
example attributed to Ennius (second century b.c.)
Do not men live contrary to Nature who grow fruit of a description of a cavern as “the enormous arches
trees on the top of a wall? Or raise waving forests of heaven” (caeli ingentes fornices). The epitome of the
upon the roofs and battlements of their houses –
the roots starting at a point to which it would be idea comes with the Pantheon, as is shown in Dio
outlandish for the treetops to reach?29 Cassius’s comment on its name when he says: “But
my own opinion of the name is that, because of its
The object of his complaint may well have been vaulted roof, it resembles the heavens.”33 The celes-
Nero’s hanging gardens, or horti pensiles, on the tial theme of the dome is accentuated by the oculus,
Palatine. The ability to create horti pensiles represented which captures the sun’s beam on a clear day. The
a freedom from traditional constructional restraints unusual number of twenty-eight coffers per row may
and followed the Roman tradition of creating a man- bring in a lunar theme as well. The Pantheon dome
made topography as an expression of power.30 The was by far the largest unreinforced dome ever con-
comments of Tacitus indicate that the genius of Nero’s structed, and it employed a sophisticated combination
architects/engineers Severus and Celer was seen as of materials and vaulting techniques, each of which
the ability “to create a semblance of what Nature had been used earlier but never in such an audacious
had refused,”31 and this was precisely what annoyed combination. In a way it stood as a symbol of the ex-
Seneca. The use of raised brick floors in conjunc- panse of the empire itself, which was then at its widest
tion with vaulted substructures allowed the Roman extent after Trajan’s victory in Parthia. Similar themes
builders to manipulate the natural environment, that related the expanse of heaven to empire had been
which was surely what Nero intended in his residence used before, albeit in a more direct manner. The most
on the Palatine. However much Seneca laments the renowned example is the allegorical figure of Caelius
move away from the natural order of things, hanging on the breastplate of the statue of Augustus from
gardens were considered wondrous precisely because Prima Porta. At the Pantheon, however, the themes of
174
P1: JPK
world domination and power were expressed through they could certainly put the products from their land-
spatial manipulation and technological prowess rather holdings in buildings, in the form of bricks. This is a
than allegorical figures. case in which construction technology was develop-
The emperors took advantage of grand vaulting hand in hand with personal, social, and economic
ing schemes for their own political purposes, but connections.
these same schemes also affected the people who The leaders of the brick industry during its initial
experienced the imperial messages and who often period of expansion were the brothers Cn. Domitius
were involved in their production. These ranged from Tullus and Cn. Domitius Lucanus, who in a.d. 59
the senatorial classes to the freeborn poor to the had inherited brickyards from their adopted father
freedmen and slaves. The massive building projects Cn. Domitius Afer, a well-known orator from Nimes
touched the lives of every class of people in the city (cos a.d. 39). Both gained favor with Vespasian, Tullus
and at times had a great influence on social mobility. commanding a legion in North Africa in a.d. 71/72
For example, J. DeLaine, in her analysis of the con- and both brothers sharing the consulship in 73. They
struction of the Baths of Caracalla, has shown that no doubt also provided great quantities of bricks for
the cost of the construction of the baths would have the construction of the Colosseum during the 70s, al-
fallen in the same order of magnitude as largesse dis- though admittedly brick stamps from this initial phase
tributed by Septimius Severus, and almost 80 percent of construction are not abundant enough in the mon-
of the cost went to the workforce.34 Such a con- ument to verify the brothers’ contribution.35 More-
struction project could therefore represent a signifi- over, the Colosseum employs the unusual addition
cant economic boon for the people in Rome and its of small pieces of pumice, which preliminary anal-
surrounding territories from where the brick, lime, yses suggest may have come from the east side of
pozzolana, and building stone came. Lago di Bolsena (Appendix 3). This is the general
One of the most obvious segments of the construc- area where the Domitii are known to have owned
tion industry where the economic effects on individ- property, though such a direct connection is impossi-
uals can be traced is in the brick industry. A proso- ble to confirm.36 We know from a letter of Pliny the
pographical analysis of the names found on the brick Younger of the vast wealth accumulated by Domitius
stamps reveals that both male and female members of Tullus by the time of his death in a.d. 107.37 The
the senatorial class were heavily involved in the indus- property of Tullus went to his niece, Domitia Lucilla,
try since they owned the land on which the clay beds who then married and produced a daughter, Domitia
were located. In the early days of brick production Lucilla Minor, who in turn was the mother of the fu-
in Rome during the first century a.d., the landown- ture emperor, Marcus Aurelius. Thus, through inher-
ers (domini) were independent (albeit wealthy) private itance, the vast majority of the brick figlinae came into
owners, but by the late second century the industry the hands of the emperor by the end of the second
had become an imperial monopoly. As seen in Chap- century.
ters 2 and 5, the late first century and early second The players lower in the hierarchy of the brick in-
century was a time when vaulting techniques using dustry, that is, the brick makers (officinatores) can also
bricks, such as brick ribs and brick linings, were de- be traced through brick stamps. The Domitius broth-
veloped. After the time of Augustus, the senatorial ers had at least twenty-one known slaves working for
elite could no longer put their names on buildings, but them. By tracing their careers, we can get a glimpse
175
P1: JPK
into the effect that the brick industry had on the peo- office were obliged by the emperor to invest a third
ple involved in boom times. Numerous cases show of their capital in property in Italy.43 One way of ex-
that an officinator started off as a slave but then gained ploiting the new required landholdings was to invest
his freedom, possibly buying it from the proceeds in the clay beds for brick making, and during this
added to his peculium. We know of one slave of the time of extraordinary building activity, such an in-
Domitii, Agathobulus, who also had his own slaves vestment would have been a very lucrative one. The
(vicarii), Trophimus and Aprilis. Agathobulus earned use of more bricks in building projects would have
his freedom in a.d. 115, taking his former vicarii as been beneficial for both the domini and the officinatores.
his own slaves. He then manumitted them by 118 and Likewise, the disappearance of the brick linings after
123, respectively, so that all three former slaves were the Severan period (Chapter 2) and the substitution
eventually working as independent officinatores.38 In at of lattice ribbing for solid brick ribbing (Chapter 5),
least one example from the Trajanic period, we find came at time when the brick industry was in decline,
an officinator who seems to have become a dominus as shown by the lack of brick stamps from the period
himself.39 Women were also involved in the brick after the death of Caracalla. When the brick industry
industry: Nearly 30 percent of domini and 6 percent was renewed under Diocletian, it was concentrated
of officinatores were women.40 Women of senatorial in the hands of the emperor, and many of the senato-
birth were more likely to inherit brickyards as prop- rial elite were no longer involved. Over time, fewer
erty, but even the low born were offered some earning slaves were involved, so the brick industry provided
potential in the industry. The brick industry, supply- less potential for social advancement for at least some
ing one of the main construction materials for the segments of society.44 This overview of the people in-
concrete vaulted structures in Rome, affected the lives volved in the brick industry and its changes over time
of people from all social classes regardless of gender. provides some idea of the tangential links between the
The changing patterns that one can trace in the in- development of vaulting techniques using large num-
formation on the brick stamps are roughly reflected bers of bricks and the effect this had on individuals at
in the changes in vaulting techniques found in the various levels of society. No doubt, other aspects of
buildings. Brick linings were first used for vaults dur- the building industry provided similar opportunities,
ing the Trajanic period, during which time there was which today remain untraceable.
a dramatic increase in the number of brick produc- By the fourth century, the social, cultural, and po-
ers. Almost 25 percent of the known types of brick litical context in which the building industry was op-
stamps were in use during Trajan’s reign compared erating was very different than it had been. The fund-
to a little less than 10 percent from the second half ing and acquisition of materials for public building
of the first century,41 and the number of named figli- had changed with Diocletian’s reforms. Moreover,
nae increased from nine during the period a.d. 80–97 under the tetrarchic system instituted in a.d. 293,
to twenty-nine during the period a.d. 98–113.42 H. Rome’s central role in the administration of the em-
Bloch has related the increased numbers of figlinae to a pire diminished, and the focus of imperial building
phenomenon noted by Pliny the Younger in the first was often diverted to new administrative centers, such
decade of the second century in which land prices as Milan, Trier, Nicomedia, and Thessalonica. Rome
around Rome went up because candidates for public was no longer a strategic point from which to oversee
176
P1: JPK
the empire, but it nevertheless remained the ideolog- (Chapter 6). The techniques, however, were often
ical center. After the Forum Romanum was damaged slightly modified to respond to the changes in the
by fire in a.d. 283, the rebuilding of the symbolic cen- building industry.
ter of the city was critical for Diocletian to demon- Aside from the Baths of Diocletian, the major
strate the rebirth that he intended to bring to the buildings constructed ex novo in Rome during the
empire. The main vaulted structure involved in the tetrarchic period are associated with Maximian’s son,
reconstruction was the Basilica Julia, the aisle vaults Maxentius, whose power base actually lay in Rome
of which were built using imported Vesuvian scoria. itself because of the unusual circumstances of his
This came about seventy years after the last known rise to power.47 After the abdication of Diocletian
use of the imported scoria at the Baths of Caracalla and Maximian in a.d. 305, the two former Caesars,
and represents a significant effort toward the renewal Galerius and Constantius, were promoted to Augusti,
of vaulted technology after the hiatus of the mid-third and Maximinianus (no relation to Maximian) and
century. Severus were made the new Caesars, thus passing
The major new addition to the cityscape of Rome over the sons of both Maximian and Constantius
involving advanced vaulting technology was the im- (i.e., Maxentius and Constantine). Severus took over
perial bath complex on the Viminal dedicated to Dio- Italy but, on October 28, 306, the people in Rome
cletian. The structure was, in fact, built on the initia- revolted when census officials were sent to Rome
tive of Maximian Herculius, the Augustus in control to register the citizens for a poll tax on the urban
of Italy, in honor of Diocletian as noted in the ded- populace, and the Praetorian Guard responded by
icatory inscription.45 Maximian, who was based in proclaiming Maxentius emperor.48 Both Severus and
northern Italy, had never been to Rome before he Galerius tried to take back Rome, but Maxentius tri-
stopped there on his way back from a successful cam- umphed and retained control of the city. His power
paign in Africa in a.d. 298, during which time he ini- base and center of command, therefore, lay in Rome
tiated work on the baths. H. Bloch in his study of the itself. He advertised his commitment to the city in his
brick stamps from the Baths of Diocletian argued that coinage from the mints of Rome and Ostia with the
the brick industry was probably reorganized specifi- legend CONSERVATORES VRBIS SVAE.49 Con-
cally to supply the bricks for this project.46 Neither trol of Rome was the source of his legitimacy, and he
Diocletian nor Maximian spent much time in Rome, used architecture to emphasize the point. Concrete
but the effort put into the construction of the Baths vaulting had been well established as the hallmark of
of Diocletian was clearly a symbolic gesture follow- the capital city during its peak, and it made a state-
ing the lead of emperors who had ruled during more ment of established longevity and traditional contin-
stable times. It represented an established means of ex- uation that could not be made in the same way with
pressing symbolic power as control of resources. The less durable materials.
bath complex was modeled on the earlier imperial By the time of Maxentius’s ambitious building pro-
thermae built by Trajan and Caracalla and employed gram, the funding and supply of material and labor
all the advanced techniques that had by then be- for imperial works had radically changed from ear-
come standard: lightweight caementa (pumice) (Chap- lier times, and the economic and social benefits that
ter 3), vaulting ribs (Chapter 5), and iron tie bars had come with large construction projects were no
177
P1: JPK
longer spread as wide as they had been. Large build- vast territory, referred to as her property ad duas lau-
ing projects, which had once functioned as a means ros (“at the two laurels”) outside the walls extending
of largess, came to be seen more as economic bur- from the Via Praenestina to the Via Latina, on which
dens than boons.50 In the case of the brick industry, her mausoleum was ultimately built.54
the figlinae had become part of the imperial prop- Near the Sessorian, and possibly even part of it,55
erty, and there was a virtual monopoly of brick pro- was one of the most impressive vaulted structures
duction by the state under the tetrarchy.51 During of the early fourth century – the decagonal pavil-
this period, the great senatorial landowners, like the ion known as the “Temple of Minerva Medica” –
Domitii during the first and early second centuries, yet the ambiguity in its dating makes it one of the
were no longer benefiting from providing bricks for most enigmatic as well. The brick stamps found on
state projects as they once had.52 Even if the construc- its dome are post-Diocletianic and could have been
tion of a project like the Baths of Diocletian or the put in place either under Maxentius or Constantine.56
Basilica of Maxentius did not provide the same social Regardless of who conceived the initial project, the
and economic benefits to various levels of society as structural modifications made to stabilize the struc-
earlier, it was still an important symbolic gesture for ture were most likely made under Constantine. The
a recovering empire. building was fitted with pipes for water and hypocaust
Constantine’s defeat of Maxentius at the Milvian floors like the triclinium of the Domus Flavia on the
bridge in a.d. 312 marked a turning point in the ide- Palatine, suggesting that it was most likely used as a
ology of building in the city of Rome and the begin- dining pavilion.57 Whoever built the structure was
ning of the decline in the construction of large vaulted thinking boldly in terms of exploiting the most ad-
public monuments. Economic factors played a role vanced vaulting technology of the day, but signifi-
in the change but, as shown by Maxentius’s building cantly it was built in the private sphere rather the
program, the knowledge existed and the supplies and public, although by this time many public functions
labor could be mustered if the priorities were high had begun to occur in ostensibly domestic settings.58
enough. The move away from building very large Constantine’s largest and most enduring projects
vaulted structures ultimately was because of a change constructed ex novo were his ecclesiastical buildings.
in the cultural climate that came with Constantine. In terms of purely civic buildings employing concrete
He did not particularly like Rome, but it was the site vaulting, he completed the Basilica begun by Max-
of his victory in 312 and the official capital of the em- entius and had a new bath complex on the Quirinal
pire so it could not be ignored. He spent very little dedicated in his name. However, if Steinby is correct
time there himself, and thus the incentive to build in her analysis of the brick stamps, it may well have
himself residential accommodations in the form of been a project initiated by Maxentius and only com-
palace/villa architecture, as Maxentius had done, was pleted by Constantine, much like the Basilica.59 The
lessened.53 However, his mother, Helena, who lived real emphasis of his building program was the devel-
in Rome from at least 317 until her death around 330, opment of an ecclesiastical architecture necessary for
functioned as the primary imperial representative in his new Christian empire. Building religious monu-
Rome and may well have been an influential person ments to helpful deities was an old pagan tradition
in encouraging his building schemes. In addition to that Constantine used to create a new architectural
the Sessorian palace within the walls, she owned a language, and with a few exceptions this language
178
P1: JPK
was one that, in its initial stages, used timber roofing foundations in Rome. It was located on Helena’s
instead of concrete vaulting. property at the third mile of the Via Labicana.62
The model for the main Christian building type The structure was built up against a cemetery basil-
was the timber-roofed basilica, and R. Krautheimer ica positioned over the catacombs of Marcellinus and
has pointed out that speed of construction was an Peter, thus creating a type of mausoleum/cemetery
important consideration in the choice.60 As seen in complex that has been found elsewhere on the out-
Chapter 6 with the Basilica Ulpia, basilicas could, skirts of Rome. Some scholars have suggested that
in fact, incorporate concrete vaulting, but doing the Mausoleum was originally intended to house the
so involved extraordinary means to ensure stability remains of Constantine himself before the founding
through the use of lightweight caementa and most of Constantinople, where he was ultimately buried.63
likely tie bars. The Basilica Ulpia was the most direct It has received much less attention than the later and
model for the earliest Constantinian basilicas of San well-preserved domed mausoleum at Santa Costanza
Giovanni in Laterano (a.d. 313–318) and San Pietro (11.4-m span) on the Via Nomentana. However, in
(a.d. 320–329) at the base of the Vatican hill, but spite of the more interesting interior spatial qualities
they did not replicate its concrete vaulted aisles. For a of Santa Costanza, which consisted of a domed cen-
building type that could be built quickly throughout tral space supported on an arcade and surrounded by
the empire, forms requiring extraordinary technol- an annular barrel vault, the dome of the Mausoleum
ogy and special materials were to be avoided (although of Helena (20.2-m span) was almost twice as large and
they would later reappear when speed was not such an thus provided a more grand interior space than the
issue). Timber roofing was easier, faster, and widely later building. Moreover, evidence from the Liber Pon-
available. The grandeur of the interior space could tificalis suggests that the complex on Helena’s prop-
then be achieved through decorations and furnish- erty ad duas lauros was one of the most important and
ings. In the give and take of establishing priorities, richly endowed of the early Constantinian projects
the symbolic longevity of concrete was replaced with in Rome. A comparison of the donations of silver
gold, silver, and porphyry. and gold liturgical vessels places the Mausoleum and
The notable exception to Constantine’s timber the adjoining basilica of SS Marcellino e Pietro sec-
roofed church architecture was the domed imperial ond only to the Lateran and ahead of those of Santa
mausoleum of his mother, Helena. The monument Croce at Helena’s Sessorian palace.64
can be securely dated to the period between 312 and The form of a domed imperial mausoleum was
326 and is the latest of the buildings treated in this an established type that developed during the tetrar-
study.61 An exploration of the significance of this chy. The Mausoleum of Diocletian at Split and the
structure and its vault serves to illustrate the symbolic Mausoleum of Romulus at the Villa of Maxentius
significance that the dome retained even after other on the Via Appia both consisted of domed rotun-
types of vaulting went out of fashion. das fronted by a projecting pediment in imitation of
The Mausoleum of Helena is not well preserved the Pantheon.65 The use of the Pantheon as a model
today, and in its own day it was soon overshadowed reflected a change both in the attitude toward the im-
by the basilicas of San Pietro at the Vatican and San perial succession as well as toward the funerary cele-
Paolo Fuori le Mura; however, when it was built it was bration itself. Under the First Tetrarchy the emperors
among the most important of the early Constantinian claimed to rule by divine authority,66 so the choice
179
P1: JPK
of a temple, especially a temple to all the gods, as Eusebius’s use of it to describe the resting place of
model for the new tomb type was appropriate. The Constantine’s soul there alongside God:
Tetrarchs, who considered their rule as divinely sanc-
. . . [our thought] reaches upwards to the very vaults
tioned, saw themselves as arriving at their rightful
of heaven, it pictures there too the thrice-blessed
place amongst the gods on their death.67 The dome soul in the presence of God, stripped of all mortal
and its celestial symbolism, therefore, accorded well and earthly attire, and brilliant in a flashing robe of
with the assumption that the emperor automatically light.71
(without the approval of the Senate, which had been
With the growth of the cult of the martyrs, the no-
an earlier requirement) assumed his place in heaven.
tions of temple, tomb, and church began to overlap.72
With the breakdown in the continuity of the im-
The Mausoleum of Helena was evidently used for
perial succession during the third century and the
liturgical ceremonies, and excavations and restora-
dispersion of the emperors throughout the empire
tions of the remains provide some idea of the sump-
outside of Rome, the imperial mausolea came to have
tuous interior decoration in which they took place.
different functions from the prototype of the impe-
The walls were revetted in colored stone including
rial dynastic mausoleum, that of Augustus. Diocletian
porphyry, and the dome was covered in glass mosaic,
specifically avoided the idea of dynastic succession,
tesserae of which have been found in blue, green,
having no son himself. During the first two centuries
and gold leaf.73 Among the furnishings listed in the
of the empire, imperial funerary ceremonies were
Liber Pontificalis are also enormous silver chandeliers.74
outside events focused on the cremation pyre. Al-
Combined with the gilt tesserae of the dome deco-
though the imagery and significance of the pyre con-
ration and large windows allowing in the light, they
tinued into the fourth century, it became separated
would have produced a twinkling effect that would
from the actual body of the deceased once inhuma-
have reflected the celestial imagery of earlier pane-
tion became the norm and burial was no longer nec-
gyric prose, such as the description of the apotheosis
essarily in Rome. The grand domed mausolea were
of Constantine’s father Constantius:
built as settings for the interior aspects of the funer-
ary ceremony, which could be entirely separate from . . . when he was about to go to the gods he looked
the burning of an effigy of the emperor.68 In the case out on Oceanus, the father of the gods from which
of the Mausoleum of Helena, the domed structure the fiery stars of heaven gain new light so that when
thence to enjoy eternal light, he should already see
provided the resting place for her sarcophagus as well
from that place an almost unending daylight.75
as the setting for Eucharist celebrations as indicated
by the liturgical vessels listed as donations in the Liber The dome retained the celestial overtones that had
Pontificalis.69 The placement of a 200-lb silver altar been long associated with it, but they could be read
in front of the sarcophagus of Helena also emphasizes in a new Christian context. The combination of
the importance of the religious ceremonies in relation tomb cum church was then transferred to Constan-
to the memory of the empress.70 tine’s burial place in the Church of the Apostles at
The use of the tetrarchic form of imperial tomb Constantinople.76
for the Mausoleum of Helena fulfilled both symbolic After Constantine’s defeat of Licinius in 324 and
and ceremonial functions. The Christians adopted the subsequent creation of Constantinople in 326–
the symbolism of the “vault of heaven” as shown by 330, the creative energy put into the construction
180
P1: JPK
of large and impressive concrete vaulted structures, of pitched brick in the Byzantine world.77 Vault-
especially public ones, faded in Rome. The dome ing continued to occur in Rome, especially for small
retained its symbolic significance, but for the most domed mausolea, but it was not used on such a grand
part the largest examples moved away from Rome, scale as it had been previously nor did it employ the
and the techniques used to build vaults were adapted same construction techniques.78 The ultimate legacy
to new environments and available materials. Hence, of the vaulted construction of imperial Rome ap-
one finds large domes and other vault forms built of peared in the new Christian capital of Constantinople
tubi fittili in northern Italy and elsewhere from the fifth with Justinian’s sixth-century vaulted masterpiece of
century and new vaulting methods using variations the Hagia Sophia.
181
P1: JPK
182
P1: JPJ
0521842026apx01.xml CB885B/Lancaster 0 521 84202 6 April 18, 2005 10:37
APPENDIX 1
CATALOGUE OF MAJOR MONUMENTS
the following catalogue includes the major Materials. The arches, approximately 6 m thick, are
monuments from Rome and environs discussed in this built of peperino (lapis Gabinus) blocks with an outer
study. The purpose of the catalogue is threefold. First, facing of travertine voussoirs.
each entry introduces the location, date, purpose, and, Centering. The arch connecting to the left bank has
if relevant, the later history of the monument. Second, two cuttings, spaced approximately 4.5 m apart, along
the details of the various constructional issues relevant the impost at either side of the intrados. Each cutting
to this study are summarized together. If the issue is contains two holes, one above the other. The top hole
discussed at length in the main text, a cross reference is is cut to receive a horizontal beam whereas the lower
provided rather than repeat the information. In cases hole is cut to receive a diagonal brace, presumably for
where the observations are my own unpublished on- the centering structure (Fig. 20, p. 32).
site observations, I include more details than if the in- bibliography: Blake 1947: 146–7; O’Connor 1993:
formation is published elsewhere. Third, the relevant 66; Galliazzo 1995: 20–3; LTUR 4: 109–10.
bibliography that deals with the construction tech-
niques used in the monument is included at the end of 2. pons cestius (first century b.c.,
each entry. I also cite the references from the Lexicon rebuilt a.d. 370)
Topographicum urbis Romae (LTUR), which the reader comments: The Pons Cestius, which originally con-
can consult for a more comprehensive bibliography. sisted of three arches (13.7-m span), connects the right
Each monument is located on Map 1 (p. 4) according bank of the Tiber to Tiber Island. It dates from the
to its catalogue number. first century b.c., probably around the same time
as the Pons Fabricius so as to complete the passage
1. pons fabricius (62 b.c.) across the island. The bridge was then restored un-
comments: The Pons Fabricius consists of two der Gratian in a.d. 370 as attested by inscriptions
arches (24.5-m span) and connects the left bank of mounted on the bridge; therefore, the material and
the Tiber to Tiber Island. Identical inscriptions on techniques of the upper part of the bridge probably
both sides of one of the arches indicate that it was represent the fourth-century reconstruction. In 1888–
built by L. Fabricus in 62 b.c. 1892, it was taken apart when the Tiber Embankment
183
P1: JPJ
APPENDIX 1. CATALOGUE OF MAJOR MONUMENTS
was built, and the central arch was reconstructed us- della via Tiberina, which probably date to the Flavian
ing about a third of the original (fourth-century[?]) period.
blocks of travertine. Centering. Brick linings remain on the vault of one
Materials. The bridge was built of travertine blocks of the third-century rooms in the courtyard.
facing a core of tuff and peperino. The travertine Metal Fittings. Three travertine blocks from the pil-
blocks were connected by pyramidal-shaped iron lars of the arcade remain that have cuttings for metal
dowels and pi clamps fixed with molten lead. The bars in the top surface (Fig. 98, p. 116) (Bauer and
foundation was built on oak piles. Pronti 1978: 113, nos. 36–8, fig. 5). Two other traver-
Centering. In the fourth row of voussoirs of the tine blocks that have been reconstructed as belonging
central arch are projecting corbels, three of which on to the dividing walls of the rooms have cuttings for
the left side and one on the right, are ancient (the bars in the top surface (Bauer and Pronti 1978: 130–1,
other ones are modern). All the original corbels have nos. 83–4, fig. 15). Bauer has reconstructed the build-
on the underside the remains of two grooves cut so ing with exposed tie bars projecting from the façade
that they are deeper towards the outer end of the block wall of the courtyard to the dividing walls of the barrel
(Fig. 21, p. 33). Piranesi reconstructed the cutting as vaulted rooms (Fig. 97, p. 115).
lodgings for centering supports (Fig. 22). bibliography: Astolfi et al. 1978: 31–100; Bauer and
bibliography: Hülsen 1889: 282–5; Lanciani 1897: Pronti 1978: 107–31; Bauer 1978: 132–46; LTUR 3:
18–19; Galliazzo 1995: 10–13; LTUR 4: 108–9. 37–8.
3. horrea agrippiana (33–12 b.c.) 4. theater of marcellus (17 b.c.)

comments: The Horrea Agrippiana was built along comments: The Theater of Marcellus was built in
the Vicus Tuscus by Agrippa sometime between his the southern Campus Martius by Augustus to finish a
first aedileship in 33 b.c. and his death in 12 b.c. project conceived by Caesar. It was probably first used
It consisted of an enclosed rectangular structure in the ludi saeculares of 17 b.c. and was then dedicated
built around an open courtyard. A two-story ar- by Augustus to Marcellus in 13 or 11 b.c. The standing
caded portico surrounded the courtyard, and bar- remains of the building consist of an outer ambula-
rel vaulted rooms on two levels opened off the por- tory on at least two levels, an interior ambulatory at
tico. The northeast enclosure wall, originally built ground level called the “Passaggio dei Cavalieri,” and
in blocks of tufo lionato, was rebuilt in opus testaceum a series of radial barrel vaulted passages forming the
when Domitian added a grand entry vestibule (Ap- substructure of the cavea (Figs. 65–66, pp. 89–90).
pendix 1.10) leading up to the Palatine. In the mid- The outer ambulatory at ground level is covered by
second century, a sacellum to the genius of the horrea an annular barrel vault, whereas at level 2 it is covered
was added (Astolfi 1978: 60–1). In the third century, by a series of side-by-side barrel vaults supported on
a series of rooms were built in the open courtyard. travertine lintels spanning the width of the corridor.
Materials. The arcaded portico was built of blocks Materials. The outer wall is built of travertine
of travertine, and the dividing walls of the vaulted whereas the inner wall of the outer ambulatory is built
rooms were built of blocks of tufo lionato. Traces of of tufo lionato with springer blocks and keystones of
the remains of vaults built against back wall of the travertine (Fig. 63, p. 88). The radial walls are built of
Domitianic Vestibule contain caementa of tufo giallo opus quadratum combined with opus reticulatum. The
184
P1: JPJ
outer wall of the “Passaggio dei Cavalieri” is built of Metal Fittings. On the blocks with cuttings for tie
opus reticulatum and the inner one, which contains a bars, see Chapter 6 (Figs. 100–101, pp. 117–18).
series of relieving arches, of opus testaceum. The wall of
bibliography: Blake 1947: 187, 336–7, 344; Bauer
opus testaceum in which the arches occur is constructed
1988: 200–12; Amici 1997: 85–95; LTUR 1: 183–7.
of fairly thick yellow tiles (?) with wide mortar joints
(module of five bricks and joints = c. 25–6 cm). The
6. esquiline wing of the domus aurea
tiles making up the relieving arches are 40 cm thick.
(a.d. 64–68)
The caementa of all vaults consist of tufo lionato.
comments: The Domus Aurea was part of Nero’s
bibliography: Calza Bini 1953: 1–44; Fidenzoni extensive palace rebuilt after the fire of a.d. 64. Part of
1970; LTUR 5: 31–5. the palace on the Esquiline was preserved when it was
5. portico of basilica aemilia (paulli) built into the foundations of Trajan’s Baths (a.d. 104–
(c. 14 b.c.–a.d. 4) 109). Most of the preserved rooms are covered with
barrel vaults except for the octagonal room, which
comments: The portico (7-m span) ran along a se-
consists of an eight-sided dome surrounded by rooms
ries of barrel vaulted shops (4.5-m span) fronting the
covered by either barrel or cross vaults.
Basilica Aemilia. It is often called the Porticus of Gaius
Materials. The walls throughout the Neronian parts
and Lucius after the monument mentioned by Sueto-
of the building are built of opus testaceum. The caementa
nius (Aug. 29.4) because of the large pieces of inscrip-
of the vaults consist of tufo lionato and broken brick
tions found nearby mentioning Augustus’s adopted
throughout.
grandsons Gaius and Lucius (CIL VI, 36908; CIL VI,
Centering. The vault of the octagonal room con-
36880; CIL VI, 36893). The structure was built to-
tains the impressions of the formwork (c. 28 cm wide)
gether with the reconstruction of the Basilica Aemilia
along its intrados (Fig. 34, p. 42). On the evidence for
after a fire in 14 b.c. (Dio Cass. 54.24.2), and the
the centering structure of the octagonal room, see
inscriptions provide a terminus post quem of the first
pages 42–3. All of the corners have large depressions
decade of the first century a.d. (Rose 2005: 40–1).
gouged out of the corners, which have sometimes
The façade wall of the portico consisted of an arcade
been interpreted as socket holes for centering beams,
with engaged columns supporting a Doric frieze, the
but De Angelis d’Ossat 1938 (pub. 1940): 241 says
metopes of which contained alternating bucrania and
that none of the holes have bearing planes and that
patera (Fig. 99, p. 117). Bauer has reconstructed the
they appear to have been gouged away later. He also
portico covered by a series of concrete cross vaults,
has noted that the profile of the intrados of the vault
though there are no surviving remains to verify their
corresponds to a circular arc only along the diago-
existence.
nals. Brick linings have been reported at the Esquilina
Materials. The walls of the barrel vaulted shops
Wing, but in fact, there are none there (p. 233 n. 33).
opening off the portico are built of tufo lionato blocks
Buttressing. See page 143 (Fig. 130).
held together with dovetail clamps (bronze or wood?).
Blake 1947: 336–7 noted the remains of “dusky red bibliography: Zander 1958: 47–64; Blake 1959: 48–
mortar” with “a soft grayish tufa” in vaults of the 52; Fabbrini 1982: 63–84; Fabbrini 1985–1986: 129–
shops. The façade wall of the portico was built of 79; Ball 1994: 183–254; Fabbrini 1995: 56–63; Lan-
white marble (Luna or Pentelic?). caster 1995a: 13–37; Ball 2003; LTUR 2: 56–63.
185
P1: JPJ
7. colosseum (a.d. 70–80; arena substructures revealed that oak was used for the
reconstructed after a.d. 217) formwork (Ghini 1988: 101–5). Dendrochronologi-
cal and carbon-14 analyses show that the wood was
comments: The Colosseum was built on the site of taken from a one-hundred-year-old tree and put in
Nero’s Lake in the Valle Labicana. The project was place while still green, which has caused speculation
evidently funded from the spoils of the Jewish Wars that the wood was used before it was properly sea-
(Alföldy 1995: 195–226). On Vespasian’s accession, soned due to the high demand such a large project
there was no permanent amphitheater in Rome be- placed on the availability of wood (Luciani 1993: 98–
cause the amphitheater of Statilius Taurius had been 9). The formwork impressions visible in the cross
destroyed in the fire of a.d. 64. The Colosseum vaults of the level 2 inner ambulatory show very few
consists of two outer ambulatories at three levels. butt joints between the boards, but those that do ap-
The lowest two levels are covered with barrel vaults, pear (in the third-century reconstruction) are located
whereas the third was covered with cross vaults, which directly above the piers where the frames would have
were rebuilt in the nineteenth century (Fig. 69, p. 92). rested (Fig. 29, p. 38). Brick linings (second century
Radially arranged barrel vaulted stairwells supported a.d.[?]) were used to repair the vaults of the niches
the cavea. The building was damaged after a fire in along the containment wall of the hypogea. Many of
a.d. 217 described by Dio Cassius (79.25.2–3), and the remaining bessales bear anepigraphic stamps.
sections of the superstructure were rebuilt. The arena Ribs. See pages 88–91 (Figs. 67–71).
has undergone more phases of reconstruction than Metal Fittings. All the travertine blocks of the piers
the superstructure both before and after the fire. and arches in the Flavian building were connected
Materials. The original Flavian vaults all consist of with pi clamps (piers) and dowels (piers and arches).
caementa of tufo giallo della via Tiberina (12–22 cm long Tuff blocks were clamped and doweled to adjacent
and 6–8 cm high), whereas the third-century recon- travertine blocks but not to adjacent tuff blocks. The
structions employ tufo lionato. The mortar of the Fla- travertine blocks of the rebuilt piers were dowelled
vian vaults at levels 2 and 3 also contains small pieces together, but the voussoirs of arches were not.
of white pumice (< 1.5 cm) added to the mixture bibliography: Gerkan 1925: 11–50; Cozzo 1928:
(Pl. V). I have not found pumice in the mortar of the 195–253; Ghini 1988: 101–5; Garello 1991: 173–95;
ground-level vaults. Luciani 1993; Lancaster 1995a: 38–63; Piraino 1996:
Centering. Formwork imprints in both the Flavian 143–55; Beste 1998: 106–18; Lancaster 1998b: 146–
and the third-century phases are approximately 25– 74; Piraino 1998: 119–25; Beste 1999: 249–76; Beste
30 cm wide. The Flavian imprints on large vaults 2000: 79–92; La Regina 2001; Rea 2002; Rea et al.
rarely ever show the grain, but the small vaults of the 2002: 341–75; LTUR 1: 30–5; LTUR 5: 223.
niches facing the arena at a level 1 have imprints of
somewhat smaller boards with knots. Imprints of the 8. domus tiberiana (a.d. 54–138)
third-century formwork in the outer ambulatory at comments: The visible remains of the Domus
level 1 show staggered butt joints, which suggest that Tiberiana on the Palatine date from the first half of
the ends of the boards did not precisely align with the first century a.d. to the early third century with
the support frames during the reconstruction. Exca- major activity under Nero, Domitian, Hadrian, and
vations in the large drains along the minor axis of the the Severans.
186
P1: JPJ
Materials. Van Deman 1924: 391 describes the Materials. The caementa throughout the Flavian
caementa of the pre-Domitianic rooms as consisting structures consist of tufo giallo della via Tiberina. In
of a friable, clear yellow tuff mixed with small pieces one of the pre-Domitianic structures under the basil-
of pumice. I have not confirmed this personally. She ica of the Domus Flavia, called the Aula Isiaca, Caret-
notes that the caementa of yellow tuff are unusually toni 1971: 326, fig. 31 noted that pieces of “pomice”
large (20–30 cm long and 5–10 cm high) and are made up the caementa in part of the vault. In the pho-
placed in rows. Some of the Neronian vaults sup- tograph he published, the caementa are visible at the
ported gardens above and were protected with raised crown and appear to be roughly fist-sized pieces of
floors. See pages 58–9, 173–4 (Fig. 44, p. 60). A similar dark scoria.
technique was used again in the Hadrianic additions Centering. De Angelis d’Ossat 1938 (pub. 1940):
(Appendix 2e.9). 226 pointed out that the shape of the octagonal pavil-
Centering. The impressions of boards (20–5 cm ion vaults of the room facing onto the lower peristyle
wide) are visible along the intrados of the Neronian of the Domus Augustana was formed by semicircles
vaults. Van Deman 1924: 387, 391 indicates that the along the middle of the flat panels, which would
formwork was covered with a thin layer of mortar suggest that the main centering frames for the vault
with unusually high lime content. were located along the flat sides instead of along the
Ribs. Two of the Domitianic rooms have the re- diagonals.
mains of bipedalis ribs in the vaults. The intrados of Ribs. The barrel vaults of the niches of the polyg-
one of the doorways in the Domitian section is built onal rooms facing onto the lower peristyle of the
with three rows of whole bipedales set upright in the Domus Augustana are all built of radially laid bipedales.
mortar at a distance of about 0.10–0.15 m, but there See pages 94–5 (Fig. 73). The Flavian arches of the
are no smaller bricks in between as in the more devel- branch of the Aqua Claudia supplying the Palatine
oped form of the lattice rib (Lancaster 1995a: 78–9, also are built of radially laid bipedales.
fig. 67C). Buttressing. The largest rooms on the Palatine are
the state halls of the Domus Flavia: the aula regia
bibliography: Van Deman 1924: 368–98; Krause
(30.4-m span), the triclinium (29-m span), the basil-
1986: 442–63; Krause 1994; Lancaster 1995a: 74–85;
ica (15-m span), and the lararium (13.5-m span), and
LTUR 2: 189–97.
the reconstruction of their roofing has been contro-
versial. The reconstructions for the group of three
9. domus augustana/flavia rooms along the northeast side (basilica, aula regia,
(a.d. 81–92) and lararium) include wooden roofs for all three
comments: The Domus Augustana and the Domus rooms (Finsen 1962: 23–34; Giuliani 1977: 91–106);
Flavia are names that refer to the palace built on the concrete barrel vaults for all three (MacDonald 1982:
south side of the Palatine, the Domus Flavia referring 56–69; Wataghin Cantino 1966); and concrete barrel
to the more residential building to the east and the vaults for the basilica and lararium with a wooden
Domus Flavia referring to the reception halls to the roof for the aula regia (Ward-Perkins 1981: 80, 104);
west. Both were built above earlier remains and often or no roof for the aula regia (Giovannoni 1938
incorporated walls from Nero’s earlier palace into the (pub. 1940): 85–94). The evidence most often cited
new design (Cassatella 1990: 91–104). in favor of concrete vaulting is the comment by the
187
P1: JPJ
eighteenth-century excavator, Bianchini 1738: 50, gave access to a group of rooms that were turned
who said that the mound under which the basilica, into the church of Santa Maria Antiqua in the sixth
aula regia, and lararium were buried was formed “da’ century a.d. On the east side of the peristyle a se-
cementi delle volte cadute.” Bianchini never speci- ries of ramps led up to the Palatine. The Vestibule
fied the exact location of the “cementi.” Finsen 1962: was built up against the Horrea Agrippiana, and a
24–5 reconstructed wooden roofs for all three rooms series of wedge-shaped rooms filled the space be-
based on the results of structural analysis (no details tween the large hall and the south wall of the Hor-
given), which indicated that a barrel vault over the rea, which was modified to accommodate the new
aula regia would have been impossible given the lack structure.
of buttressing at either side. MacDonald 1982: 56–89 Materials. The caementa of the vaults in the wedge-
restored the aula regia with a barrel vault and sug- shaped rooms along the northeast wall of the Horrea
gested that the end walls of the rooms could have Agrippiana consist of bricks and tufo lionato.
acted as ties to help resolve the horizontal thrusts. He Centering. Delbrueck 1921: 11 noted that the vaults
acknowledged that the theoretical lines of thrust pass of some rooms, which he specifies only as “un-
outside of the external face of the supporting walls zugänglichen Räumen,” were constructed using an
for all of these structures but suggests that the foun- earth centering and that traces of fine sand, bricks,
dations, which are wider than the walls, could have and lime were found along the intrados as well as im-
compensated since they are “structurally inseparable pressions of horizontal floor planks, which must have
from the walls.” Giuliani 1977: 91–106 proposed a formed the surface holding the earth centering.
trussed gabled roof above the aula regia and flat ter- Ribs. One of the vaults (6.5-m span) of the wedge-
races above the two side halls. Given the lack of but- shaped rooms served as buttressing for the back wall
tressing, the most likely solution for the aula regia of the Horrea Agrippiana and was reinforced with
seems to me to be a wooden truss, which, with a bipedalis ribs. See pages 91, 94 (Fig. 72).
span of 30.4 m, would have been one of the largest in Buttressing. Delbrueck 1921: 16–17 suggested that
the Roman world and no small feat of engineering. the heavy buttressing along the north and south walls
bibliography: Giovannoni 1938 (pub. 1940): 85– of the large hall were intended to support a massive
94; Carettoni 1949: 48–79; Finsen 1962; Wataghin barrel vault (32.5-m span) running in the east–west
Cantino 1966; Finsen 1969; Carettoni 1971: 300– direction, which was never built, as confirmed by
26; Romanelli 1973: 208–18; Giuliani 1977: 91–106; excavations by Hurst 1986: 476.
MacDonald 1982: 56–69; D’Elia and Buranelli 1985: bibliography: Delbrueck 1921: 8–33; Krause 1985:
176–8; Luciani 1985: 148–55; Cassatella 1990: 91– 73–136; Hurst 1986: 470–8; Lancaster 1995a: 74–85;
104; DeLaine et al. 1994: 67–97; Lancaster 1995a: LTUR 2: 197–9.
69–72; LTUR 2: 42–5.
11. structure under san clemente
10. domitianic vestibule (a.d. 81–92) (a.d. 70–96)
comments: The Domitianic Vestibule was part of comments: The large structure underneath the
the modifications to the Domus Tiberiana on the basilica of San Clemente, which has been identified
northwest corner of the Palatine. It consists of a large as either a horreum or a mint, consisted of a series of
hall (31 × 21 m) leading into a peristyle, which barrel vaulted rooms (4.30 m wide × 5.60 m long)
188
P1: JPJ
facing onto an open courtyard and dates from the vaults (14.2-m span), which were connected to the
Domitianic period. semidomes of the apse, show only signs of tufo giallo.
Materials. The outer wall of the structure consists Other fallen fragments on the site also have alternating
of blocks of tufo lionato whereas the interior walls sup- rows of Vesuvian scoria and tufo giallo: a barrel vault
porting the vaults are built of opus mixtum. (c. 8.10-m span), a cross vault (c. 8.75-m span), and
Centering. Each of the barrel vaulted rooms has a semidome (c. 12-m dia) (Pl. VIII and Appendix 3).
square holes (25 × 25 cm) built 26 cm below the The Trajanic builders also protected the earlier vaults
spring of the vault (Fig. 24, p. 35). The holes (1.35 m of the Domus Aurea by covering them with cocciopesto
apart) extend all the way through the wall, and the and building raised floors of bipedales raised on suspen-
inner surfaces are not faced with brick. The imprints surae (Appendix 2e.8). The places where this tech-
of the formwork boards show that they were about nique was used corresponds to the areas that would
10 RF long and 1 RF wide. See further, page 34. The have been part of the garden area between the outer
floors of the rooms were later raised, and a layer of enclosure wall and the main bath building.
cocciopesto covered the holes, which were presumably Ribs. Section E also contains anomalous “ribs” of
filled. vertical bipedales are placed in the remains of bar-
bibliography: Guidobaldi 1992: 55–68; LTUR 3: rel vaults at either end of room. See page 96 (Fig.
280–1. 77). The recent discovery of two relieving archs in a
tunnel on the west side of the outer enclosure have
12. baths of trajan (a.d. 104–109) painted graffiti noting dates, and Volpe 2002: 383–
comments: The Baths of Trajan were built on the 91 has shown that these probably relate to work days.
Oppian hill after a fire in a.d. 104 ( Jer., Ab Abr. 2120). The sequence shows that both the arch and the filling
The platform on which they were built incorporated in the lunette progressed alternately in stages.
parts of the Esquiline Wing of Nero’s Domus Aurea. Metal Fittings. In the remains of section K at the
Only fragments of the superstructure of the Baths sur- crown of the barrel vault that once surrounded the
vive, but the plan is known from Renaissance draw- east palaestra, there are two holes (c. 4 m apart), one
ings and from the Severan Marble Plan, so the dis- of which still contains the stone block for anchoring
parate parts that do survive can be put into context. the metal tie bars that ran across the crown of the
The letters on Pl. VI referring to various sections of vaulted portico. See pages 116–18 (Fig. 102).
the complex follow those assigned by de Fine Licht Buttressing. The triple cross vaulted frigidarium was
1974. surely buttressed in the same way as the frigidarium
Materials. See Pl. VI. During excavation at the base in the Baths of Caracalla and in the Baths of Diocle-
of exedra L, large amounts of volcanic scoria were tian, but there are no material remains to verify what
found on the ground, which probably belonged to form the buttresses took. The three standing apsidal
the upper part of the dome (de Fine Licht 1974: 21). halls of section D (29.5-m span), section H (30.6-m
Section E had numerous different types of caementa span), and section L (28.8-m span) are the largest
used in different parts. The apse preserved at one end spanned structures remaining at the Baths. Originally
contains Vesuvian scoria used together in alternating they had other vaulted structures at either side that
rows with tufo giallo della via Tiberina. Some pieces of would have provided some buttressing against the lat-
tufo rosso a scorie nere are also identifiable. The barrel eral thrusts generated by such great vaults. Substantial
189
P1: JPJ
cracking is visible in the upper parts of the walls of apse on all four sides that served as buttresses to counter
D, indicating that at some point spreading did occur the outward pressure of the vast amounts of water
(Fig. 4, p. 8). Unfortunately, the forms of the extrados contained within.
are not preserved well enough to give an indication bibliography: Cozza 1974–1975: 79–101; de Fine
of whether step-rings like those at the Pantheon were Licht 1990; Lancaster 1995a: 134–46; LTUR 5: 68–9.
used to create a surcharge on the haunches of these
semidomes. 14. forum of trajan (a.d. 106–113)
bibliography: de Fine Licht 1974; Fabbrini 1982: comments: The Forum of Trajan, which was built
5–24; Lancaster 1995a: 134–46; Volpe 2002: 377–94; from the spoils of the Dacian Wars, is the last and
LTUR 5: 67–9. largest of the imperial fora. The complex consisted
of a large open area bounded by an exedra and por-
13. sette sale (a.d. 110) tico on the east and west and by the Basilica Ulpia
comments: Sette Sale is the name given to the large on the north. Beyond the basilica to the north lay
cistern that served Trajan’s Bath. It consists of nine a colonnaded courtyard surrounding the Column of
barrel vaulted rooms (5.4-m span) on each of two Trajan with libraries opening off the east and west
levels (plan on Pl. VI). The top rooms are intercon- side.
nected by doorways and served as the water contain- Materials. The aisle vaults (6.2-m span) of the
ment basins, whereas the lower rooms acted as sub- Basilica Ulpia were built of tufo giallo della via Tiberina
structures that raised the basins in order to generate and Vesuvian scoria in alternating rows (Appendix 3).
the necessary water pressure. A small piece of vaulting with scoria and tufo giallo
Materials. The walls are faced with brick, and the was also found in the west library. In the recent ex-
caementa in the walls and vaults consist of tufo lionato. cavations at the south end of the Forum of Trajan,
Centering. The lower series of vaulted rooms were chunks of concrete containing the same type of sco-
built on centering formed of piled earth (Cozza 1974– ria combined with pieces of tufo lionato (?) have been
75: 90 n. 19, fig. 13). The vaults of the upper chambers found. The caementa in the latter chunks consisted
were built on brick linings of sesquipedales with their predominantly of scoria and were not laid in alter-
joints covered by a grid of bessales. The dividing walls nating layers like that of the Basilica Ulpia vaults. In
between these chambers have a series of rectangu- contrast, the caementa of the substructure vaults of the
lar holes (45 cm high × 30 cm wide) built 15 cm Column Portico, which would have undergone much
below the impost of the vaults and spaced 2.29 m more traffic of men and materials, consisted of the
(7.75 RF) apart (Cozza 1974–75: fig. 13; de Fine Licht heavier tufo lionato.
1990: 31–2). The holes, which must have originally Centering. Formwork imprints visible on fallen
extended through the wall, were filled with brick fac- vaults of Basilica Ulpia are 30–5 cm wide.
ing soon after construction and plastered over. They Ribs. The substructure vaults of the Column Por-
were probably used to hold the centering frames dur- tico contain a bipedalis rib (6.4 m wide), which I
ing construction as in the Domitianic structure below have argued elsewhere was used to reinforce the vaults
San Clemente (Appendix 1.11). against the weight of the blocks of the Column when
Buttressing. The exterior of Sette Sale was designed it was being built (Lancaster 1999: 423–4). Recent in-
so that it was surrounded by a series of apsidal niches vestigations, however, have uncovered two Hadrianic
190
P1: JPJ
brick stamps, CIL XV, 1013, 1209b, dated to a.d. 123 16. trajan’s markets (a.d. 106–113)
and 125, respectively, that show that it was added after
the Column was erected. I thank Roberto Menegh- comments: Trajan’s Markets make up the complex
ini and Elisabetta Bianchi for bringing these stamps built into the Quirinal on the east side of the Forum
to my attention and for helping me take a cast of one of Trajan and, like the renovations of the Forum of
in situ. Caesar, were part of the Trajanic transformation of
Metal Fittings. See pages 118–25 (Figs. 106–109). this area. They are built on six different levels and are
bibliography: Amici 1982; Meneghini 1989: 541– separated into two buildings by the Via Biberatica.
57; Milella et al. 1989: 27–291; Packer 1997; Lancaster The lower building consists of the large Hemicycle
1999: 419–39; Meneghini 2001: 245–61; Bianchi facing onto the east apse of the Forum and the wing
2001: 82–120; Bianchi and Meneghini 2002: 395– projecting off it to the north (i.e., the North Wing).
417; Packer 2003: 109–36; LTUR 2: 348–56. Structurally, the most complex section is the Aula
located at the north end of the upper building at level
15. trajanic renovations at the forum 4. It consists of a two-story hall (8.5-m span) covered
of caesar (a.d. 106–113) by six cross vaults flanked by barrel vaulted rooms on
comments: The Forum of Caesar was renovated in either side. Most of the other spaces in the Markets
conjunction with the construction of the Forum of are covered by barrel vaults except for the four apsidal
Trajan. The barrel vaults of some of the rooms on the halls covered by semidomes: two at the north end of
west side of the Forum of Caesar were demolished, the Hemicycle (17.1-m dia, 13.5-m dia), one at the
and new vaults were added to support a semi-elliptical south end (14.2-m dia), and one at level 4 of the upper
latrine. To the north of the latrine, the Basilica Argen- building (10.7-m dia).
taria was built at forum level, and new vaulted taber- Materials. All the vaults employ a mix of broken
nae facing onto the Clivus Argentaria were added bricks and tufo lionato except for the cross vaults of
above it. the Aula, which was built exclusively of tufo giallo della
Materials. Tufo giallo della via Tiberina was used for via Tiberina. The large semidome at the north end of
the caementa of the vaults of the Basilica Argentaria and the Hemicycle was covered by a 15-cm thick layer of
the tabernae along the Clivus Argentaria. Some pieces cocciopesto, whereas the somewhat smaller semidome
of tufo rosso a scorie nere are also visible. In contrast, the behind it was protected by roof tiles set into cocciopesto
rebuilt tabernae vaults supporting the latrine employ and then covered with another layer of cocciopesto.
primarily broken bricks. Vaults with a flat extrados were usually covered by
Centering. The vaults supporting the latrine also a layer of cocciopesto on which was set a layer of opus
have the remains of brick linings made of sesquipedales spicatum, which, in turn, was often covered by an-
with the joints covered by a grid of bessales, each of other layer of cocciopesto with mosaic of leucititic lava
which has a hole chipped out of the center. on top (Appendix 2e).
Ribs. The latrine was built with a complex system Centering. Two groups of rooms, one on level 2
of ribs and relieving arches. See further, pages 98–9 of the North Wing and the other on level 3 of the
(Figs. 78–79). Aula, were built using brick linings of bipedales with
bibliography: Amici 1991b: 77–136; Lancaster the joints covered with bessales with hole chipped out
2000: 779–84; LTUR 2: 299–306. of center, as at the Trajanic latrine (Appendix 1.15).
191
P1: JPJ
All the other rooms in the Markets were built with that is, tufo giallo della via Tiberina. The vaults of the
wooden formwork, though the impressions of the building on the south side of the rotunda are some-
boards do not survive. Elsewhere, I have suggested what different and consist of caementa of alternating
that these two groups of rooms represent the work of rows of yellow tuff and brick. The lower ledge (above
a single team of builders (Lancaster 1998a: 283–308). the third exterior cornice) was protected by a series of
Ribs. Bipedalis ribs occur underneath walls in the marble roof tiles (Cozza 1983: 109–18), whereas the
Aula, in vaults in the North Wing, and as radially dome itself was covered by bronze roof tiles, which
laid bipedalis vaults at ground level of the Hemicycle were stolen in antiquity and have been replaced by
(Lancaster 2000: 755–85). lead sheeting.
Buttressing. For the step-ring on the large Centering. The vaults of the niches and void spaces
semidome to the north of the complex and the within the rotunda wall were built on linings of
buttressing arches in the Aula, see Chapter 7 and bipedales. The joints were probably covered by bessales,
Lancaster 2000: 755–85. but the linings are still intact and cannot be checked.
bibliography: Ricci 1929: 541–55; Lugli 1929– The dome itself would have been built using wooden
1930: 527–51; Blake 1973: 19–28; MacDonald 1982: formwork, though the plaster now covers any evi-
75–93; Bianchini 1991: 102–21; Lancaster 1995b: 25– dence for its arrangement. See further, pages 44–6.
44; Ungaro 1995: 126–35; Lancaster 1998a: 283–308; Ribs. The rotunda wall incorporates two systems
Lancaster 2000: 755–85; LTUR 3: 241–5; LTUR 5: of ribs. See pages 96–8 (Fig. 80). Another rib, un-
276. connected to the rotunda series, occurs in the south
building. It is a radially laid bipedalis vault (5.20-m
17. pantheon (a.d. 118–128)
span, two bipedales thick), which originally spanned
comments: The Pantheon was originally built by
the gap between the projection on the rotunda and
Agrippa in 27 b.c., but it burned down in the fire that
a corresponding projection on the wall of the struc-
destroyed much of the Campus Martius in a.d. 80 and
ture identified by G. Gatti as the Basilica Neptuni,
was rebuilt by Domitian. It was damaged by lighten-
which contains bricks contemporary to those found
ing under Trajan and then rebuilt in its present form
in the Pantheon (de Fine Licht 1968: 160–6; Gatti
under Hadrian from a.d. 118–128. The main struc-
1938 (pub. 1940): 61–74). The vault supported four
ture consists of a large dome (43.3-m dia) supported
parallel walls running perpendicular to the rotunda
by a 6-m thick cylindrical wall into which are built
wall and was evidently intended to stabilize the ro-
niches such that the weight of the dome is concen-
tunda and to support the four walls on top of it.
trated onto the eight piers between them. The porch
Buttressing. See pages 141, 158–61.
and its intermediate block on the north side of the
building are bonded to the rotunda wall in the lower bibliography: Beltrami 1898: 88–90; Colini and
part but not in the upper parts. The building on the Gismondi 1926: 67–92; De Angelis d’Ossat 1930:
south side is not bonded to the rotunda wall at all, 211–15; Rosi 1931: 227–9; Terenzio 1932: 52–7;
but brick stamps show that is roughly contemporary. Lugli 1933: 273–4; de Fine Licht 1968; MacDon-
Materials. For the distribution of materials, see ald 1982: 94–111; Cozza 1983: 109–18; Mark and
Fig. 46. Gioacchino De Angelis d’Ossat 1930: 211–15 Hutchinson 1986: 24–34; Wilson Jones et al. 1987:
determined that the scoria used in the dome was from 133–53; Martines 1991: 3–18; Haselberger 1994: 279–
Vesuvius and the yellow tuff from the Sabatini system, 308; Wilson Jones 2000: 177–212; LTUR 4: 54–61.
192
P1: JPJ
18. nymphaeum in horti sallustiani been excavated, but numerous walls remain stand-
(c. a.d. 126) ing, and pieces of fallen wall and vaults remain scat-
comments: The nymphaeum of the Horti Sallustiani tered throughout the area. Brick stamps date it to the
is located at the high point of the valley separating the Hadrianic period, and Ashby and Lugli 1928: 190–1
Quirinal from the Pincian. The property was origi- speculated that it was owned by Q. Servilius Pudens
nally owned by Sallust but had come into imperial based on the stamps on lead pipes found in the area.
hands under Tiberius. The nymphaeum, however, Materials. The vaults have caementa of tufo lionato
dates to the Hadrianic period as shown by the brick and the walls are faced in brick.
stamps found there. Amphoras. The villa has the earliest known exam-
Centering. The central space is covered by a seg- ples of amphoras (Dressel 20) built into its vaults. The
mental dome (11.2-m dia) in which the eight seg- vaults of all the standing remains contain amphoras,
ments alternate between flat and concave. and numerous fragments of vaulting around the site
Ribs. Lehmann-Hartleben and Lindros 1935: 210 contain them. In one wall, two amphoras were below
report that the lower haunches of the flat segments the level of the impost (Fig. 51, p. 73), so apparently
of the dome are built of brick, but that the upper the amphoras were not confined to the vaults.
parts remain covered by plaster. They also note the bibliography: Ashby 1907: 74–8; Ashby and Lugli
presence of ribs in the barrel vault opening along the 1928: 183–92.
main axis of the structure.
Buttressing. Excavations on the extrados revealed a
series of three concentric channels surrounding the 20. villa di sette bassi
dome (Tedone 1990: 169–70). The two inner ones are (a.d. 140–150 = phase 3)
60 cm wide and are set over the dome while the outer
comments: The Villa di Sette Bassi was built in three
one is 90 cm wide and is set over the supporting walls.
phases during the mid-second century a.d. just past
The channels are not continuous all the way around
the fifth mile of the Via Latina. It was one of the
the dome but rather are divided into curving com-
largest of the villas on the outskirts of Rome along
partments 1.3–2.5 m long. Each compartment was
with the nearby and contemporary Villa dei Quintilli
built like a drain with a covering of two bipedales set
on the Via Appia. It consists of two wings of struc-
“a cappuccina.” The channels created hollow spaces
tures facing onto an unexcavated hippodrome. The
in the thick part of the vault that formed a flat terrace
most imposing structure remaining is a bath building
above and, like amphoras, would have reduced the
from phase 3, which has been dated by brick stamps
amount of material necessary for the thickest parts of
to a.d. 140–150. A barrel vaulted cryptoporticus ran
the vault.
along its south side and today provides access to the
bibliography: Lehmann-Hartleben and Lindros remaining substructures of the bath. A few impos-
1935: 196–227; Tedone 1990: 169–70; LTUR 3: 80–1. ing remains of the superstructure are also standing.
The villa was presumably built by a wealthy senato-
19. villa alla vignaccia rial family, though the owner remains unknown. It
(c. a.d. 123–130) takes its name from a toponym already in existence in
comments: The Villa alla Vignaccia is located at the tenth century derived from “Septimius Bassianus”
the fourth mile of the Via Latina. The villa has not (Ashby 1907: 111).
193
P1: JPJ
Materials. The caementa of the phase 3 vaults consist 21. hadrianeum – temple of divine
of peperino in the haunches and tufo lionato at the hadrian (a.d. 145)
crown. comments: The Hadrianeum, dedicated to Hadrian
Centering. Formwork imprints (28-cm wide) re- by Antoninus Pius in 145 a.d., was located within
main on cryptoporticus barrel vault. Brick linings of an enclosed precinct in the central Campus Martius
sesquipedales and bessales were used on all other barrel facing east towards the Via Lata. It took the com-
vaults of the substructures and on the few remaining mon form of a peripteral temple, but both aisles and
vaults of the superstructure. cella were made of concrete vaulting instead of the
Ribs. The barrel vault of the cryptoporticus (4.7-m more typical wooden construction. The columns,
span) was divided by projecting ribs that corre- cella wall, and part of the cella vault of the north flank
sponded to buttressing piers along the outer wall are preserved and form one side of the Piazza di Pietra
about every 4.5 m. Parts of six of these ribs are pre- today. Some vaulted substructures are also preserved.
served showing that they projected 6 cm at the im- Materials. The barrel vault of the cella (c. 18-m
post increasing to 30 cm at the crown. The projecting span) consists of yellow tuff in the lower parts and
parts of the ribs are faced with bricks and were cov- alternating layers of brown (Vesuvian?) scoria and
ered with plaster, parts of which still remain. Unlike yellow tuff in the upper part. The cella walls are built
typical ladder ribs, however, the outer bricks do not of blocks of peperino (lapis Albanus?) once clamped
project into the core of the vault (Fig. 83, p. 102), and together.
the intrados of the ribs are crossed only occasionally Buttressing. The large barrel vault of the cella was
by a larger brick (i.e., the rung of the ladder). A cross buttressed by the colonnaded aisles on either side,
vault (7.60-m span) in the substructures of the bath which were covered with small barrel vaults.
building has ladder ribs (c. 65 cm wide) along groins bibliography: Cozza 1982; LTUR 3: 7–8.
with bipedales extending all the way across every 6–10
bricks (Fig. 91, p. 107). The intrados of the vault is not 22. severan baths on the palatine
preserved so the ribs project, but the uneven face of (a.d. 193–211)
the external bricks suggests that they were once flush comments: The baths built by Septimius Severus on
with the vault (or with brick linings ?). The some- the southeast corner of the Palatine were raised on
what smaller adjacent cross vaulted room (Fig. 84, enormous concrete vaulted substructures overlooking
p. 103) does not employ ribs along the groins, but the the Circus Maximus. The remains were enlarged by
barrel vaults along its sides (4.50-m span) have ladder Maxentius (see Baths of Maxentius, Appendix 1.29),
ribs with rungs formed of two bipedales. These were but the core of the platform belongs to the earlier
probably used to aid in the construction process by Severan work. At the base of the hill, a series of the
allowing the barrel vaults to be built before the cross barrel vaulted rooms running along the present Via
vault was added. dei Cerchi probably also belongs to the Severan mod-
Buttressing. The cryptoporticus has along its south- ifications associated with the nearby Septizodium.
ern side a series of buttresses aligned with the vaulting Materials. The caementa of the vaults consist of tufo
ribs. lionato.
bibliography: Ashby 1907: 97–112; Rivoira 1925: Centering. The vaults of the substructures were
140–8; Lupu 1937: 117–88; Bianchi 2000: 124–5. built on brick linings of bipedales covered with a solid
194
P1: JPJ
layer of bessales, some of which bear anepigraphic ceilings hanging from T-shaped metal bars embedded
stamps (Fig. 86, p. 104). in the concrete (DeLaine 1987: 150–3).
Ribs. The barrel vaults of the substructures have Buttresses. The remains of freestanding buttressing
either one or two three-arch lattice ribs distributed arches can be seen above the springing of the frigidar-
along the length of the vault (Figs. 85–86). The barium vaults (Fig. 120, p. 137). The corners of most cross
rel vaulted rooms along Via dei Cerchi have three- vaulted rooms are reinforced with projecting piers.
arch lattice ribs at regular intervals along the intrados. bibliography: Brödner 1951; DeLaine 1985: 195–
These are set unusually close together (the space be- 206; DeLaine 1987: 147–56; Conforto 1991: 43–8;
tween the ribs (c. 1.2 m) is about equal to the width Lombardi and Corazza 1995; DeLaine 1997; Bianchi
of the ribs themselves). 2000: 125–9; LTUR 5: 42–8.
bibliography: Massaccesi 1939: 117–33; Carettoni
1972: 96–104. 24. baths of agrippa (third/fourth
century a.d.)
23. baths of caracalla (a.d. 212–216) comments: The Baths of Agrippa were originally
comments: The Baths of Caracalla were built at the part of Agrippa’s building program in the Campus
point where the Via Appia entered the city of Rome. Martius during the late first century b.c., but the
Brick stamps for the central building show that it building burned in the fire of a.d. 80 and was re-
was completed under Caracalla, whereas the outer built immediately afterwards. Other restorations are
enclosure wall was completed after his death (SHA, attested under Hadrian (SHA, Had. 19, CIL VI, 9797)
Heliogab. 17.8–9). and in a.d. 344–345 under Constantius and Constans
Materials. The caementa used in the vaults and walls II (CIL VI, 1165). The surviving remains consist of
mainly consist of tufo lionato, but in some places Vesu- part of the large domed structure (24-m dia) now
vian scoria was used at the crown of vaults: frigidar- embedded in the housing along Via dell’Arco della
ium (?), rooms 3W, 2E, and 14E (Pl. XIII). The large Ciambella. The date of this structure is somewhat un-
exedrae (c. 26.8-m span) opening onto the palaestrae clear. The style of construction suggests that it dates
employ a distinct gradation of brick in the lower third, from the third or fourth century. The structure is of-
tufo lionato in the middle third, and scoria in the upper ten dated to the reign of Alexander Severus (a.d. 222–
third (Pl. IX and Appendix 3). The vaulted porticoes 235), who restored the nearby Baths of Nero (Rasch
around three sides of each palaestra employed scoria 1991: 350; LTUR 3: 42), though there is no other
alone in the upper three quarters of the vault and documentation of a reconstruction at this time. The
brick in the lower quarter. only brick stamp evidence from the zone dates to the
Centering. Most of the surviving vaults were built Maxentian period (a.d. 306–312) (Steinby 1986: 123,
on brick linings. The solid layer of bessales, many of 142).
which bear anepigraphic stamps, is preserved in many Materials. The caementa in the dome consist of tufo
places, and probably covered a layer of bipedales that lionato.
is now missing. Ribs. The remains of the dome are now built into
Ribs. See pages 98, 100, 102–3, 107, 111. the surrounding structures, but almost half of the
Metal Fittings. For tie bars, see page 118. The vaults lower part is still visible displaying four-arch lattice
of some of the rooms appear to have had suspended ribs to either side of openings along the diagonal axes
195
P1: JPJ
(Figs. 93–94, pp. 108–9). Rasch 1991: 352 notes that the corbels were clearly not meant to be decorative
the placement of the existing ribs is unbalanced and and must have been used to support centering frames,
proposes three-arch lattice ribs also along the main as was typical in the construction of aqueducts (the
axes above the windows, where the vault no longer remains of the Severan reconstruction of the Aqua
remains. This proposed configuration results in a Claudia in the Villa Wolkonsky has a similar detail).
twelve-part division of the dome with ribs at inter- Ribs. The corbels are aligned with a series of
vals of 4.0–4.5 m, which would be appropriate for the pseudo-brick ladder ribs. Only the lower fourteen to
centering frames supporting radial formwork. At least eighteen brick voussoirs of the ribs survive, but they
one course of bipedales runs continuously between the show no signs of cross pieces. The corbels and the
ribs, but it is not aligned with the cross pieces in them. ribs appear to be related since both occur together at
Presumably, the bricks of this course are laid flat unequal spacing in order to avoid being located over
whereas the crosspieces of the ribs are laid voussoir- the openings in the wall below. See page 101.
like, so the two need not be related. Whether this co- bibliography: Tedeschi Grisanti 1977; LTUR 3:
urse relates to the form of the extrados is not known. 351–2.
bibliography: Hülsen 1910; Rasch 1991: 350–3;
Bianchi 2000: 133–4; LTUR 5: 40–2. 26. basilica julia (post-a.d. 283)
comments: The Basilica Julia was originally built
25. nymphaeum alexandri “trophies of under Caesar and then rebuilt after a fire under Au-
marius” (a.d. 222–235) gustus and dedicated in a.d. 14. The fire in a.d. 283
comments: The Nymphaeum Alexandri was built under Carinus again destroyed the building, and it
by Alexander Severus at the intersection of the Via was rebuilt under Diocletian. It consisted of two cross
Labicana and the Via Praenestina on the Esquiline, vaulted aisles (5.1-m span) supported on an arcaded
now at the Piazza Vittorio Emmanuele. It was dubbed structure. The nave was covered by a trussed roof
the “Trophies of Marius” because the statues of tro- structure. At the south corner of the building, there
phies presently mounted on the railings flanking the are remains of partially buried fallen vaulting. These
main stair up to the Campidoglio were originally must belong to the same vaults that have been re-
found in the niches of the fountain. It was proba- ported in past centuries. Fredenheim found pieces of
bly supplied by either the Claudia or the Anio Novus vaulting with coffers and stucco in 1780–1789 (LTUR
and was built in three levels such that the water was 1: 178). Canina 1860: 187 reported that remains of the
channeled through a variety of pipes and basins before fallen vaulting belonging to the Diocletianic restora-
collecting in a large basin at ground level. tion were found on the ground, and later Lanciani
Materials. The caementa throughout consist of tufo 1897: 277 noted the existence of pieces of fallen vaults
lionato. with stucco.
Centering. The remains have travertine corbels run- Materials. These pieces of vaulting consist of cae-
ning below what was once a barrel vault (2.8-m span) menta of Vesuvian scoria (Appendix 3). They are for
that supported basins at the central level. The corbels the most part still buried, so it is difficult to tell if they
are spaced either 2.0 m or 2.8 m apart in accordance belong to the crown or haunches of the vault.
with the openings below (Fig. 87, p. 104). As the bibliography: Canina 1860: 179–94; Rivoira 1925:
vault covered a utilitarian space inside the fountain, 202–4; LTUR 1: 177–9.
196
P1: JPJ
27. baths of diocletian (a.d. 298–306) room east of the natatio retains the imprints of radially
arranged formwork.
comments: The Baths of Diocletian on the Viminal Ribs. For a plan showing locations of ribs, see
were initiated by Diocletian’s co-Augustus, Maximian Pl. XIII. Every surviving cross vault displays lattice
Herculius, probably in a.d. 298. An inscription (CIL ribbing along the groins. The few barrel vaults used
VI, 1130 = ILS 646) notes that they were dedicated in the building also have the remains of lattice rib-
between May 1, 305, and August 306. Like the Baths bing, most notably those covering the niches to ei-
of Caracalla and the Baths of Trajan, they consisted ther side of the frigidarium. The vaults of two domed
of a central bath building surrounded by an outer rooms remain standing, the Church of San Bernardo
enclosure wall. (22.40-m dia) and the “Planetarium” (21.65-m dia in-
Materials. Heres 1983: 234–5 describes the mortar scribed). The ribs at San Bernardo are no longer visi-
as “tenacious and rather well-sifted” containing both ble, and only conflicting descriptions of them remain.
red and black pozzolana. The caementa used in the Based on earlier descriptions (Paulin 1890: 13; De
lower parts of vaults typically consist of tufo lionato. A Angelis d’Ossat 1938 (pub. 1940): 248 n. 58), Rasch
variety of types of pumice was used in the crowns of 1991: 352 proposes a reconstruction of the ribs similar
some of the vaults (Pl. XIII). These are rarely visible to that of the Baths of Agrippa with wide ribs be-
today and the attributions are based on observations tween the niches and narrower ones over the niches.
of De Angelis d’Ossat 1938 (pub. 1940): 245, Rivoira The “Planetarium” intrados is clearly visible, show-
1925: 206, De Angelis d’Ossat 1946: 21–2 (white?), ing eight three-arch ribs along the groins and eight
and Caraffa: fig. 5. The only examples that I have been more along the flat panels over the niches (Fig. 94,
able to examine personally (with the aid of binocu- p. 109). The latter ones occur directly over the crowns
lars) occur at the crown of the barrel vaults in the east of the niches in the walls. On the northeast corner of
palaestra, which contains large pieces of dark black- the enclosure walls of the complex, the remains of a
ish gray pumice (Pl. X). Candilio 1985: 528 reports large apse (42-m span) have a series of evenly spaced
finding similar pumice in the excavation of the west three–arch lattice ribs along the meridians.
palaestra. Metal Fittings. For tie bars, see page 118 (Fig. 104).
Centering. The formwork imprints are visible on a Amphoras. Ward-Perkins 1981: 436 reports am-
number of the domes and semidomes. The intrados of phoras used in the vaults, but I find no evidence for
the large semidome (c. 42-m span) in the northeast them.
corner of the outer enclosure wall employed hori- bibliography: Paulin 1890; Rivoira 1925: 204–10;
zontal formwork on a series of closely placed frames. Caraffa 1942; Brödner 1951: 28–9, 35–6; Heres 1983:
The length of the boards in the lower parts of the 233–7, Candilio 1985: 525–32; Rasch 1991: 373;
semidome are 88 cm (3 RF) long and then some- Bianchi 2000: 136–40; LTUR 5: 53–8.
what longer higher up. The use of horizontal form-
work was typical for the semidomes at the Baths of 28. basilica of maxentius
Diocletian, as is evident from the imprints on other (a.d. 307–315)
smaller apsidal vaults in the niches both inside and comments: The Basilica of Maxentius was begun
outside of the dome of the “Planetarium.” How- after a fire destroyed the buildings on the Velia in
ever, one semidome (c. 22-m dia) in the double-apsed a.d. 307. The Basilica must have been conceived as
197
P1: JPJ
part of the rebuilding of the Velia including the Tem- ditional buttressing on the west side of the building.
ple of Venus and Rome. It was unlike all previous See further, page 138 (Figs. 121–122).
basilicas in its incorporation of imperial bath con- bibliography: Choisy 1873: 54–5, figs. 24–25, pl. 3;
struction, particularly in the triple cross vaults cover- Rivoira 1925: 213–14; Minoprio 1932: 1–25; Heres
ing the nave. 1983: 223–32; Bianchi 2000: 148–50; Amici 2003:
Materials. The pozzolana used in all the mortar is 17–29; Calabresi and Fattorini 2003: 81–9; Coccia
the red variety with occasional black. The vaults em- and Fabiani 2003: 30–51; Amici 2005: 21–74, 125–
ploy a variety of different types of caementa includ- 60; LTUR 1: 170–3.
ing tufo giallo della via Tiberina, tufo lionato, brick, and
pumice. The fallen section of vaulting lying in the 29. baths of maxentius on the
nave has an abundance of tufo giallo above the ribbed palatine (a.d. 306–312)
sections. Recent excavations and core samples on the comments: Maxentius renovated and expanded the
roof of the north barrel vaults reveal the use of light Severan Baths on the Palatine by enlarging the exist-
gray pumice in the vaults. The upper part of the west ing substructures with a series of very tall cross vaulted
wall of the nave employs some leucititic lava as cae- and barrel vaulted structures and then constructing a
menta. The use of the heavy leucititic lava was per- new bath building above.
haps an attempt to provide a surcharge to counter the Materials. Caementa of tufo lionato are used in all the
thrusts of the vaults. The extrados of the central nave vaults.
were covered with terracotta roof tiles, as can been Centering. Projecting travertine corbels (25 cm
seen from fallen fragments on the ground. wide, 40 cm high, 1 m apart center to center) were
Centering. Both barrel vaults and cross vaults had used to support the centering frames of the cross vaults
octagonal coffers with diamonds in between, which (4.2-m span). The formwork imprints are still visible
had to be incorporated into the formwork. In some on the vaults (Fig. 92, p. 107).
places, the imprints of the boards within the coffers Ribs. Three-arch lattice ribs occur along the groins
can be seen. See also page 34 (Figs. 26–27). of the cross vaults, and continuous lattice ribbing
Ribs. The barrel vaults (24.5-m span) to either side occurs along the entire intrados of the barrel vaults
of the cross vaulted nave employ a double layer of (Figs. 85, 89, and 92).
lattice ribbing. See further, page 106. bibliography: Carettoni 1972: 96–104; Herrmann
Amphoras. Rivoira 1925: 213–14 mentioned the 1976: 403–24; Heres 1983: 238–41; Bianchi 2000:
existence of amphoras in the barrel vaults but did 140–2; LTUR 5: 60.
not describe them. Recent excavations on the roof
revealed one example of an Almagro 51c. There are 30. villa of maxentius on the
no examples in the existing fragments on ground, and via appia (a.d. 306–312)
they are not shown in any of the drawings of earlier The Villa of Maxentius is located at the third mile of
fragments that have since been removed. the Via Appia and consisted of a circular mausoleum
Buttressing. On the roof of the barrel vaults are the with a pedimented porch within a walled enclosure, a
remains of the freestanding buttressing arches that large circus, and the imperial palace, which is largely
once helped stabilize the cross vaults. The sloping unexcavated. It is attributed to Maxentius on the basis
nature of the terrain also created the necessity for ad- of an inscription (CIL VI, 1138) recording a dedication
198
P1: JPJ
to his deified son, Romulus. Brick stamps confirm the was decorated with marble revetment and the dome
dating. with painted plaster depicting scenes illustrating a
Materials. The walls are built of opus vittatum con- magistrate, which Rasch 1993: 83 argues was a mem-
sisting of one or two courses of brick (often reused) ber of the senatorial elite but not necessarily of the
alternating with a course of rectangular tuff blocks. imperial family. The porch was very unusual be-
The mortar is made with red pozzolana sometimes ing roofed with concrete cross vaults supported on
in large pieces up to 2.5 cm. The mix also contains a columnar structure (Pl. XII). Near the mausoleum
pieces of Luna marble (Heres 1983: 243, 314). The was later built a Christian cemetery basilica similar
caementa are of tufo lionato. to the one attached to the Mausoleum of Helena
Amphoras. E. Rodriguez Almeida 1999: 235–45 (Appendix 1.36), which has also raised many unan-
has recently published a study of the amphoras in swered questions about the owner of the mausoleum.
this monument and has calculated that the ramping Materials. The caementa in the dome consist of tufo
vault supporting the seating for the circus spectators lionato. Rasch 1991: Abb. 9, 16 noted that those in the
contained anywhere from six thousand to ten thou- crown are larger than those at the haunch and that the
sand amphoras. The vast majority are Dressel 23 am- change occurs at the point that the ribs begin.
phoras, but some sections contain significant numbers Centering. The formwork imprints left along the
of Dressel 20 amphoras as well. They are placed in the intrados show that the dome was built with radially
thick upper part of the vault and are staggered to fit as arranged formwork. The boards of the first three tiers
many as possible into the available space. Amphoras were 4 RF long and the uppermost tier 3 RF long.
were probably used in the dome of the mausoleum Ribs. Four ladder ribs were built along the major
as well. Only the podium of the mausoleum survives, cross axes of the dome (Fig. 94, p. 109). They begin
but in his study of the monument, Rasch 1984: 68 about halfway up the dome and merge at the crown.
found part of an amphora in a piece of fallen vault. Amphoras. For amphoras used in the porch roofing,
see pages 78–80 (Pl. XII).
bibliography: Heres 1983: 242–4, 312–14; Rasch
1984; Ioppolo and Pisani Sartorio 1999; Rodrı́guez- bibliography: Frazer 1969: 45–8; Rasch 1991: 315–
Almeida 1999: 235–45. 23; Rasch 1993: 51–2; Bianchi 2000: 143–7.
31. tor de’schiavi (a.d. 305–309) 32. octagonal hall of the “villa
comments: The Tor de’Schiavi was a domed mau- of the gordians” (early
soleum (13.2-m dia) with a pedimented porch lo- fourth century a.d.)
cated at the so-called Villa of Gordians on the Via comments: The Octagonal Hall of the “Villa of
Praenestina. The brick stamps found in the structure the Gordians” stands about 250 m west of the Tor
are Diocletianic, and Rasch 1993: 78–9 has proposed de’Schiavi. The monument has traditionally been
a date for the beginning of construction of around dated to the mid-third century based on a literary de-
a.d. 305 and extending no later than 309. The dome scription (SHA, Gord. 3.32), but the similarity in the
was still standing in the eighteenth century but has distinctive brick cornice detail to that on the nearby
since partially fallen. It was pierced with four circular Tor de’Schiavi and at Minerva Medica, both of which
windows in its haunch. The interior of the rotunda have brick stamps from the early fourth century,
199
P1: JPJ
suggests that the monument dates later than tradition- bibliography: Durm 1905: 297; Lugli 1915: 164–
ally thought. Luschi 1989–1990: 407–14 has come to 5; Mazzucato 1970: 348–9; Luschi 1989–1990: 407–
the same conclusion based on her study of the con- 34, esp. 407–14; Rasch 1991: 342–6; Bianchi 2000:
struction techniques, building typology, and stucco 142–3.
decoration. The structure was covered by an octag-
onal dome (11.40-m dia), which had a circular win- 33. “temple of venus and cupid” at
dow in the haunch of each of the eight panels, much the sessorian palace (a.d. 312–330)
like those at the Tor de’Schiavi. The structure was comments: The “Temple of Venus and Cupid” is
later adapted and changed into a medieval tower in actually the apsidal end of an audience hall built at
the thirteenth century by adding a second floor sup- the Sessorian palace of the empress dowager Helena.
ported by a central circular pilaster and an annular It was given this misnomer in the sixteenth century.
barrel vault (Fig. 58, p. 79). The only remains are the semidomed structure that
Materials. The caementa of the dome are all tufo formed the apse (17.35-m wide, 10.50-m deep). Re-
lionato and are laid radially within the cells of the ribs, cent research by D. Colli suggests that it dates from
whereas the caementa of the thirteenth-century addi- the Constantinian period.
tions consist of peperino. Materials. The brick faced walls employ much
Centering. The formwork boards of the dome were reused brick, and the caementa of the vault consist
arranged radially so that the first three tiers were 3 RF of tufo lionato.
long and the uppermost 2 RF long (compare the Amphoras. Two amphoras, one above the other, are
horizontal formwork used for the octagonal dome of visible at the corner of the semidome. The shape sug-
the Domus Aurea). gests Dressel 23 types, though I have not measured
Ribs. The dome contains lattice ribbing that them.
starts 34–57 cm above the impost and is continu- Ribs. The semidome was built with lattice ribbing,
ous throughout the preserved parts of the intrados which Rivoira 1925: 147 described as “rather irreg-
(Fig. 94, p. 109). The cross pieces of the ribbing ular ribs set according to the meridians, consisting of
form continuous rings every six to seven bricks and a single chain of brick.”
are set at an angle (rather than laid flat as at Minerva Buttressing. The apsidal structure is buttressed on
Medica). the exterior by walls that appear to have been added
Amphoras. Parts of the vaults have fallen revealing after the original construction, as they block the orig-
Dressel 23 amphoras (c. 40–3-cm internal dia, 59-cm inal windows. They are attached to the structure by
height [internal base to rim]). Some amphoras, placed means of travertine blocks partly embedded into the
upside down, are also visible in the haunches of the walling, like ties (Fig. 132, p. 145). More detailed study
vaults of the standing structure (Fig. 58, p. 79). A is required to determine whether these travertine ties
rough estimate of ten amphoras per side gives a total of were inserted into the original wall later (as seems
around eighty for the entire vault. The medieval addi- likely) or whether they were built into the wall with
tions also incorporated pots in the vaults, but these are the intention of adding the buttressing in a second
much smaller (35.4-cm internal dia, 33.5-cm height) phase.
than the Roman transport amphoras. See further, bibliography: Rivoira 1925: 147; Colini 1955: 164–
page 80. 8; Colli 1996: 771–815; LTUR 4: 308.
200
P1: JPJ
34. “temple of minerva medica” changes to pumice in the crown (Fig. 57, p. 78). The
(first half of fourth century a.d.) bipedales that form the rungs of the lattice ribbing are
placed radially to form the voussoirlike compartments
comments: The decagonal domed structure (24.84- within the rib, whereas the rings of bipedales are laid
m dia corner to corner; 23.5 m dia inscribed) is lo- horizontally; therefore, the radially laid rungs of the
cated in what was the Horti Liciniani. The dome lattice ribs are independent of the rings and do not
itself has fallen, but much more of it was standing unusually align with them. In some places, there is a
til 1828 when the final collapse took place (Fig. 143, clear delineation between the edges of the ribs and
p. 163). Nine of the sides were pierced by semicircular the adjacent mortar (Pl. XI). If the ribs and the ad-
niches (6.2-m dia) covered by concrete semidomes. jacent caementa infill were laid separately, the radially
The building was later modified by the addition of laid ribs must have been at least one step ahead (and
large buttresses between the niches on the southeast possibly more) of the horizontally laid concrete in-
side, by two large semicircular exedrae (c. 19.3-m dia) fill between them. The two panels to either side of
on the northeast and southwest sides, and by a biapsi- the cross axis have the remains of what appear at first
dal entry vestibule on the northwest side. The interior to be infilled windows (Fig. 57, Pl. XI), but the lad-
walls were revetted with marble facing, and the dome der ribs defining them at either side run up past the
was originally decorated in glass mosaic, which was rudimentary arch, and both Caraffa 1944: 19–20 and
later covered by plaster. In its second phase, the struc- Rasch 1991: 333 point out that they could never have
ture was supplied with fountains in the niches and the functioned as true windows and must have been used
two exedrae. briefly during the building process.
Materials. The caementa of the vaulted apses consist Amphoras. The horizontal courses of bipedales were
of tufo lionato, whereas the dome employs both tufo also used as planes on which to set four amphoras (1.35
giallo della via Tiberina in the haunches and grayish m apart at each of two levels) above each of the large
pumice in the upper part (Pl. XI). window openings (Frontispiece and Fig. 57). Caraffa
Centering. The imprints of the radially arranged 1944: 19, fig. 8 gives the dimensions of the amphoras
formwork boards show that the first tier was 7.5 RF, found: external diameter of 56 cm and heights of
the second tier was 6 RF, and the uppermost tier was 70 and 76 cm, which imply Dressel 20 amphoras as
7 RF (Fig. 95, p. 111). opposed to the smaller Dressel 23 type. The amphoras
Ribs. Each of the ten groins of the vault has a are placed in the concrete containing the caementa of
five-arch lattice rib at the base, which becomes a tufo giallo just below the level at which it changes
three-arch rib higher up (those to either side of the to pumice. If this arrangement was repeated for all
main axis extend further up as five-arch ribs) (Fron- ten sides, originally there would have been a total of
tispiece and Fig. 94, p. 109). The ribs are connected at least forty amphoras. However, Choisy 1873: 97,
at intervals by horizontal courses of bipedales that run pl. 11 notes that he saw an amphora embedded in the
through the thickness of the vault to form the cover- wall above one of the “archivolte” of the doorways, in
ing of the step-rings along the extrados. These courses which case there may be more amphoras lower down
seem to mark stages in the construction because they in the structure (see p. 238 n. 38).
occur at points where the intermediary ribs end and Buttressing. In the first phase, the only buttress-
also where the yellow tuff in the haunch of the vault ing consisted of the nine apsidal projections. In a
201
P1: JPJ
second phase, large buttressing walls were added to 9 wide. If these notations refer to oncie and minuti
the structure along with additional semicircular rooms (1 oncia = 1.86 cm = 5 minuti), the pots would be
that also acted as buttressing. See further, pages 144–5 21.6 cm high × 16.74 cm wide, which is much smaller
and Frontispiece. than the amphoras used by the Romans and would
bibliography: Choisy 1873: 96–8, pl. 11; Giovan- suggest that they were medieval. If so, this would be
noni 1904: 3–39; Caraffa 1944; De Angelis d’Ossat another example of medieval builders imitating the
1945: 3–6; Heres 1983: 356–60; Rasch 1991: 329–36; ancient technique visible in the remaining parts of the
Cima 1995: 53–69; Bianchi 2000: 134–6; Biasci 2000: structure, as at the Gordian Octagon (Appendix 1.32).
67–88; LTUR 3: 66–7. bibliography: Uggeri 1800: 55, pl. 13; Choisy 1873:
tav. VII; Töbelmann 1915: Taf. XVII; Tedone 1993
35. arch of janus (early to (pub. 1998): 195–202; Pensabene and Panella 1994–
mid-fourth century a.d.) 1995 (pub. 1998): 25–67; Bianchi 2000: 142–3;
comments: The quadrafrons arch (12-m square) is LTUR 3: 94.
located in the Forum Boarium. It is usually associated
with the Constantinian arch (arcus Divi Constantini) 36. mausoleum of helena “tor
listed in the Regionary Catalogues in Region 11. The pignattara” (a.d. 325–330)
central part between the four piers is covered by a comments: The Mausoleum was built on the prop-
cross vault (c. 7-m span). erty of Helena, the mother of Constantine, at the
Ribs. The central cross vault has ladder ribs along third mile of the Via Labicana. The mausoleum was
the groins (unlike the more typical three-arch lattice a domed rotunda added to the cemetery basilica con-
ribs). The cross pieces occur about every nine bricks. nected to the graves of the martyrs Marcellinus and
The outer edges of the barrel vaults of the four arch- Peter. It may have been conceived as a dynastic tomb
ways were also built with rudimentary ladder ribbing. for Constantine, but he was later buried in Con-
Amphoras. Excavations on the roof of the monu- stantinople, and the Mausoleum contained only the
ment revealed the use of amphoras (“vasi di terra- porphyry sarcophagus of Helena, who died around
cotta”) embedded in the crown of the vault over north a.d. 330. A coin found in the mortar for the set-
arch. The type of amphora is not given, but the mea- ting bed of the revetment of the mausoleum dates to
surements (50-cm dia and c. 75–80-cm high) suggest a.d. 324–326, and the brick stamps are Constantinian
the Dressel 20 type. An upper story thought to be a (CIL XV, 1569). The dome (20.20-m dia) appears to
medieval addition was destroyed in 1830 by Valadier, have been changed soon after it was begun. Horizon-
but some doubt exists as to whether it was actually tal bipedalis courses extend through the vault to form
medieval or part of the original fourth-century build- the tops of the original step-rings along the extrados,
ing. It is today often assumed to have been part of the which were then covered in the second phase. It was
original attic (Pensabene and Panella 1994–1995 (pub. decorated on the interior by mosaic, the glass tesserae
1998): 31). I add one more piece to the puzzle with of which consisted of blue, green, and gold.
the following observation: Detailed drawings of the Materials. The mortar is made with pozzolanella in
demolished upper part of the arch by A. Uggeri in the walls, red pozzolana in the dome, and black poz-
1800 show that pots were also embedded in it. On zolana in the bedding mortar for a course of bipedales
his drawings, these pots are labeled as 11 3 high by at the top of the rotunda wall and for another at the
202
P1: JPJ
bottom of the wall. The caementa consist of tufo lionato Lago di Albano. The remains are now part of the pa-
throughout. pal villa at Castelgandolfo. The main preserved struc-
Centering. The formwork boards were arranged ra- tures include a theater, the so-called palazzo imperiale
dially with the lower two tiers 5 RF long and the sur- (largely unexcavated), and a cryptoporticus (7.45-m
viving upper tier 4 RF long. The ends of the boards span), which stretched for over 300 m and acted as
align with the spring of the brick ribbing and the a retaining wall for one of the terraces on which the
course of bipedales that covered the first phase step- villa was built. Brick stamps confirm the Domitianic
rings (Fig. 42, p. 57). dating of the theater and the cryptoporticus. Less is
Ribs. The surviving ribbing, which begins at 2.90 known about the “palazzo imperiale,” which is now
m above the impost, consists of six-arch lattice ribs very overgrown. My own observations of the brick-
with cross pieces about every eight bricks. Only one work of the walls (25 cm = five bricks and five joints)
of the ribs remains (along one of the diagonal axes). and the highly developed form of the brick linings on
Originally there would have been either four or eight the some of the vaults (containing upright bessales)
ribs. lead me to believe that at least parts of it were built
Amphoras. The monument takes its modern name in the second century a.d.
of “Tor Pignattara” from the amphoras that are now Materials. The caementa for all concrete in the villa
visible in the remains of its dome. They are distributed is the local lapis Albanus (peperino) into which the
in three rows at two different levels (Figs. 42, 47, villa is built. The mortar contains mainly granules of
pp. 57, 69). The lower level contains two rows of black pozzolana.
amphoras while the upper level contains one. All are Centering. One stretch of the cryptoporticus is built
placed upside down on the courses of bipedales that with coffers in the barrel vault, parts of which have
covered the step-rings. Rodrı́guez-Almeida 1984: fallen. In some places where the coffers can be exam-
167 has identified them as Dressel 23 amphoras. A ined, the remains of horizontal formwork imprints
recent publication by Vendittelli 2002: 775 identifies can be seen on the central panel. Lugli 1957: 681 lists
them as Dressel 20 types, but the hollows in the base the theater at the Villa of Domitian as having brick
seen in Fig. 47 suggests that the Dressel 23 type is linings, but in fact the brick linings only occur in the
most likely. Rasch has calculated that there would “palazzo imperiale.”
have originally been 180 amphoras. Ribs. For the ribs in the cryptoporticus vault, see
bibliography: Heres 1983: 309–11; Rasch 1991: pages 95–6 (Figs. 74–76).
346–8; Chiari et al. 1996: 1–36; Rasch 1998; Bianchi bibliography: Lugli 1918: 57–63; Hesberg 1978–
2000: 150; Vendittelli 2002: 771–92. 1980: 305–24.
outside of immediate environs 38. arco di malborghetto (first

of rome quarter of fourth century a.d.)
comments: The quadrifrons arch (14.9 × 11.9 m)
37. villa of domitian at is located on the Via Flaminia about 19 km out-
castelgandolfo (a.d. 81–96) side of Rome and is generally assumed to have com-
comments: Domitian built a villa overlooking the memorated Constantine’s victory over Maxentius in
Ariccia valley along the west crest of the crater of a.d. 312, which occurred in this area. A Diocletianic
203
P1: JPJ
brick stamp (a.d. 292–305) found in the vault accords it was combined with fragments of travertine and
with such a date. The structure was once covered brick.
in marble revetment, most of which has been spoli- Ribs. The central cross vault has three-arch lat-
ated. It was later turned into a farmhouse. The cen- tice ribs along the groins. The barrel vaults contain
tral cross vault and the four adjacent barrel vaults still haphazardly placed lattice ribbing. The arches of the
remain. ribbing often are not continuous and seem to be re-
Materials. The arch is built into banks of tufo gi- lated to different layers of work divided by courses of
allo della via Tiberina and contains the same material bipedales.
as caementa in its walls and vaults. The tufo giallo della bibliography: Töbelmann 1915; Messineo 1989;
via Tiberina was also used in the foundations where Bianchi 2000: 147–8.
204
P1: JPJ
APPENDIX 2
CATALOGUES OF BUILDING
TECHNIQUES
the following catalogues provide my own not cited directly, so I have included the lists
collection of examples of the techniques discussed in this Appendix for those interested in further
in the main chapters of the book. Many are information.
catalogue 2a. formwork imprints on barrel vaults
# Date Monument Width cm Length m Reference

1 Caes? Forum Romanum 25–35 1.0–5.0 Carettoni 1956–1959: 26 n. 7
2 Aug Forum of Caesar 27 – Personal observation
3 Ner Domus Tiberiana 20–5 – Van Deman 1924: 387
4 Dom Colosseum 25–30 – Personal observation
5 Dom S. Clemente 27–9 2.95 (10 RF) Personal observation
6 Tra Basilica Ulpia 30–5 – Personal observation
7 Ant Sette Bassi 28 ∼4.5 max Personal observation
205
P1: JPJ
APPENDIX 2. CATALOGUES OF BUILDING TECHNIQUES
catalogue 2b. formwork imprints on domes
Board Tiers ×
# Date Monument Dia. m orient. length (RF) Reference
1 Ner Domus Aurea, 13.5 hor – Personal
Octagon observation
2 Had Hadrian’s Villa, Large ∼13.5 rad – Rakob 1988:
Baths Caldarium Taf. 115.3
3 Had Hadrian’s Villa, Piazza d’Oro – rad – Rakob 1988:
Octagon Taf. 116
4 Max Tor de’Schiavi 13.2 rad 3 × 4 Rasch 1991:
∼7 × 3 Tab. 2, Abb. 7
5 Max/Con Gordian Octagon 11.4 rad 3 × 3 Rasch 1991:
∼10 × 2 Tab. 2, Abb. 29
6 Con Minerva Medica 24.3 rad 1 × 7.5 Rasch 1991:
1 × 6 Tab. 2, Abb. 20
5 × 7
7 Con Mausoleum of Helena 20.3 rad 2 × 5 Rasch 1991:
∼10 × 4 Tab. 2, Abb. 31
catalogue 2c. formwork imprints on semidomes

# Date Monument Dia. m Board orient. Reference
1 4th c. Domus Parthorum ∼8.0–8.5 vert Personal observation
2 Dio Baths of Diocletian, Semidome 4.3 vert Personal observation
3 Dio Baths of Diocletian, Semidome outer enclosure ∼42 hor Rasch 1991: Taf. 83.3
4 Dio Baths of Diocletian, “Planetarium” ∼6.5 hor Personal observation
∼4.2
5 Max Basilica of Maxentius, Niche in enclosure wall ∼3.5 vert Personal observation
206
P1: JPJ
catalogue 2d. brick linings on the intrados of vaults
Rome
Form/ Brick Bes Bes Upright

# Date Monument span m type grid solid bricks Reference
1 Tra Trajan’s Markets barrel bes/bip × Lancaster 1998a:
2.3–5.0 299–305
2 Tra Forum of Caesar barrel bes/bip × Lancaster 1998a:
Latrine 5.0 304
3 Tra Sette Sale barrel ses/bip × Cozza 1974–1975:
5.4 79–101
4 Tra Atrium Vestae barrel bes/bip × Personal
observation
5 Had? Baths of Titus barrel bes/bip × Personal
observation
6 Had? Domus Tiberiana barrel Personal
observation
7 Had Pantheon barrel bip Personal
10.4 observation
semi- bip
dome
9.2
9 Had Mausoleum of barrel Lugli 1957:
Hadrian pl. CCIV, 3
10 Had Janiculum Baths barrel Blake 1973: 93
1.5
11 Had/Ant Insula on Palatine barrel Blake 1973: 86
12 Ant? S. Lorenzo in Lucina, barrel Personal
Insula observation
13 Ant/Sev Colosseum Recon barrel bes/bip × × Lancaster 1998b:
5.0 160
14 Sev Lateran, Sessorian barrel bes/bip × × × Personal
Palace observation
15 Sev? Domus Flavia, barrel bes Personal
Cryptoporticus observation
16 Sev Palatine Balneum barrel bes/? × Cassatella and Iacopi
1991: 134–5, fig. 10
17 Sev Basilica Argentaria cross bes/bip Personal
observation
18 Sev Amphitheatrum barrel bes/ses? Personal observation
Castrense
19 Sev Monte del Grano Lugli 1957: 681
(continued )
207
P1: JPJ
catalogue 2d (continued )
Rome
20 Sev Severan Baths Palatine barrel bes/bip × Personal observation
21 Sev Baths of Caracalla barrel bes/bip × × DeLaine 1997: 165–6;
Personal observation
cross
semidome
22 Sev Titulus Equiti cross bes/bip × × Personal observation
23 Sev? Horrea Agrippiana, barrel Personal observation
Courtyard
Environs of Rome
24 Tra Cistern, Via C. Ramieri 1992: 92–3;
Columbo Ramieri 1993: 72–3
25 Had? Villa of Domitian, barrel bes/bip × × × Personal observation
Castelgandolfo
cross
26 Had Terme Taurine barrel bes/bip × × Personal observation
cross bes/bip × × Personal observation
27 Ant Villa dei Antonini barrel bes/bip × Cassieri and Ghini 1990: 176,
figs. 13, 19
28 Ant Villa di Sette Bassi barrel bes/ses? × × Lugli 1957: pl. CCVI, 2; Lupu 1937:
155; Personal observation
29 Ant Villa dei Quintili barrel bes/bip × × Personal observation
30 Sev Magazzini barrel bes/bip Personal observation
Severiani, Portus
31 Temple of barrel bes/bip × Lugli and Filibeck
Portunus, Portus 1935: fig. 60
semidome bes/bip ×
32 Sev Ponte Lupo barrel ? Personal observation
Hadrian’s Villa
33 Had Maritime Theater barrel bes/bip × × Personal observation
34 Had Heliocaminus Baths semidome bes/bip × × Personal observation
208
P1: JPJ
Hadrian’s Villa
cross
35 Had Large Baths cross 12 ×20 × Personal observation
36 Had Serapeum barrel ses Personal observation
37 Had Torre Roccabruna semidome Personal observation
Ostia
38 Had Capitolium 3.0–3.6 bes/bip × Personal observation
39 Had? I.1.3 barrel bip Personal observation
40 Had Insula dei dipinti (I.4.4) barrel bes/ses? × Personal observation
1.18
41 Had I.16.2 fallen vault Personal observation
42 Had Terme del Mithras (I.17.2) barrel bes/bip × × Personal observation
1.0–2.0
cross ∼4.0 bes/bip ×
43 Had Magazzini “Traianei” (I.20.1) barrel bes/bip × × Personal observation
∼5.0
cross bes/bip × ×
44 Had? Terme di barrel bes/ses? × Personal observation
Nettuno, cistern
45 Had? Case di Giardino barrel bes/bip × × Personal observation
(III.9.9,13–18,22) 1.5–3.0
46 Had? Caseggiato del barrel bes/bip × Personal observation
Serapide (III.10.3) 3.0–6.0
cross ∼5.0 bes/bip ×
47 Had Caseggiato di Annio (III.14.4) barrel ∼3.0–5.0 bes/bip × Personal observation
cross bes/bip ×
48 Had Loggia di Cartilio (IV.9.1) cross Blake 1973: 214
49 Ant Casa di Diana (I.3.3–4) barrel bes/bip × × Personal observation
cross bes/ses × ×
50 Ant Terme del Foro, Praefurnium barrel 1.9 bes/bip × × Personal observation
(I.12.6)
51 Ant I.8.9 barrel Personal observation
cross bes/bip ×
(continued )
209
P1: JPJ
catalogue 2d (continued )
Ostia
52 Ant Domus di Apuleio (II.8.5) barrel 1.2 bes/bip × Personal observation
53 ? III.2.11 barrel bes/bip × Blake 1973: 214
∼3.0–5.0
54 Ant Caseggiato delle Trifore (III.3.1) barrel ∼1.5 bes/ses × Personal observation
55 Ant Caseggiato degli Aurighi (III.10.1) barrel bes/bip × × Personal observation
2.0–3.0
56 Ant Casseggiato dell’Ercole (IV.2.3) cross 5.4 bes/bip × Personal observation
57 Ant? Terme delle 6 colonne 1.3 bes/bip × Personal observation
(IV.5.10–11)
3.8 frags ×
58 Ant Scuola di Traiano (IV.5.15) barrel bes/ses × × Personal observation
1.5–3.5
59 Ant Terme della Tricrinaria (III.16.7) barrel 2.5 bes/bip × Personal observation
60 Ant Theater (II.7.2) barrel 4.7 bes/bip × × Personal observation
61 Ant Macellum (IV.5.1) barrel bes/bip × Personal observation
1.6–2.6
62 Ant Temple (V.11.1) cross 5 bes/bip × × Personal observation
63 Sev Terme di Sette Sapienti (III.10.2) cross bes × Lugli 1957: pl. CCVI,1
3.0–4.0
64 Sev? I.12.5 barrel bes Personal observation
65 Sev? Molino di Silvano (I.3.1) barrel 6.2 bes/bip × Personal observation
66 Sev? Casa delle Volte Dipinte, barrel 1.5 bes/bip Personal observation
Cistern (III.5.1)
210
P1: JPJ
catalogue 2e. uses of cocciopesto to protect the extrados of vaults

Protection of extrados
# Date Monument Context (bottom to top) Reference
1 Tib Castra terrace (1) cocciopesto, Blake 1959: 15
Praetoria (2) opus spicatum,
(3) cocciopesto,
(4) coarse mosaic
2 Ner Domus garden (1) 33-cm thick cocciopesto, Krause 1994: 44–8,
Tiberiana (2) suspensurae 3 bes high 77–8, 96–7; Krause
w/1-bip thick floor 1986: 444–5
3 Dom Domus garden as above Krause 1994:
Tiberiana 121–2
4 Dom? Domus garden (1) mosaic (lava) w/ Carettoni 1971:
Flavia tesserae 1–1.5 cm 321–3
per side (earlier phase?),
(2) suspensurae 1–2 bes
high w/1-bip floor,
(3) 12-cm thick cocciopesto,
(4) opus spicatum
5 Dom Domus terrace (1) 3-cm thick cocciopesto, Personal
Augustana (2) mosaic (lava), (3) 20-cm observation
Stadium thick cocciopesto, (4) opus
spicatum
6 Tra Forum of Caesar, terrace (1) 12-cm thick cocciopesto Amici 1991b: 109,
Basilica Argentaria w/broken brick up to 7 cm fig. 194; personal
long, (2) opus spicatum, observation
(3) 7.5-cm thick cocciopesto
w/broken brick up to 3 cm,
(4) mosaic (lava) w/tesserae
1–1.5 cm, (5) 7-cm thick
cocciopesto
terrace (1) 20-cm thick cocciopesto
containing pieces of
broken brick up to 10 cm,
(2) opus spicatum, (3) cocciopesto
w/broken bricks up to 5 cm,
(4) opus spicatum
7 Tra Trajan’s Markets large semidome (1) cocciopesto, (2) opus spicatum, Personal
terrace (3) cocciopesto, (4) mosaic (lava) observation
small semidome (1) cocciopesto, (2) tegulae/imbrices, Lancaster
(3) cocciopesto 2000; 766
large semidome (1) 15-cm thick Lancaster
cocciopesto 2000; 767
(continued )
211
P1: JPJ
catalogue 2e (continued )
Protection of extrados
# Date Monument Context (bottom to top) Reference
8 Tra Trajan’s Baths Section E (1) cocciopesto, (2) opus spicatum, de Fine Licht 1974: 34;
terrace (3) cocciopesto, (4) mosaic (lava) personal observation
Section H (1) cocciopesto, (2) mosaic (lava) de Fine Licht 1974: 38
stair landing
South exedra (1) cocciopesto, (2) opus spicatum De Romanis 1822: 19 no. 22
terrace
garden? (1) cocciopesto, (2) suspensurae 1-bes Fabbrini 1982: 15,
above Domus high with 2-bip figs. 21–23
Aurea thick floor
9 Had Domus Tiberiana garden (1) cocciopesto, (2) suspensurae of bes Krause 1994: 78;
w/2-bip thick floor Krause 1986: figs. 145–147
10 Had Pantheon above front (1) 46-cm layer of cocciopesto, de Fine Licht 1968: 63,
porch (2) 4-cm thick mosaic, 68 fig. 74
(3) 10-cm layer of concrete
of broken brick and
chalk, (4) layer of brick,
(5) 27-cm layer of cocciopesto,
(6) concrete with caementa of
bricks, tuff, and marble
fragments
ledge (1) cocciopesto, (2) Proconnesian Cozza 1983: 110
marble roof tiles: 1.52 m long,
0.16 m thick
dome (1) layer of semilateres Terenzio 1932: 54;
overlapping each other like de Fine Licht 1968: 240 n. 4
scales, (2) 12–15-cm thick
cocciopesto, (3) gilt bronze
roof tiles
11 Had Hadrian’s Villa, flat roof (1) opus spicatum, Salza Prina Ricotti 1988: 15–16
Tre Esedrae (2) cocciopesto, (3) mosaic
12 Sev Severan Baths, terrace? (1) cocciopesto (2) opus spicatum, Carettoni 1971: 313 n. 1
Palatine (3) cocciopesto, (4) mosaic
212
P1: JPJ
catalogue 2f. lightweight caementa

Vault Type/
# Date Monument TG Sc Pu Span m Comments Reference
1 Cae Forum of × barrel scoria used alone Personal observation
Caesar 4.9 Amici 1991b: 52, 162
2 1st c. Vaults on ×? ? Middleton 1892b: 55
b.c.? Palatine facing
Circus Maximus
3 Aug- Aula Isiaca ×? barrel nature of “pomice” Carettoni 1971:
Ner not described 326, fig. 31
4 Ner Neronian Structures × barrel Blake 1959: 47
on Velia
5 Ner Domus × ×? barrel nature of “pumice” Blake 1959: 22
Tiberiana, not described; yellow
Sector 9 tuff ranges from
20–30 cm long
and 5–10 cm high
6 Ves Colosseum × all Personal observation
Colosseum, × barrel pea to grape sized Personal observation
Levels 2 & 3 Stairs 3.5–3.9 pieces of white
pumice added
to mortar
7 Dom Domus Augustana × all Personal observation
8 Dom Domitianic Vestibule × barrel Personal observation
9 Dom Aqua Claudia × barrel Personal observation
Palatine 6.8
10 Dom Building under × barrel Blake 1959: 97
S. Clemente
11 Dom Porticus of the × barrel Blake 1959: 128
Dei Consentes
12 Tra Trajan’s Baths × most Personal observation
Trajan’s Baths, × × barrel 8.1 scoria alternated Personal observation
Section E cross 8.8 with rows of
semidome 12 tufo giallo
13 Tra Trajan’s Markets, × cross Lancaster 2000: 776
Aula 8.5
14 Tra Basilica Ulpia × × barrel scoria alternated Meneghini 1989:
6.6 with rows of 541–7; personal
tufo giallo observation
15 Tra Bibliotecae Ulpia × × barrel? Packer 1997: 452;
personal observation
(continued )
213
P1: JPJ
catalogue 2f (continued )
Vault Type/
# Date Monument TG Sc Pu Span m Comments Reference
16 Tra Forum of Caesar × barrel Personal observation
17 Tra Ostia (I.20.1/I.19.4) × ? scoria alone in Personal observation
vault fragment remaining fragment
18 Had Pantheon × × dome scoria alternating with De Angelis d’Ossat
43.3 tufo giallo in crown 1930: 211–15
19 Ant Hadrianeum × × barrel scoria alternated with Personal observation
∼18 tufo giallo in crown Cozza 1982: fig. 25a
20 Sev Baths of Caracalla, × barrel scoria alone in crown DeLaine 1997: 159
Palaestrae ∼7.3
Baths of Caracalla, × semidome scoria alone in crown DeLaine 1997: 159
Exedrae 13 25.5
Baths of Caracalla, × cross scoria alone in crown DeLaine 1997: 159
Frigidarium ∼21
21 Dio Basilica Julia × scoria alone Personal observation
22 Dio Baths of Diocletian, × dome type not indicated Caraffa: fig. 5
“Planetarium” 21.7
Baths of Diocletian, × dome white/yellow pumice De Angelis d’Ossat
Tepidarium 19.1 1946: 21–2
Baths of Diocletian, × barrel gray pumice Personal observation
Palaestrae ∼5.8
Baths of Diocletian, × cross black/white pumice Rivoira 1925: 106
Cross Vaults 14–21
23 Max Basilica of Maxentius × × barrel gray pumice Coccia and Fabiani
24.5 2003: 38, fig. 8
24 1st 1/2 Minerva Medica × × decagonal dome gray pumice De Angelis d’Ossat
of 4th c. 24.3 (diag) 1945: 3–6
25 2nd 1/2 Santa Costanza × × dome scoria and tufo giallo Giovannoni 1931: 29
of 4th c. 11.4 placed together
irregularly
214
P1: JPJ
catalogue 2g. amphoras in vaults

Vault type/ Pots in
# Date Monument span m Pot type walls # of Pots Reference
1 Had Magazzini “Traianei,” cross Dr 20 no > 12 Personal observation
Ostia 4–6
2 Had Villa alla barrel Dr 20 yes 8 Ashby 1907: 74–8;
Vignaccia crossdome personal observation
dome ∼10 Dr 20 5
barrel/cross Dr 20 13
5–11
3 mid- Tomb just past IV cross ? ? ? Rivoira 1925: 164–5
2nd c. mile of Via Appia
4 late “Casa di via barrel Africana no > 10 Packer 1968–69:
3rd c.? Giulio Romano” 1 tav. LIX, 2
5 Max Circus of Maxentius ramp barrel Dr 20/23 no 6,000– Rodrı́guez-Almeida
∼5 10,000 recon 1999: 235–45;
6 Max Basilica of barrel Almagro no 1 Coccia and Fabiani:
Maxentius 25 51c 38 n. 17, fig. 8
7 Max Mausoleum of no Rasch 1984: 65
Romulus (Appia)
8 Max? Gordian dome Dr 23 no > 10 Mazzucato 1970: 348–50;
Octagon 11.4 personal observation
9 Max Tor de’Schiavi cross Dr 23 (?) no 315 recon Rasch 1993: 51–2
10 1st 1/2 Minerva Medica dome Dr 20 yes? 40 recon Caraffa 1944: 19–20, fig. 8
of 4th c. 23.7 Choisy 1873: 97, pl. XI
11 Con “Temple of Venus semidome Dr 23? no 1 Rivoira 1925: 147;
and Cupid,” 17.5 personal observation
Sessorian
12 1st 1/2 Arch of Janus cross Dr 20 no ? Uggeri 1800: 55
of 4th c. ∼7
13 Con Mausoleum dome Dr 23 no 180 Rasch 1998: Taf. 22.4;
of Helena 20.3 or 20? recon Rasch 1991: 346–8;
Vendittelli 2002: 75;
14 Hon Aurelian Wall, ? ? ? ? Cozza 1987:
east tower of 42 n. 15, fig. 44
Porta Asinia
15 4th– S. Maura, semidome Dr 23 no 10 Quilici 1974: fig. 1709;
5th c. Via Casilina 5.4 personal observation
215
P1: JPJ
catalogue 2h. vaulting ribs

Rib type (ˆ=#bricks
Vault type/ between rungs
# Date Location Monument span m of lattice) Reference
1 1st c. b.c. Tivoli Sanctuary of barrel travertine Giuliani 1970: 184.
Hercules Victor ∼10
2 Ner Rome Esquiline barrel solid brick Personal observation
Wing of ∼3.9 (56) (unclear if they are original)
Domus Aurea, ∼2.9 (58)
Rms 56 & 58
3 Ves Rome Colosseum, barrel travertine Lancaster 1998b: 167,
Hypogea 3.3–3.9 figs. 19, 27, 28, 31
Colosseum, barrel ladder Lancaster 1995a: 57
Stair vaults ∼3.9
Colosseum, annular solid bip Lancaster 1998b: 153–6
Level 2 barrel
annular barrel 4.4–5.1
vaults of
ambulatories
Colosseum, barrel solid bip Lancaster 1998b: 156, fig. 14
Level 2 stairs 3.5
4 Dom Rome Domus barrel solid bip Lancaster 1995a: 78–9
Tiberiana, ∼6
Krause 1985a:
fig. 92, Sect 12,
Rms. 65, 67
5 Dom Rome Domus barrel solid bip Delbrueck 1921:
Tiberiana, ∼4.4, 14, 17; Lancaster
Krause 1985a: ∼6.5 1995a: 79–80
fig. 104, Sect. 15,
Rm 232
6 Dom Rome Domus Augustana, barrel projecting Lancaster 1995a: 68
Cryptoporticus arch
7 Dom Castel- Villa of Domitian, barrel solid bip Lugli 1918: 61;
gandolfo Cryptoporticus 7.5 w/roof tiles Lancaster 1995a: 188
8 Dom Chieti Cistern barrel rudiment. Colosimo 1938 (pub. 1940):
5.8 solid ribs 254; Blake 1959: 149–50
9 Dom? Cumae Arco Felice barrel solid bip Blake 1959: 142;
10 Dom Rome Aqua Claudia barrel solid bip Colini 1944: 105, tav. III;
on Palatine ∼6.8 personal observation
216
P1: JPJ
Rib type
(ˆ=#bricks
11 Tra Rome Forum of barrel ∼4.9 solid bip Amici 1991b: 116–18, fig. 210;
Caesar Lancaster 2000: 779–84,
Latrine figs. 27, 29
12 Tra Rome Trajan’s Markets, barrel solid bip Lancaster 2000: 774–5
Aula Level 4 rooms 4.1–6.1
Trajan’s Markets, barrel solid bip Lancaster 2000: 765–6,
Small hemicycle 1.8–4.7 figs. 3, 11
Trajan’s Markets, barrel solid bip Lancaster 2000: 765–6, fig. 9
Stair vault south 4.6
end of hemicycle
Trajan’s Markets, barrel solid bip Lancaster 2000: 771–2
Level 1 hemicycle 2.7–3.4
13 Tra Rome Trajan’s Forum barrel solid bip Personal observation
(against Terrazza
Domizianea)
14 Tra Rome Trajan’s Forum, barrel solid bip Amici 1982: 64–5, fig. 104;
Library Portico 5.2 Lancaster 2000: 423–4, figs. 2.A, 3
15 Had Rome Pantheon, barrel solid bip de Fine Licht 1968: 96–100,
Rotunda wall 11.8, 5.4 137, fig. 99
South building barrel solid bip de Fine Licht 1968: 161–2,
5.2 figs. 180–183
16 Had Tivoli Hadrian’s Villa, barrel solid bip Bianchi 2000: 130–1 n. 66;
Cryptoporticus: personal observation
Salza Prini Ricotti
1973: tav. I, 21
Hadrian’s Villa, barrel solid bip Personal observation
Praetorium
17 Had/Ant? Rome Palatine barrel solid bip Romanelli 1973: 209, fig. 3
Cryptoportico 4.1
18 Had Civita- Terme Taurine barrel solid ses Lugli 1957: 625, 682; Köhler 1999:
vecchia 365–76; Bianchi 2000: 130–1
n. 66; personal observation
19 Had/Ant? Rome Palatine private barrel ladder Blake 1973: 86; Lugli 1946:
structure 450–1; tav. VIII, 4
20 Ant Rome Villa di Sette cross ladder ˆ6–10 Ashby 1907: 105; Rivoira 1925:
Bassi 7.6 140–3; Lupu 1937: 153–5, fig. 31;
Blake 1973: 108–9; Bianchi 2000:
124–5; personal observation
(continued )
217
P1: JPJ
catalogue 2h (continued )
Rib type
(ˆ=#bricks
Villa di Sette barrel ladder
Bassi 4.7 ˆ6–10
Villa di Sette barrel ladder
Bassi 4.5
Villa di Sette Bassi, barrel rudiment.
Cryptoporticus 4.5 ladder
21 Ant? Rome Baths of “Severus barrel ladder or Castagnoli 1949–1950: 125–7
and Commodus” 3.4–3.7 lattice?
22 Sev Palatine vaults along barrel ladder Personal observation
Via dei Cerchi ∼5–6 ˆ4–8
23 Sev Rome Domus Augustana, barrel lattice Personal observation
Hippodrome
24 Sev Rome, Ponte Lupo barrel lattice Ashby 1935: 120
Environs 3.6 ˆ4–6
25 Sev Portus Temple of dome projecting Lugli and Filibeck 1935: 93–4,
Portunus 14.5 ribs fig. 60; De Angelis d’Ossat
1938: 143; Rasch 1991: 371
26 Sev Rome Aqua Claudia in barrel lattice Colini 1944:
Villa Wolkonsky 93–7; personal
(Severan recon) observation
27 Sev Rome Baths of Caracalla, cross lattice Cozzo 1928: 72, figs. 93, 99, 102;
Room 5w 16 × 16 DeLaine 1997: 160–5; personal
observation
Baths of Caracalla, barrel ladder
Room 21w 18.0 ˆ5–6
Baths of Caracalla, barrel lattice
Pier between 8.5 ˆ3–4
tep and cald
Baths of Caracalla, barrel solid bip
Caldarium piers 8.5
Baths of barrel solid bip
Caracalla, 13.8–15.2
Frigidarium
niches & natatio
central niche
Baths of barrel solid bip
Caracalla, 8.5
Rooms 3a, b, c, d
218
P1: JPJ
Rib type
(ˆ=#bricks
28 Sev Rome Nympheum barrel 2.8 ladder
Alexandri
29 Sev Rome Colosseum, annular lattice Lancaster 1998b:
Level 2 barrel ˆ3–4 153–6
ambulatories 4.4–5.1
Colosseum, barrel 3.5 lattice Lancaster 1998b:
Level 2 stairs ˆ3–4 156, fig. 15
Colosseum, barrel lattice Lancaster 1998b:
Hypogea 3.3–3.9 167–9, fig. 27
Colosseum, annular ladder Lancaster 1998b:
Level 1 & 2 barrel ˆ6 158–60
outer ambulatories 5.1
30 3rd/4th c. Rome Baths of dome lattice Choisy 1873: 81;
Agrippa ∼24.0 ˆ10–12 Rasch 1991: 350–3;
Bianchi 2000: 133–4;
31 mid-3rd c. Rome, Tomb of dome lattice Rivoira 1925: 189–90;
Via Appia Calventii 9.5 Rasch 1991: 371,
373
32 mid-3rd c. Rome Baths of semidome ladder La Follette 1994:
Trajan Decius ∼5.0 44, fig. 8
33 Dio Rome Baths of cross lattice Rivoira 1909–1910:
Diocletian, ∼20.0 359
Frigidarium and ∼23.7
surrounding
cross vaults
Baths of Diocletian, semidome lattice Personal observation
Exedra of outer wall ∼42.0
Baths of Diocletian, octagon lattice De Angelis d’Ossat
“Planetarium” 21.7 1933: 12; personal
observation
Baths of dome solid brick Paulin 1890: 13;
Diocletian, ∼22.4 De Angelis d’Ossat
San Bernardo 1938 (pub. 1940): 248;
Rasch 1991: 371, 373
(continued )
219
P1: JPJ
catalogue 2h (continued )
Rib type
(ˆ=#bricks
34 Max Rome Baths of cross solid brick Carettoni 1972: 102, n. 14;
Maxentius, 5.4 on groins Bianchi 2000: 140–2; personal
Palatine observation
Baths of cross lattice
Maxentius, 4.9
Palatine
Baths of barrel lattice
Maxentius, 5.6 ˆ9
Palatine
35 Max Rome Basilica of cross lattice Choisy 1873: 56, fig. 3;
Maxentius ∼30.0 ˆ6–9 Bianchi 2000: 148–50;
Amici 2005; personal observation
Basilica of barrel lattice
Maxentius 24.5
36 Max Rome, Tor dome ladder Rasch 1991: 315–23
Via Prenestina de’Schiavi 13.7
37 Max? Rome, Gordian octagonal continuous lattice Rasch 1991: 342–6;
Via Prenestina Octagon dome begins 34–57 cm Bianchi 2000: 142–3;
11.4 above impost personal observation
38 1st 1/2 Rome Minerva decagonal lattice ladder Choisy 1873: 86, pl. XI;
of 4th c. Medica dome Giovannoni 1904: 17–21;
∼23.9 Caraffa 1944: 15–16; personal
observation
Minerva semidome lattice
Medica 6.3–6.9 lattice
Minerva barrel lattice
Medica 3.2–3.3
39 Con Rome “Temple of Venus semidome lattice Rivoira 1925: 147; Colli 1996:
and Cupid,” Sessorian 17.3 784–6; personal observation
40 Con Rome, Arco di cross lattice Messineo 1989: 47–52;
Via Flaminia Malborghetto 6.0 × 5.4 Toebelmann 1915: 3, 13–16;
Bianchi 2000: 142–3
Arco di barrel lattice
Malborghetto 5.4–6.0
41 Con Rome Mausoleum of dome 20.3 lattice Rasch 1991: 346–50;
Helena personal observation
220
P1: JPJ
Rib type
(ˆ=#bricks
42 1st 1/2 Rome Arch of Janus cross ladder Tedone 1993 (pub. 1998):
of 4th c. 7.3 195–202; personal observation
43 mid-4th c. Tivoli “Tempio della dome rudiment. Giuliani 1970: 203;
Tossa” 12.4 lattice Rasch 1991: 323–8
44 4th c. Rome “Domus semidome lattice Mancioli et al. 1993: 56,
Parthorum” ∼8.0–8.5 ˆ7–8 fig. 5; personal observation
45 mid-4th c. Rome, Via S. Costanza dome lattice Rivoira 1925: fig. 294; De Angelis
Nomentana 11.4 d’Ossat 1938 (pub. 1940): 248;
Giovannoni 1938: 215; Rasch 1991:
336–42
46 mid-4th c. Rome Temple of Romulus, dome ? Rasch 1991: 336–42
Forum Romanum 14.7
catalogue 2i. blocks with tie bar cuttings
Entab. Anch. Vault Bar Support

# Date Monument block block span m space m type/material Reference
1 Aug Horrea × 3 ∼5.0–5.3 arcade Bauer 1978: 139–46;
Agrippiana (?) tufo lionato/ Bauer and Pronti
travertine 1978: 113
2 Aug Basilica × 7 ∼6 arcade Bauer 1988: 200–12
Aemilia (?) white marble
3 Tra Trajan’s Baths × 6 ∼4 colonnade, Amici 1997: 85–95;
Palastrae granite (?) DeLaine 1990: 407–24
4 Tra Basilica Ulpia × 6.6 ? colonnade, Amici 1982: 22;
pavonazzetto giallo Packer 1997: 441
antico, Mons
Claudianus granite
5 Tra Trajan’s Column, × 5.7 1.5, 1.0 colonnade, Amici 1982: 84–5;
Portico pavonazzetto Packer 1997: 446
6 Sev Baths of Caracalla, × 7.3 3.0–4.8 colonnade, Mons DeLaine 1985: 198–202
Palaestrae Claudianus granite
7 Dio Baths of Diocletian, × 5.9 3.5–10.2 colonnade, gray Brödner 1951: 28;
Palaestrae granite Amici 1997: 85–95
221
P1: NDZ/KDF
APPENDIX 3
SCORIA ANALYSIS
the intention of the analysis was to determine Because the color of the scoria varies from dark
the most likely provenance of the scoria samples taken brown to reddish brown (Pl. VIII, IX), samples from
from the caementa of vaults of five buildings in Rome each end of the spectrum were included. The size
dating from the mid-first century b.c. to the late third of the vesicles can also vary, so samples with both
century a.d. The samples are visually very similar to large (> 1 mm) and small (0.1 mm) were chosen. The
the scoria used in the dome of the Pantheon, which sample (Basilica Julia B) with small vesicles is at the
was determined by Gioacchino De Angelis d’Ossat red end of the spectrum making it look very similar
in 1930 to have been a product of Vesuvius.1 Because to a large piece of pozzolana rossa. This piece was
there has recently been some suggestion that a sim- chosen as it most closely resembled material from the
ilar looking material produced by the Colli Albani Colli Albani system. As a control sample, one piece
system just south of Rome also may have been used of dark brown scoria from an opus caementicium wall
for vaulting in Rome,2 this analysis is designed to at Pompeii also was included.
determine whether the material from Vesuvius con- The results show that none of the scoria samples
tinued to be imported over a long period or whether was compatible with material from Colli Albani be-
it was replaced by a local but similar-looking ma- cause all contain sanidine in the groundmass as well as
terial. In his later study of the lightweight material significant amounts of plagioclase (Table 8).3 Neither
from the “Temple of Minerva Medica,” De Angelis of these minerals is present in deposits erupted from
d’Ossat found pumice produced by the Sabatini the Colli Albani volcano.4 In addition, the presence of
system north of Rome. In the present study, olivine and the lack of phenocrysts of sanidine rules
petrographical analyses of thin sections were used to out the Sabatini system.5 Rather, the samples seem
identify the crystal fragments within each sample, and to have originated in Vesuvian deposits as shown by
then the resulting mineralogical profile was compared the similarity to the sample from Pompeii with its
to the compositional data for volcanic deposits from phenocrysts of pyroxene and plagioclase. The lack of
the three most likely volcanic districts to have sup- sanidine phenocrysts and the presence of sanidine mi-
plied the Roman builders: Vesuvius, Colli Albani, and crocrysts in the groundmass is characteristic of the
Sabatini. products of Vesuvius from minor eruptions in the
222
P1: NDZ/KDF
APPENDIX 3. SCORIA ANALYSIS
table 8. Scoria analysis (see Glossary for geological terms)
Forum of Trajan’s Basilica Baths Baths Basilica Basilica

Source Pompeii Caesar Baths E Ulpia Caracalla A Caracalla B Julia A Julia B Colosseum
color brown brown brown brown red brown brown red white
vitrophyric × × × × × ×
Texture
porphyric × × × × × × × × ×
scoriaceous × × ×
Vesicule Size > 1 mm > 1 mm > 1 mm > 1 mm > 1 mm > 1 mm > 1 mm 0.1 mm 0.1 mm
Porphyricity Index low low med low-med low-med low-med low-med low med
pyroxene ++ + ++ ++ ++ ++ ++ + −
plagioclase + ++ ++ ++ + + + ++ +
leucite + + + +
Phenocrysts
relics of olivine + + + + + + +
magnetite − − − −
phlogopite + +
sanidine ++
mica −
pyroxene ∗ ∗
plagioclase ∗ ∗ ∗ ∗ ∗ ∗ ∗
Groundmass
leucite ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗
magnetite ∗ ∗ ∗ ∗ ∗ ∗
sanidine ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗
glass zeotilised zeotilised zeotilised zeotilised zeotilised zeotilised zeotilised zeotilised zeotilised
oxidized
Classification tephrite tephrite- tephrite tephrite tephrite tephrite tephrite tephrite phonolite-
phonolite trachyte
tenth/ninth century b.c. (the last eruptive activity same quarry. Supporting this hypothesis is the fact
before the 79 eruption) and from the 79 eruption that the groupings occur according to date. Both
itself; however, recent studies of the 79 eruption do Trajanic samples are distinguished by the presence of
not indicate the presence of the same type of dark phenocrysts of phlogopite and microcrysts of pyrox-
brown/red scoria as those analyzed.6 ene in the groundmass. The samples from the Baths of
Some differences between the samples allow Caracalla and one from the Basilica Julia (A) are distin-
groupings that may indicate provenance from the guished by leucite and magnetite phenocrysts. These
223
P1: NDZ/KDF
APPENDIX 3. SCORIA ANALYSIS
groupings suggest that there was a change of quarry testing. More important is the confirmation that im-
sites between the early second century and the early ported scoria was being used continuously for over
third century. Further analyses of Hadrianic sam- three centuries with a probable change in quarry
ples from the Pantheon and the Hadrianeum could sources during that time.
provide additional evidence for this hypothesis. The One sample of white pumice (Pl. V) found in the
reddish sample with small vesicles from the Basilica mortar of the vaults of the Colosseum was also in-
Julia (B) was chosen as the most likely sample to cluded in the analysis. It is different from the scoria
have come from the Colli Albani; however, the re- samples in that it contains an abundance of sanidine
sults show that it is rich in plagioclase and sanidine, phenocrysts and some mica. Thus far, the most com-
thus ruling out Colli Albani. In spite of the differ- patible materials I have found in terms of mineral-
ence in color and texture, it has a similar profile to ogy come from the Vulsini volcanic district around
the other samples, indicating that it too came from Lago di Bolsena (Map 2, p. 13), which produced great
Vesuvius. quantities of pumice. The most likely deposits con-
The analysis shows that scoria from Vesuvius was centrate along the northeastern rim of the lake.7 The
imported into Rome for vaulting from the mid-first dates of the events producing the Vulsini pumice de-
century b.c. until the late third century a.d. Given posits have been determined, so further analysis to
the small number of samples from each monument, determine the date of emplacement for the pumice
the presence of locally quarried material cannot be used at the Colosseum could be used to cross-check
ruled out, but it remains to be found through further these preliminary results.
224
P1: KNP
APPENDIX 4
THRUST LINE ANALYSIS
the funicular polygon method of thrust line unit weight of the material used. The weight (W)
analysis became a standard way of analyzing arches in kg must then be translated into units of force in
in the nineteenth century when the introduction of Newtons (N) by multiplying by 9.8 m/sec/sec.
graphical methods made it more user-friendly than These are the force vectors, each of which is re-
the mathematical methods with which the principles ferred to by the number of its corresponding sec-
were developed. I use it in Chapter 8 to test the effi- tion, for example, F1, F2, F3, and so on.
cacy of various techniques discussed in this study. In 3b. To draw the force vectors, choose a convenient
what follows, I provide an example of how to perform scale for the vectors so that each unit in the draw-
a simple thrust line analysis for a barrel vault. It can ing equals a certain number of Newtons (e.g.,
be done with a pencil and paper, but I use AutoCad scale above Drawing 2). Draw each calculated
for greater accuracy and ease. force vector as point load located at the cen-
ter of gravity of its section. (A program such as
AutoCad can calculate the center of gravity of
part 1 unsymmetrical shapes automatically.)
steps for constructing the thrust line 4. Out to the side of the drawing of the vault
through a barrel vault (fig. 145): (Drawing 1), draw the force vectors end to end
1. Draw a scaled profile of the vault to be analyzed (Drawing 2), one above the other at the same
(shaded area in Drawing 1). scale as represented on Drawing 1. Then pick an
2. Divide the vault into an odd number of vertical arbitrary point (the trial pole-O ) and connect
sections (the more sections the more accurate the the ends of the vectors labeled a, b, c, and so
final curve) so that the middle section is centered on, to O so that you have a series of radiating
on the crown of the vault. Number each section lines: aO , bO , and so on (the thin solid lines on
starting with “1” at the far left. Drawing 2).
3a. Determine the unit weight of the material used 5. Below Drawing 1, project the line of action of
for each section in kg/m3 . Calculate the mass (M) each force down to a base line (Drawing 3). Then
of each section in m3 and multiply it times the starting at the left where the line of action of
225
P1: KNP
APPENDIX 4. THRUST LINE ANALYSIS
145. Drawing demonstrating the steps in making funicular polygon analysis of a barrel vault.
226
P1: KNP
145 (continued )
227
P1: KNP
F1 intersects the baseline at point x , draw a line provide the pole point O. The horizontal dis-
parallel to line bO (from Drawing 2) until it in- tance from O to the vertical line of vectors (aj)
tersects the line of action of F2 (ignore line aO represents the horizontal thrust acting on the arch
for now). At this point, draw the next line paral- or vault.
lel to line cO until it intersects the line of action 9. On Drawing 2, draw lines connecting the pole
of F3. Continue until the diagram looks similar point O to the ends of the vectors creating lines
to Drawing 3. Then connect points x , y , and z aO, bO, cO, and so on (thick dash–double dot
to form a triangle x y z . These points represent lines).
the intersections of the curve just drawn and the 10. The final step is to plot the line of thrust for the
lines of action of the middle (F5) and the two vault. On Drawing 3, draw a line parallel to line
outermost (F1, F9) force vectors. aO from Drawing 2 starting at point x (draw
6. Go back to Drawing 2 and draw lines parallel down and to the left). This is the vector repre-
to lines x y , y z , and x z (represented as single senting the thrust on the abutment. Go back to
dash-dot lines on both Drawings 2 and 3) so that point x and draw a line parallel to line bO from
they pass through the trial pole O . Drawing 2 until it intersects the line of action of
7. Return to Drawing 1 and determine the three F2. Then draw a line parallel to line cO until it
points (x, y, and z) through which the thrust curve intersects the line of action of F3. This process
should pass. All three points must be located on is a repeat of that in Step 5. The curve (thrust
the line of action of one of the force vectors. To line 1) can then be transferred to Drawing 1 to
determine the minimum thrust (i.e., the steepest see how it relates to the vault. (The curve can
possible curve that will fit within the arch) for a actually be drawn directly onto Drawing 1, but it
barrel vault, one point (y) should be at the crown is easier to see on Drawing 2.) If the final curve
of the extrados. The other two points (x and z) is not tangent to the intrados of the arch, points
should be on the lines of action of the outermost x and z must be moved up or down along their
(F1, F9) force vectors. These locations can be es- respective lines of action. Moving the points up
timated to get an idea of the path of the thrust line will spread the curve out making it less steep,
and then adjusted later. (In Drawing 1, I chose and moving them down will bring the curve in
to start with the intersections at the impost of making it steeper. In the example illustrated, the
the vault.) Connect the three points with lines curve needs to come in to touch the intrados, so
forming a triangle xyz. Ultimately, you want to the points are moved down. Connect the new
define the curve that is tangent to the arch in points, r and s, so that they form triangle rys.
three points. These are the three “hinges” of the Then repeat steps 8–10 as shown in Drawing 4
three-hinge arch explained in Chapter 8. to create thrust line 2. Once the curve is tangent
8. Go to Drawing 2 and draw lines parallel to lines to the vault in three points, it is at the state of min-
xy, yz, and xz from Drawing 1 (dotted lines) such imum thrust, which is the limit state indicating the
that they pass through the intersections of the point of stability because of the three hinges. This
vertical line of vectors (aj) and the lines x y , x z , is a trial and error process and may take more than
and y z . The intersection of the lines xy, yz, xz one try!
228
P1: KNP
part 2 3. At the intersection of line rn and the line of ac-

plotting the line of thrust from the vault tion of Fb, mark point t. Using the same scale
down through the abutment: In the previous used for the thrust line of the vault, draw a vector
section, the minimum thrust of a stable three-hinged equal to Fb from t straight down along the line of
arch was established. The final part of determining the action to k, such that tk = Fb. At point k, draw a
stability of the structure is to see whether the line of line parallel and equal in length to line rn to cre-
thrust remains within the thickness of the abutment. ate line ka. Finally, draw a line from t to a. This
Because the drawing is symmetrical, the instructions vector ta represents the magnitude and direction
are given only for the left abutment. of the thrust on the abutment. If the line of ac-
tion of vector ta remains within the thickness of
the abutment, the structure is in equilibrium and
1. Calculate the mass (Mb) of the abutment and will stand; however, a margin of safety is always
multiply it times the unit weight of the mate- desirable.
rial used. Translate this weight into a force (Fb) The process just described provides the steps
into Newtons by multiplying by 9.8 m/sec/sec. involved in creating a funicular polygon diagram,
2. Determine the center of gravity of the abutment. but it does not explain the reasoning behind these
For a rectangular abutment of the same material steps. For further explanation of the funicular
throughout, it will be located on the vertical cen- polygon method, see W. Zalewski and E. Allen,
terline. This is the line of action of the force Fb. Shaping Structures (1998).
229
P1: KNP
230
P1: NDZ
0521842026end.xml CB885B/Lancaster 0 521 84202 6 April 18, 2005 12:8
NOTES
chapter 1. introduction opus spicatum, cocciopesto, and mosaic, so I would suggest that
they are simply pieces of the extrados of a vault similar to
1. MacDonald 1982: 25–31. Figure 43 (see also Appendix 2e).
2. Lamprecht 1987: 46–65; Malinowski 1979. 19. Lugli 1957: 664; MacDonald 1982: 161 n. 47.
3. Blake 1947; Lugli 1957; Rakob 1976: 366–86; MacDonald 20. Van Nice 1965; Mainstone 1988; Mark and Çakmak 1992.
1982: 3–19. 21. Barrel vaulted Republican structures at Ferentino: Gullini
4. Coarelli 1977: 1–23. 1954: 187–9, tav. 53; Annular barrel vault at the Sanc-
5. Blake 1947: 235; Lugli 1957: 417; Boethius 1987: 170. tuary of Fortuna at Palestrina: Fasolo and Gullini 1953:
6. Fasolo and Gullini 1953: 301–23; Lugli 1954: 51–87; Gullini 8, fig. 339; Segmental dome at the Horti Sallustiani:
1973: 760–79. Lehmann-Hartleben and Lindros 1935: tav. 5, 8.3. Repub-
7. Degrassi 1969: 126. lican barrel vault at Hadrian’s Villa: Manieri Elia 1999:
8. Tivoli: Reggiani 1998: 41; Sulmona: van Wonterghem 1976: 175–81.
241; Terracina: Lugli 1926: col. 156. For an evocative recon- 22. Mainstone 1975: 118; Mark and Hutchinson 1986: 32–3.
struction of the Sanctuary of Hercules Victor at Tivoli, see 23. Fasolo and Gullini 1953: 252–4.
Mari and Fiore 2005: 32–41. 24. When these architraves were excavated they were supported
9. Gatti 1934: 123–49. on fill, which explains how they survived for so long (Fasolo
10. Liv. 41.27.8. and Gullini 1953: fig. 7).
11. Tuck 2000: 175–83. 25. Fasolo and Gullini 1953: 252–4, fig. 264.
12. CIL VI 1314 = ILS 35 = ILLRP 367. 26. Gullini 1973: 784–5.
13. Delbrueck 1907: Taf. 4. 27. For the practice of material estimates before a building project
14. Boethius and Carlgren 1932: 181–208; Gullini 1954: 185–216. commences, see Dio Cassius 60.11.3 on estimating the cost of
15. Pollitt 1986: 230–3. the Claudian harbor at Portus, and Dig. 11.6.7.3 on penalties
16. Cifani 1994: 194; Carandini et al. 1986: 436. of fraud involving estimations of materials.
17. Middleton 1892: 66. 28. For measuring sticks, see Ciarallo and De Carolis 1999: Cat.
18. Blake 1959: 163, Ward-Perkins 1981: 101; Adam 1994: 177. 300, 310, 380; Donati 1998: Cat. 99.
Salza Prina Ricotti 1988: 15–26; 2001: 231–40 has argued that 29. Heron, Stereometrica II, 28–40.
some large slabs of concrete found at the Edificio a Tre Esedre 30. Downey 1948: 113.
at Hadrian’s Villa were used as flat vaults covering a span of 31. Archimedes: Plut., Vit. Marc. For a brief overview of the di-
9 m. According to the calculations presented, the slabs (45–70 chotomy between “theoretical” and “practical” mathematics,
cm thick) would have undergone tensile stresses in the range see Cuomo 2000: 91–4.
of 6.27–13.2 kg /cm3 in the most tenable of the reconstruc- 32. Heron, Mechanics I, 24–31; Heath 1921: 350–1. Heron
tions proposed, but even this would be extreme for modern is sometimes accused of being mathematically inaccurate:
concrete, which has a maximum tensile strength of 15–25 Warren 1976: 5, but in the context of explaining how to es-
kg /cm3 (see Chapter 8 n. 21). The slabs are composed of timate materials, there need not be a direct correspondence
the same sequence of layers typically used for waterproofing: between calculable geometrical forms and the final result.
231
P1: NDZ
NOTES TO PAGES 12–20
The results were estimates rather than mathematical proofs. 66. Robinson 1992: 54 believes that it functioned primarily for
See also Tannery 1883: 347–69. maintenance rather than new construction, whereas Martin
33. Frank 1924; Blake 1947; Lugli 1957. 1989: 62 suggests that it was in charge of organizing builders,
34. DeLaine 1995: 555–62. laborers, and supplies. Eck (1992: 242; 1993: 391) believes
35. DeLaine 1997: 219. that other persons outside of the cura operum publicorum were
36. Quilici 1974: 62–78, 105–18, 143–69. appointed to oversee new buildings. Kolb 1993: 53–7 argues
37. Strabo 5.3.11. that the office was probably responsible for new projects but
38. Frank 1924: 30–1. admits that there is little direct evidence to prove this. Bruun
39. Ashby 1924: 135–9. 1996: 737 disagrees, pointing to lack of evidence and to the
40. The lapis Gabinus was transported to Collatia (modern fact that the high senatorial officials acting as curatores usually
Lungazza) and then loaded onto river boats (Lugli 1957: 307, only held their position for one year.
Quilici 1986: 210). 67. Frontinus, Aq. 2.119.
41. DeLaine 1995: 556, fig. 2. Weights from Ventriglia 1971: 209, 68. Brunt 1980: 81–100.
219, 221. 69. Martin 1989; Kolb 1993; Daguet-Gagey 1997.
42. Mari 1983: 10–14, 361–70; Quilici 1986: 209. 70. Josephus, BJ 3.10.10 mentions that Vespasian took six thou-
43. DeLaine 1995: 560, fig. 1; DeLaine 1997: 88–9; Quilici 1986: sand Jewish captives to work on the canal at Corinth. This
211. passage is probably the source of the mistaken but often
44. On the construction costs for the Baths of Caracalla, DeLaine quoted idea that the Colosseum was built by means of Jewish
1997: 216 gives the figure of 42 percent for the supply of fuel slave labor. The labor for digging a canal is quite different
for processing the brick and lime. from the skilled labor of carpenters, masons, and bricklayers.
45. Plin., HN 36.176. 71. CIL VI, 9034. The suggestion has been made that the phras-
46. Faventinus 4. ing implies the establishment of a department of imperial
47. Gerding 2002: 48; Blake 1959: 161 notes that the Tomb of works by the late first century a.d. called the Opera Caesaris
Hirtius found under the Cancelleria was faced with broken (Strong 1968: 105; Anderson 1997: 69, 89, 91 n. 83); how-
tiles and can be dated to 43 b.c. ever, a redemptor on the imperial staff makes little sense, as the
48. Blake 1959: 162; Lugli 1957: 546. title itself implies that he took part in some sort of competi-
49. Juvenal, Sat. 6.344; Lanciani 1897: 41–2. tive bidding process and was not a salaried employee: Pearse
50. E.g., the bricks of the praedia Statonensis would have been near 1974: 38–9.
ancient Statonia, the figlinae Subortanae were probably located 72. CIL VI, 607.
near Horta (modern Orte), the figlinae Narnienses near Narnia 73. Haterii tomb: CIL VI, 19148; Evagogus: CIL VI, 9408.
(modern Narni) (Steinby 1978: col. 1509). 74. Roman citizens typically had three names, praenomen (first
51. Graham 2002; Bianchi 2001; Olcese 1993; Olcese 1994. name), nomen (family name), and cognomen (distinguishing
52. Vitr., De arch. 2.9.6; Plin., HN 16.42, 16.200–1, 16.222–3. name), whereas slaves only had one name. When a slave was
53. Etruria and Campania: Vitr., De arch. 2.10.1–2; Plin., HN freed he took the praenomen and nomen of his master and
16.191; Pisa: Strab. 5.2.5. added his slave name as the cognomen. For the obligations of
54. Theophr., Hist. pl. 5.8; Plin., HN 3.74; Dion. Hal. 20.15; freedman to former master, see Treggiari 1969: 68–81. For
Meiggs 1982: 462–6. a study of relationships between former slaves and masters in
55. Sidonius 5.441–5. the building industry at Ostia, see DeLaine 2003: 723–32.
56. Meiggs 1982: 255. 75. D’Arms 1981.
57. Quilici 1986: 205–7; Chevallier 1976: 197–201. 76. Martin 1989: 103–13.
58. Plin. HN 3.53–4. Some remains of this lock system have 77. Dig. 19.2.51.1.
been found, but damage during World War II and subsequent 78. Rea et al. 2002: 370–4.
reconstruction has covered much of it, so any indication of 79. Steinby 1993: 139–43.
use over time has been lost (Quilici 1986: 215–17). 80. Helen 1975: 108–9.
59. Strab. 5.2.5, 5.3.7. 81. Brunt 1980: 81–100; Garnsey 1981: 359–71; DeLaine 1997:
60. DeLaine 1997: 211. Hopkins 1983: 104 gives ratios of 220–4.
sea:river:land of 1:6:55. 82. Waltzing 1896: 255–9, 397–408; de Robertis 1938: 416–17.
61. Dion. Hal. 20.15.2. See also Theophr., Hist. pl. 5.8.1. See also Alföldy 2001: 8.
62. Fontana 1995: 563–70; Quilici 1986: 205. 83. More 1971: 203–5.
63. Anderson 1997: 79–88. 84. CIL VI, 1060, 9405, 10300; DeLaine 2000: 121–2, Anderson
64. Suet. Aug. 37. 1997: 115–16.
65. Gordon 1952: 280–3. 85. Pearse 1974: 124–5.
232
P1: NDZ
86. Pearse 1974: 69. 11. Krautheimer et al. 1980: 104–5.

87. DeLaine 2000: 132–6; DeLaine 2003: 727–31. 12. Ucelli 1950: fig. 150, tav. 5. The beams were approximately
88. Martin 1989: 49–50, 103–13. 35 cm deep with 125 cm long overlap.
89. Dig. 19.2.24 pr. Roman jurists do not name the architect as 13. Curle 1911: 111–13, pl. lxvii, 5.
one typically performing probatio, though there is evidence 14. Mercer 1929: 272, fig. 227.
that this was the case in a fourth-century-b.c. Greek con- 15. Heron, Mechanica 3.5; Drachmann 1963: 102. For ropes, see
tract from Delos, in which the architect is named along with O’Connor 1993: 48; Lancaster 1999: 435.
the contracting agents as responsible for the final inspection 16. Healy 1978: 233.
(Burford 1969: 98). 17. Strabo 5.1.8; Horace, Odes 1.16.9, Epodes 17.71.
90. Payment of architect: Martin 1989: 49; Incompetence: Dig. 18. Maddin et al. 1991: 5–23.
11.6.1.1; 11.6.7.3. 19. Goodman 1964: 118; Gaitzsch 1980: 184.
91. Jones 1964: 61–6, 708, 839–40. 20. The pit-saw depicted from the Bottega del falagname at
92. Barnes 1982: 218–19. Pompeii dates from the first century a.d., and a relief de-
93. Chastagnol 1960: 27–30, 348–9; Ward-Perkins 1984: 38–48. piction of a frame saw on an altar found near the Forum
94. Ward-Perkins 1984: 14–15. Holitorium is roughly dated to the same period (Colini 1947:
95. Lactantius, De mort. pers. 7.8–10; Williams 1985: 134–5. Jones 25–6, fig. 1).
1964: 838–40, 858. 21. Plin. HN 16.227.
96. MacMullen 1964: 49–53 argues that these requirements often 22. Gaitzsch 1980: Taf. 60 No. 295.
were ignored. 23. Fitchen 1961: figs. 2, 13, 21. An example of the process of
97. Cod. Theod. 14.8.1; Waltzing 1896: 259–67, 360–1, 378–92; removing centering is shown in the film, Roman Bath (2000),
de Robertis 1938: 424; Williams 1985: 134–5. which is part of the NOVA series, Secrets of Lost Empires. See
98. The political aspect of the fourth category is my own addition also, Yegül 2003: 164–6.
to the four categories proposed by Schlebeker 1977: 641–55 24. Waddell and Dobrowolski 1993: 25.16.
and used by White 1984: 21 in his study of Greek and Roman 25. Harrison 1995: 21.
technology. 26. Alberti, On the Art of Building in 10 Books, 3.14. Fitchen 1961:
32–4, fig. 12 also discusses some other less common tech-
niques.
chapter 2. centering and formwork 27. O’Connor 1993: 174.
28. Fitchen 1961: 33, fig. 12.
1. In many cases in which apparently similar arches or vaults are 29. Cozza 1986: 113 n. 24. The tower is located along the Lun-
constructed in a series, the length and width of each bay var- gotevere della Farnesina and is on the property of the sisters
ied, so each one had to be constructed to a slightly different “Figlie di S. Giuseppe.” The plank, which is visible just above
form. For examples at Trajan’s Markets, see Lancaster 2000: the impost of the vault in Cozza 1986: fig. 15, is 0.15 m thick,
773. 0.05 m wide, and 1.46 m long. I am grateful to Lucos Cozza
2. Saalman 1959: 151–4. for bringing this example to my attention and for accompa-
3. For a photo of the piece, see Coarelli 2001: fig. 3, who dates nying me to the tower to obtain a sample for analysis. I thank
it to the Augustan period. Rodriguez-Almeida 1994: 215–17 Ian Gourlay of the Oxford Institute of Forestry and Plant
suggests a somewhat later date between the first and second Sciences for examining the wood and making the identifica-
centuries based on technical aspects and on the molding type. tion.
4. Adam 1994: 209–12; Coulton 1977: 157–9. 30. Society 1986: 66.
5. Vitr., De arch. 4.2.1, 5.1.9. 31. The use of reed mats has been cited for the Roman period,
6. Dio Cassius (58.13) gives its dimensions, and using the infor- but in many cases these examples can be shown to be later,
mation from both sources, O’Connor 1993: 142–4 recon- such as the impressions cited by Middleton 1892: 67 of a vault
structs the bridge with twenty bays, each with a clear span of under the Basilica of Maxentius, which recent investigation
approximately 32 m (110 feet). show is in fact much later. I thank C. M. Amici for this
7. Plin., HN 16.202, 36.102. information. Other examples that are taken to be Roman
8. San Pietro: Fontana 1694: 99; Krautheimer et al. 1980: are cited by Adam 1994: fig. 434; Blake 1947: 346 n. 41;
fig. 198. San Paolo Fuori le Mura: Krautheimer et al. 1980: Storz 1994a: Taf. 31.3.
fig. 140. For a discussion of the types of trusses used in other 32. Mango 1992: 48.
early churches in Rome, see Valeriani 2003: 2023–34. 33. Lugli 1957: 681 reported an early example of the use of brick
9. Rival 1991: 162–5, pls. 34, 37. linings at the Domus Aurea, but he was probably referring to
10. Dobson 1849: 116–18. Sette Sale when it was still considered to be part of the Domus
233
P1: NDZ
Aurea (a.d. 64–68). The attribution was then taken up by modified to a horizontal bedding so that wedges could be
later scholars (Ward-Perkins 1981: 89, 100 n. 5; Sear 1982: used.
80), but, in fact, there are no brick linings at the Esquiline 53. Rasch 1991: 359–60.
Wing. At the Villa of Domitian at Castelgandolfo, the linings 54. De Angelis d’Ossat 1938 (pub. 1940): 241.
are used throughout the vaults of an unexcavated bath build- 55. Rakob 1988: 280, n. 49; Rasch 1991: 365, 369–70; Leacroft
ing belonging to the “palazzo imperiale,” but the module of and Leacroft 1969: fig. 12.13; Taylor 2003: 195–208; Viollet-
the brickwork (25 cm) is much smaller than the typical Fla- le-Duc 1868: 473; Heene 2004.
vian brickwork (28 cm), suggesting that they probably date 56. Rasch 1991: 365, 369–70; Rasch 1992: 20–1. He follows
later than the Flavian period. Heilmeyer 1975: 316–47 in accepting that Apollodorus ac-
34. DeLaine 1997: 166. tually designed the Pantheon, although there is no direct
35. For general discussions on this technique, see Choisy 1873: evidence for this.
60–7; Cozzo 1928: 179–82; Giuliani 1990: 96; MacDonald 57. Vitr., De arch. 10.13.5.
1982: 159; Lugli 1957: 668–9, 681, pl. CCVI, 2; Rivoira 58. Viollet-le-Duc 1868: 473.
1925: 93; Giovannoni 1925: 38. 59. Taylor 2003: 195–208.
36. Choisy 1873: 60–7. Followed by Rivoira 1925: 93, Giovan- 60. Heron, Mechanica 3.5.
noni 1925: 38, and MacDonald 1982: 159. 61. For elevators, see Carettoni 1956–1959: 23–44; Beste 2001:
37. Choisy 1873: 70–1. See further, Lancaster 1998a: 300. 294–7. Taylor 2003: 195 argues against the use of the central
38. Cozzo 1928: 181–2. Followed by Lugli 1957: pl. CCVI, 2 tower and points to a number of problems that he sees in its
and Blake 1959: 94, 163. use; however, I would argue that a solution for which there is
39. Vitr., De arch. 7.4.2–3. See also Plin., HN 36.177. documented evidence is much preferable to the difficulties in
40. See Rasch 1991: n. 169 for the reuse of formwork. organizing the logistics of simultaneously operating twenty-
41. Giuliani 1975: 329–42. Similar type bars have been found eight cranes.
for the suspended ceilings at the Baths of Caracalla (DeLaine 62. The imprints of the formwork coffer boxes rarely sur-
1987: 150–3). vive, but some imprints remaining from the cryptoporti-
42. Van Deman 1924: 387, 391 indicates that the formwork at cus at the Villa of Domitian at Castelgandolfo show that
the Domus Tiberiana was covered with a thin layer of mor- the flat panel of the inner coffer box consisted of side-by-
tar with unusually high lime content. This was probably inside boards laid horizontally. I thank Henner von Hesberg
tended to produce a dense and smooth outer surface. for pointing this out to me during a visit to the site in
43. Guidobaldi 1992: 55–66. March 2002.
44. de Fine Licht 1990: 31–2, fig. 32. 63. Martines 1991: 3–10.
45. Dobson 1849: 34. 64. Wilson Jones 2000: 183.
46. Fitchen 1961: figs. 54, 55, 65, and 71. 65. Taylor 2003: 204 argues that the number of coffers was orig-
47. DeLaine 1997: 166–8, fig. 84; Taylor 2003: 186–90, fig. 108. inally intended to be thirty-two (a multiple of sixteen) but
48. On crane capacities, see O’Connor 1993: 49; Meighörmer- that they were limited to twenty-eight because of the num-
Schardt 1990: 59; Lancaster 1999: 426, 436. ber of cranes that had to be used in his reconstruction of how
49. Rakob 1988: Taf. 116.1. the frames were lifted.
50. The imprints of the radial formwork are best preserved in 66. Plut., Vit. Marc.; Cic., Tusc. 5.64–66.
the Mausoleum of the Tor de’Schiavi, but the profiles of 67. Dio Cass. 69.4; Brown 1964: 57–8. The word used,
other domes suggest that radial formwork was not uncom- λ , has been variously translated as “pumpkin,”
mon (Rasch 1991: 363, fig. 5). “gourd,” and even “nasturtium”(?) (MacDonald 1982: 135,
51. The date has been placed by various authors anywhere be- n. 45).
tween the Augustan and the Flavian periods. Rakob 1988: 68. Horti Sallustiani: Lehmann-Hartleben and Lindros 1935:
290–1 settles on an Augustan date. 198–9, 210–11; Otricoli: Pietrangeli 1978: 64–75; “Temple
52. Rakob 1988: 280–3, n. 49. He notes that the inner proof Venus” at Baiae: Rakob 1961: 114–49; room of bath build-
file of the dome is a flattened curve composed of the two ing at “Villa of the Gordians”: Lugli 1915: 153, 158–60, Luschi
different arcs and proposes that their intersection marks the 1989–1990: 434–41.
point at which there was a change in the centering forms. In 69. For a list of examples surviving as well as recorded in draw-
the revised version that I propose in Figure 33 (p. 41), I have ings, see Rakob 1961: 140–3. The dome of the “Plane-
adopted a similar scheme but have placed the change between tarium” at the Baths of Diocletian does not actually have
the frames at the base of the windows in the dome, which groins and would have been built on normal domical
were not shown in his proposal. He also showed a vertical centering.
connection between the radiating frames, which I have 70. Rasch 1991: Abb. 7, 20, 25, 29, 31.
234
P1: NDZ
71. Rasch 1991: 370–9; Rasch 1985: 137. See also Mainstone 21. The association with Puteoli is first made by Seneca (QNat
1986: 673–4. 3.20.3.) in the mid-first century a.d., when he uses the term
72. Vitr., De arch. 5.10.3–5. puteolanus pulvis.
73. On pitched brick vaulting in the Honorian towers in the 22. Gazda 1987: 76 n. 5. Gazda and McCann 1987: 76 n. 5, 145
Aurelian Walls, see Cozza 1987: 43, fig. 51; on the tubi fittili n. 44, 337; Gazda 2001: 163 suggests that the date could be
at San Stefano Rotondo, see Storz 1994b: 669–93. moved to the second quarter of the first century b.c. Further
74. MacDonald and Pinto 1995: 157, figs. 192–196. Oddly, the study on samples of hydraulic mortars from Mediterranean
authors continue to refer to the decoration as a shell motif port facilities is being conducted by the Roman Maritime
even after describing the decoration as ending in tassels and Concrete Study (ROMACONS) directed by J. P. Oleson,
suggesting a large tent. R. L. Hohlfelder, and C. Brandon and may help clarify ques-
75. Athenaeus, 5.196b–c; Hemsoll 1990: 21–4, 31–3; Lehmann tions on exportation of pulvis puteolanus in the future.
1945: 1–27. 23. Vitr., De arch. 2.4.2–3.
76. For mosaics, see MacDonald and Pinto 1995: 114; Salza Prina 24. Pliny, HN 16.202.
Ricotti 2001: 241. 25. Scrinari 1963: 534–5; Scrinari 1979: 9 notes that in the ex-
77. Salza Prina Ricotti 2001: 420–1, 255; Salza Prina Ricotti cavations of the port of Claudius, the concrete used to form
1988–1989: 224–5. the substructures for the lighthouse were made of pozzolana
and tuff from Campania. She also points out that the
pozzolana in the later additions to the port is different
implying that it is local, which would suggest that pulvis
chapter 3. ingredients:
puteolanus was not always used for harbor construction
mortar and caementa during the mid-first century a.d. On recent investigations
of the concrete at Portus, see also Oleson et al. 2004: 199-
1. Ferretti 1997: 70. 229; http: //web.uvic.ca /∼jpoleson /Harbour%20Concrete/
2. Lea 1970: 9. ROMACONS02.html.
3. Lea 1970: 433. 26. Vitruvius (De arch. 2.4.2–3) does not mention harena fossica
4. Lea 1970: 307 fig. 80; Harrison 1995: fig. 4. as having hydraulic properties whereas he (De arch. 2.6.1)
5. Vitr., De arch. 2.6.4, (translation from Rowland and Howe associates this characteristic specifically with the “pulvis” from
1999: 38). the Bay of Naples.
6. Mainstone 1975: 53. Lechtman and Hobbs 1987: 98 discuss 27. Pliny (HN 36.175) recommended using less lime than did
the chemical and morphological transformations that occur Vitruvius (De arch. 5.12.8–9), who was writing about a cen-
during the hydration process. tury earlier.
7. Lechtman and Hobbs 1987: 96–9. 28. Brandon 1999: 169–78.
8. Lea 1970: 435. 29. Blake 1947: 317.
9. Smith 1981: 22. 30. Van Deman 1912: 251 gave the earliest examples as the The-
10. Vitr., De arch. 5.12.3–4. ater of Pompey, the Curia Julia, and the Forum of Caesar.
11. Frontin., Aq. 2.123. For the use of “structura” to refer to mor- 31. Coarelli 1981: 24.
tar work, see Vitr., De arch. 2.8.1. On the Baths of Trajan, see 32. Blake 1947: 317 and Van Deman 1912: 230–51.
Volpe 2002: 388–90, fig. 10. By the fifteen century, builders 33. Ventriglia 1971: 31; Chiari et al. 1996: 6.
certainly understood that water was good for vaults. Alberti 34. Vitr., De arch. 2.4.1. This also may imply that it should have
(3.14) noted that “it must be said that nothing is better for been washed before being tested with the cloth.
a vault than for it to have plenty of water and to never go 35. For the sieving of pozzolana in mortar, see Blake 1947: 314;
thirsty.” (translation from Rykwert et al. 1988: 87). Davey 1974: 195; DeLaine 1997: 110 n. 35.
12. Dio Cass. 48.51. 36. Blake 1947: 314, 316.
13. Bazant and Wittmann 1982; Nilson and Darwin 1997: 37. Van Deman 1912: 415 n. 4, 8.
33–52. 38. Alberti 2.12.
14. Mainstone 1988: 87, 166, fig. 3. 39. Healy 1978: 144; Diod. Sic. 5.27.2; Strabo 3.2.10.
15. Burnell 1850: 47–8. 40. DeLaine 1997: 110 n. 35, 40.
16. Boynton 1980: 164–5. 41. Beltrami 1892: 88–9.
17. Alberti 2.11 (translation from Rykwert et al. 1988: 55). 42. Van Deman 1912: 423.
18. Plin. HN 36.176. 43. Heres 1983: 98–111.
19. Scavizzi 1983: 29–30. 44. Chiari et al. 1996: 7–14.
20. Spinazzola 1953: 446 n. 350. 45. Guyon 1986: 306–8.
235
P1: NDZ
46. Chiari et al. 1996: fig. 2. closest matches are in the pumices produced by the volcanic
47. For example, the mixture of crushed terracotta and pozzolana district around Lago di Bolsena, but further research into the
for waterproofing of floor or walls occurs at the Mausoleum details of the material is required to make a firm verification
of Helena (Chiari et al. 1996: 13), the Basilica Argentaria in of its provenance.
the Forum of Caesar, and the Colosseum. 62. de Fine Licht 1974: 21. The physical characteristics were
48. Giuliani 1990: 171–2; Giuliani 1992: 89–94. not described, but it was presumably the same type found in
49. Chiari et al. 1992: 127. At Trajan’s Markets, they were able to section E.
determine that the terracotta in the modern cocciopesto used 63. Lancaster 2000: 776.
for restoration was from crushed pottery, because the process 64. Steinby 1986: 106.
of throwing the pots resulted in an alignment of crystals that 65. Meneghini 1989: 541–57.
one does not find in bricks or tiles. A more systematic analysis 66. They occur in a barrel vault (c. 8.10-m span), a cross vault
of this type for ancient cocciopesto could yield useful results on (c. 8.75-m span), and a semidome (c. 12-m dia). This section
the reuse of material. also contains examples of tufo rosso a scorie nere mixed in with
50. Cato, Agr. 18.7 advised that broken terracotta be combined the tufo giallo della via Tiberina.
with lime to form the floor of a press. Blake 1947: 322–3 67. De Angelis d’Ossat 1930: 211–15 determined that the sco-
cites a mid-second-century b.c. example of cocciopesto from ria was from Vesuvius and the yellow tuff from the Sabatini
Tiber Island. system.
51. Vitr., De arch. 7.1.3–4. For a reconstruction drawing, see 68. de Fine Licht 1968: fig. 99. The vaults of the South Building
Rowland and Howe 1999: 269 fig. 100. consist of alternating rows of yellow tuff and brick.
52. Vitr., De arch. 7.1.4. 69. Cozza 1982: fig. 25a. I thank Lucos Cozza for bringing this
53. Pantheon: Lugli 1938: 142; Trajan’s Markets: Ungaro 1995: example to my attention.
135; Palatine: Carettoni 1971: 313 n. 1. 70. DeLaine 1985: 200; DeLaine 1997: 87, 159.
54. Krause 1994: 44–8, 77–8, 96–7; Krause 1986: 444–5. Two 71. Canina 1860: 187. Lanciani 1897: 277 reports pieces of fallen
of the bipedales bear Neronian brick stamps. vaults with stucco. Fredenheim found pieces of vaulting with
55. Palatine: Krause 1994: 78; Krause 1986: figs. 145–147; coffers and stucco in 1780–1789 (LTUR 1: 178).
Carettoni 1971: 321–3, Carettoni 1949: 58–9. Trajan’s Baths: 72. White in tepidarium dome (De Angelis d’Ossat 1946: 21–
Fabbrini 1982: 15, figs. 21–23. She does not indicate another 2); black and white in webs of most cross vaults (Rivoira
layer of cocciopesto on top of the bipedales. Trajanic brick stamps 1925: 206). Paulin 1890: 13 is vague noting only that pieces
were found on the bipedales: CIL XV, 1390 and 1096a. of very light volcanic stone were used in the vaults. Caraffa
56. Tomb of Caecilia Metalla: Gerding 2002: 46; Basilica 1942: fig. 5 points to the presence of some unspecified
Aemilia: Blake 1947: 344. Neronian structures on Velia: Blake type of pumice in the upper part of the “Planetarium”
1959: 47. Horrea Agrippiana: Blake 1947: 344 cites the yellow dome.
tuff in the vaults, but I believe that these vaults belong to 73. Candilio 1985: 528. The caementa in the east palaestra is still
the reconstruction associated with the project of the Domi- visible, and dottoressa Candilio has kindly informed me that
tianic Vestibule, whereas Astolfi et al. 1978: 56–7 argue that the pumice excavated in the west palaestra also was grayish
the original Agrippan vaults were left in place and that the in color.
Domitianic wall was built up against them. Domus Aurea: 74. Coccia and Fabiani 2003: 38, fig. 8. I thank C. M. Amici for
Van Deman 1912: 404 reports “large pieces of yellowish gray the information on the cores.
tufa,” but the caementa visible in the vaults of the Esquiline 75. Vitr., De arch. 2.6.2–3.
Wing of the Domus Aurea are pieces of tufo lionato of mod- 76. Ranieri 1998: 135–41; Ranieri and Yokoyama 1997: 33–50;
erate size. Kawamoto and Tatsumi 1992: 92–7.
57. Blake 1947: 349. 77. Strabo 5.4.8.
58. Van Deman 1912: 404. 78. I have not yet found any similar type of scoria that belongs to
59. Amici 1991: 52, 162. For a decorative use of scoria on a tomb the a.d. 79 eruption, although further investigations around
from Ostia (first half of the first century a.d.), see Squarci- Vesuvius will be the subject of a future publication.
apino 1958: 35–8. 79. De Angelis d’Ossat 1938 (pub. 1940): 245.
60. For the use of tufo giallo della via Tiberina at the Colosseum, 80. Cod. Theod. 14.6.3.
see Lancaster 1998b: 148. Van Deman 1912: 408 n. 1 says 81. Quilici 1986: 210–11 argues that some remains along the
that dark pumice stone was used in the reconstruction, but Fosso di Grotta Oscura belong to artificial secondary tribu-
she mistook weathered tufo lionato for brown scoria. taries that were created to transport the material down to the
61. The pumice was found to have much sanidine, which is a Tiber.
characteristic of the volcanic districts north of Rome. The 82. Pagano 1995–1996 (pub. 1998): 35–9, fig. 1.
236
P1: NDZ
83. Foss 1988: 129. The plant life returning on the pyroclastic 15. Vitr., De arch. 5.5.1–5; For further discussion of the unlikeli-
flows of Mt. St. Helen’s twenty years after the eruption of hood of the amphoras as sound resonators, see Amici 1991a:
1980 shows that only small scrub growth has returned to the 46–7.
area once covered by old growth forests. 16. Ousterhout 1999: 229–30.
84. Cantasini et al. 2003 (http: //www.ims.demokritos.gr / 17. Whether it was actually a very grand house or some type of
archae /Abstracts.html). Rocks and sand were sometimes semipublic structure has been debated. The construction tec-
used as ballast for ships and therefore provided some added hniques used, however, are much more representative of the
incentive to export them (Parker 1992: 28; Bruni 2000: 42– public structures, as no other house employed concrete vaults.
3; Giachi and Pallecchi 2000: 350–1). Pumice (unspecified 18. Remesal Rodrı́guez 1998: fig. 1.
provenance and weight) has even been found used as a pack- 19. Roldán Gómez 1988: 121–2, Roldán Gómez 1993: 190–5.
ing material between amphoras in a shipwreck from Albenga 20. I thank Ted Peña for pointing out this use of amphoras.
dating to 100–80 b.c. (Parker 1992: 50). 21. Laubenheimer 1998: 47.
85. Casson 1965: 31–9. 22. Pesavento Mattioli 1998.
86. Williams 1985: 119. 23. Quilici Gigli 1987: 159–61.
87. Cod. Theod. 14.6.3, 15.1.19. 24. Hesnard 1980: 141–56; Boersma et al. 1986: 96–9.
25. Dressel 1879: 192–5.
26. Promis 1862: 141 attributes these walls to the Augustan pe-
chapter 4. amphoras in vaults riod and possibly to a basilica, but his dating criteria are vague.
Mezzena 1981: 81–4 notes that more recent excavations have
1. For further discussion of the vaulting tubes, see Wilson 1990: revealed a bath building nearby and that Promis’s foundations
97–129; Storz 1994a. may belong to a surrounding wall of the bath complex. Strati-
2. E.g, Lugli 1957: 689–90; Rivoira 1925: 178; Ward-Perkins graphical excavations place all of the early structures in the
1981: 436; Adam 1994: 183, fig. 441. area in the first century a.d.
3. Choisy 1873: 96–8. 27. According to Jashemski 1979: 188–9, figs. 279, 280, 282, the
4. Rasch 1998: 22 after Deichmann and Tschira 1957: 61; Amici reports of the excavation of this site were never published, and
1991: 46–7. the vessel type and possible contents has not been identified
5. Monneret de Villard 1924: 149–54. to my knowledge.
6. Alberti 3.14 (translation from Rykwert et al. 1988: 86). 28. Rasch 1998: 22 after Deichmann and Tschira 1957: 61, Amici
7. Peacock and Williams 1986: 133. 1991a: 46–7; Cozza 1987: 42.
8. Coccia and Fabiani 2003: 38, n. 17. 29. Dio Cass. 48.51.
9. Peacock and Williams 1986: 154; Packer 1968–1969: tav. LIX, 30. Alberti 3.14 (translation from Rykwert et al. 1988: 87).
2. 31. Bost et al. 1992: 118: Dressel 20 (thirty-four examples), Dres-
10. I thank Ted Peña for discussing with me his current work in sel 23 (sixteen examples) and Almagro 51c (sixteen exam-
progress on the history of the reuse of amphoras. ples).
11. On the reuse of wine amphoras, see Peña 1999: 43 n. 88, 32. Rodrı́guez-Almeida 1999: 244–5, tav. LVI.
177–8. 33. Rodrı́guez-Almeida 1999: 244–5.
12. Rodrı́guez-Almeida 1999: 239–40. 34. This is a high estimate based on the drawings in Rodrı́guez-
13. The laconicum of the Stabian Baths at Pompeii (first century Almeida 1999: tav. LII and LVI.
b.c.) is often wrongly cited as the earliest example of the 35. These calculations are made assuming the following: weight
use of amphoras in vaults, but this mistaken attribution stems of empty Dressel 23 = 62.4 kg; volume displaced by Dressel
from the confusion between two techniques. According to 23 = 0.0453 m3 ; weight of tufo giallo = 1,400 kg/m3 ; weight
Lugli 1957: 671 and Rivoira 1925: 38, the technique used of tufo giallo and scoria = 1,200 kg/m3 ; concrete mixture
at the Stabian Baths is that of building vaults with rows of consists of 45 percent mortar and 65 percent caementa. Pro-
interlocking terracotta tubes and pottery jars; however, Lugli portion of lime to ballast in mortar is 1:2 resulting in concrete
later equates the two techniques causing further confusion. mixture of 15 percent lime. Assuming a 25 percent loss of
14. The Magazzini “Traianei” (I.20.1) are only partially exca- lime volume during curing gives 19 percent of total volume
vated and remain unpublished other than a brief mention needed for lime (DeLaine 1997: 123–4); one single-yoke ox-
in Calza 1953: 138, 219, although a new study is underway cart can carry 400 kg (DeLaine 1997: 108).
(Bartolini and Turchetti 1999: 26–8). The main publications 36. I ignore the Dressel 20 amphoras for ease of calculation.
of the Villa alla Vignaccia are Ashby 1907: 74–8 and Ashby 37. These figures are calculated on the assumption that one cart
and Lugli 1928: 157–92. Both monuments are dated by brick drawn by two oxen or mules could carry 400 kg and that the
stamps. concrete consists of 45 percent mortar, with a proportion
237
P1: NDZ
of lime to ballast of 1:2 and 25 percent lime loss during 59. Peña 1999: 25.
processing: DeLaine 1997: 108, 123–4. The assumed masses 60. Symmachus, Relat. 35. See further, Peña 1999: 25–6.
are as follows: tuff = 1,600 kg/m3 ; pozzolana = 1,750 kg/m3 ; 61. If the Dressel 20 amphoras had been destroyed because of
hydrated lime = 480 kg/m3 . the tituli picti on them, there was then no official reason to
38. Choisy 1873: 97, pl. XI. He argues that its placement in the destroy the Dressel 23 amphoras, which did not have them.
wall shows that it was not intended to lighten the structure, With the incentive to destroy the bulbous Baetican amphoras
but I would suggest that in this case its placement over a gone, there would have been more intact vessels of this type
window should be seen in light of Caraffa’s later finds of available.
amphoras in the dome. 62. SHA, Sev. 18.3; Peña 1999: 23–4.
39. De Angelis d’Ossat 1945: 3–6. 63. Chron. a. 354, 148.
40. Caraffa 1944: 19–20 argued that these “windows” were used 64. Peña 1999: 24–8 presents the evidence for both the Sev-
as passages during construction and then closed as the dome eran influence in Tripolitania and the takeover by Domitius
was constructed because they do not go all the way through Alexander.
the thickness of the dome. 65. Cod. Theod. 9.17.2 (a.d. 349), 15.1.14 (a.d. 365), 15.1.19 (a.d.
41. Rasch 1993: 51–2. 376), 15.1.40 (a.d 398). For a discussion of the divergent
42. de’Ficoroni 1744: 173: “Il portico era ricoperto da volta attitudes, both ancient and modern, toward spoliation in late
costrutta di materiale framezzato da olle vacue per renderla antiquity, see Coates-Stephens 2003: 341–58.
leggiera.” 66. Cozza 1987: 42 n. 15, fig. 44.
43. Frazer 1969: 45–8 reconstructed the porch with the columns 67. Mazzucato 1970: 358–9; Clark et al. 1989: 314–15.
supporting a series of arches springing from architrave blocks 68. The technique of using pots in vaults later becomes very
over each column, which would have resulted in a very un- common in Byzantine architecture: Ousterhout 1999: 227–
stable structure. However, Rasch’s reconstruction of the ped- 31.
iment supported on a continuous entablature is based on his 69. Remesal Rodrı́guez 1986; Blázquez Martı́nez 1992: 173–88;
finding of a piece of one of the surviving Proconnesian archi- Carreras Monfort and Funari 1998: figs. 26–32.
trave blocks (Rasch 1993: 38–9). This reconstruction provides 70. Gerkan 1951: 216.
much more stability. 71. Remesal Rodrı́guez 1986: 31. Remesal Rodrı́guez 1991: 359
44. Rasch 1993: 51–2. notes that more than twelve hundred amphoras were used
45. Rasch 1993: 54, 87. in the building, but based on the published drawings and
46. For a nineteenth-century argument against their use as a descriptions of Gerkan 1951: Abb. 4, this seems an unlikely
means of lightening the vault, see Choisy 1873: 96–8. high estimate. I have not had the opportunity to examine the
47. Mazzucato 1970: 348–50, fig. 16. monument in person, but I am grateful to Ute Verstegen at
48. Peña 1999: 22, n. 187. the University of Cologne, who is writing a dissertation on
49. An inscription (CIL II, 1180 = ILS 1403) from a.d. 161–169 St. Jerome, for answering my questions about it.
lists an official dealing with oil as part of the office of the 72. Remesal Rodrı́guez 1986.
praefectus annonae. Another one (CIL VI, 1625b = ILS 134)
dated to approximately a.d. 144–146 reveals ties between oil
merchants from Baetica and a man who had served as praefectus chapter 5. vaulting ribs
annonae, suggesting that Baetican oil was part of the annona
by this time. 1. Bianchi 2000: 105–62 provides a detailed review of the use of
50. Blázquez Martı́nez et al. 1994: 137–42. ladder and lattice ribbing in vaults, but because she consid-
51. Blázquez Martı́nez 1992: 185. ers bipedalis ribbing a different phenomenon, we sometimes
52. Peña 1999: 20–2. come to different conclusions regarding the development of
53. Panella 1983: figs. 49–50. the technique.
54. Mattingly 1988: 55. 2. Colosseum: Lancaster 1998b; Forum of Caesar: Amici 1991;
55. Rodriguez-Almeida 1980: 115–20 presents evidence for a no Trajan’s Markets: Lancaster 2000; Baths of Caracalla: DeLaine
longer existing mound of potsherds near Monte Testaccio 1997; late Roman domes: Rasch 1991.
that he calls “piccolo Testaccio,” but its date is unclear. 3. Blake 1959: 94, 163; Ward-Perkins 1981: 68, 434–5;
56. Rodrı́guez-Almeida 1989: 112 no. 119. MacDonald 1982: 159.
57. Bost et al. 1992: 124 notes the total absence of all tituli picti on 4. De Angelis d’Ossat 1938 (pub. 1940): 239, 247–8; Giovannoni
the cargo of the Cabrera 3 shipwreck, which has a terminus 1925: 39–41; Lugli 1957: 668; Rivoira 1925: 144.
post quem of a.d. 257 based on the coins found. 5. Giuliani 1990: 94–6.
58. Panella 1999: 196–8. 6. Cifani 1994: 194; Carandini et al. 1986: 436.
238
P1: NDZ
7. Adam 1994: 160, fig. 382; Napoli 1965: 205–11; Napoli 1966: 33. Amici 2003: 19, 26, fig. 10; Amici 2005.
215–20; De Magistris 1995: 87–93. 34. Rivoira 1925: 141.
8. Blake 1947: 330; Nielsen 1993: 244. 35. Bianchi 2000: 125 also agrees with this assessment.
9. Blake 1947: 179. 36. Choisy 1873: 40–1.
10. Giuliani 1970: 184. I thank C. M. Amici for bringing this 37. Casson 1994: 31–5, 106.
example to my attention. 38. Cod. Theod. 13.4.2, 14.6.3, 15.1.19.
11. An example of double relieving arches occurs in the Tomb 39. Cozza 1987: 42, fig. 44. In the fourth century, Gregory of
of Caecilia Metella on the Via Appia (Gerding 2002: 32–3, Nyssa (PG XLVI, 1097) sent a letter to the bishop of Ico-
fig. 16). nium in southwest Asia Minor requesting workmen skilled
12. Calza Bini 1953: 9, figs. 10, 15. in constructing vaults without centering because of a dearth
13. The technique of using cut stone vaulting ribs continues of wood in the area. His request probably refers to pitched
outside of Rome in places where bricks were not as readily brick vaults found in Greece and Asia Minor as early as the
available, as can be seen in the vaults of the amphitheaters at Trajanic period: Aupert and Ginouvès 1989: 151–5; Ward
Arles and Nimes, although in these examples the ribs do not Perkins 1981: 303–4. On pitched brick vaulting from the
support anything and may have been intended to aid in the Near East, see Van Beek 1987: 78–85.
construction rather than to reinforce a particular part of the 40. Rasch 1991: 370–2.
vault. 41. Steinby 1986: 123, 142.
14. Lancaster 1998b: 153–6. 42. The use of vertical formwork boards of regular lengths of
15. The travertine blocks of the wall piers continue upwards, so whole or half RF is typical in many late Roman domes
the blocks placed above the spring of the vault flare out in (Rasch 1991: 363, Tab. 2).
line with the curve of the vault to create a platform from 43. The calculation is based on the following assumptions: that
which the ladder rib can spring. These springer blocks also 1 m3 of concrete contains 250 pieces of caementa; that a mason
occur on the adjacent pier where there is no rib, which may can lay 1,500 pieces of caementa/day; that masons are spaced
suggest that there was a lack of coordination between the approximately 2 m apart. The methodology used is that set
stone carvers and the brick and concrete workers. forth by DeLaine 1997: 103–9.
16. They seem to be precursors to similar ones at the Hemicycle
at Trajan’s Markets (Lancaster 2000: 768–71).
17. Lugli 1918: 57–61; Lugli 1957: pl. CCV, 1. chapter 6. metal clamps and tie bars
18. This ribbing at the crown is similar to that found in the large
barrel vaults (14-m span) at a cistern at Chieti (Colosimo 1938 1. Martin 1965: 238–79; Orlandos 1968: 102–9; Dinsmoor
(pub. 1940): 254). 1922: 148–58.
19. For a detailed analysis of the ribs at Trajan’s Markets, see 2. Remains from the acropolis walls at Tivoli suggest that the
Lancaster 2000: 755–85. travertine quarries were opened as early as the third century
20. Amici 1991b: 118, fig. 210. b.c. (Giuliani 1970: 48). The earliest datable use as a building
21. Lancaster 2000: 779–84. stone in Rome probably occurs in the reconstruction of the
22. Dio Cass. 69.4.1. Temple of Concord by L. Opimius in 121 b.c. (Lugli 1957:
23. Heilmeyer 1975: 316–47. 321). For the early use of travertine, see also Mari 1983: 367.
24. DeLaine 1997: 164. For marble, see Pensabene 1994: 275–9.
25. Choisy 1873: 40–1. 3. Romualdi 1991: 79–80; Maggiani 1981: 191–2.
26. Cozzo 1928: 175–9. 4. Nielsen and Poulsen 1992: 88–9, 99. A block of tufo lionato
27. Giuliani 1990: 96. was also found with pi clamp cuttings. The tuff substructures
28. Rudimentary forms of lattice ribbing occur in the arch of a were held together with dovetail clamps (of wood?) the cut-
Flavian doorway of the Domus Tiberiana (Lancaster 1995a: tings of which are 23.5 cm long (including both sides), 5 cm
79, fig. 67C) and in two vaults at the Pozzuoli amphitheater wide at center, 9 cm at end, and 5.5 cm deep.
where a series of upright roof tiles appear along the entire in- 5. Rakob and Heilmeyer 1973: 12, Bei. 9, 10.
trados of the vaults (Lugli 1957: pl. CCV, 2; Lancaster 1995a: 6. Tabularium: Durm 1905: Abb. 27, 183; Forum of Augustus:
193). Ganzert and Kockel 1988: 167, Kat. 56, and Abb. 67; Ganzert
29. Lupu 1937: 154, fig. 31. 1985: 205–8, Abb. 1, Taf. 79.6; Temple of Mars Ultor: Lugli
30. DeLaine 1997: 162, fig. 80b. 1957: 239.
31. Cf., Bianchi 2000: 132–3. 7. Vitruvius (De arch. 2.8.3) recommends that marble or-
32. For photographs and drawings of the ribbing in the Arco di thostates be attached to tuff filler blocks with iron clamps
Malborghetto, see Messineo 1989: 47–57. and lead.
239
P1: NDZ
8. Blake 1947: 187. in the Farfa register such as the Licinianae, Marcianae, Oceanae,
9. Bauer and Pronti 1978: 130, fig. 15. and Platanianae.
10. Bauer 1988: Abb. 102–103. The bucranium block is about 32. DeLaine 1990: 410–21.
5 cm taller than the patera block, but the bucranium block has 33. Fant 1989; Pensabene 2002: 15–20.
the fine workings of the tooth chisel on the bottom surface, 34. Peacock 1994: 209–30; Peacock and Maxfield 1997: 333–4;
whereas the patera block has the rough working of the point Pensabene 2002: 23–6.
chisel suggesting that it has been cut down later. 35. Delbrueck 1907: Abb. 29 illustrates a block from the Tab-
11. Bauer 1988: Abb. 98, 102, 103. ularium but does not discuss it. P. Verducchi presented an
12. de Fine Licht 1974: 42–3; Brödner 1951: 28. impost block with cuttings for bars that she found in Ostia
13. de Fine Licht 1974: Abb. 5. at the Terme di Sette Sapienti (III.10.2) at the conference,
14. DeLaine 1985: 198–202. Cantieri antichi, held at the German Archaeological Institute
15. Brödner 1951: 28; DeLaine 1985: 198–202. (October 25, 2001). I also found an impost block with cut-
16. In personal conversations, Roberto Meneghini and Elisabetta tings that might have been used for metal bars lying in the
Bianchi, who are currently involved in reconstructing the lower level of the Domus Augustana: It has a recess for a
complex, have expressed their doubts as to the existence of marble slab along the bottom of a flat arch and a central spine
the tie bars. on the oblique sides but no holes at the top for lodging metal
17. Amici 1982: 84–5. bars as at Hadrian’s Villa.
18. Packer 1997: 446, folio 28. He sees the Column Portico cov- 36. Guidobaldi 1995: 419–41.
ered with a wooden roof instead of Amici’s vaulted terrace, 37. Pensabene 2001: 122–3.
but the recent discovery under the Church of S. Nome di 38. Ousterhout 1999: 210–16.
Maria of a stair along the north wall of the East Library sup- 39. DeLaine 1990: 421–2.
ports the idea of an accessible vaulted viewing gallery above 40. Sen., Ep. 115.8–9. Illos reperti in litore calculi leves et aliquid
the Column Portico (Meneghini 1993: 16, figs. 5, 7). habentes varietatis delectant, nos ingentium maculae colum-
19. Inv. 2706 = Packer 1997: Cat. 167D. Packer 1997: 433–4; narum, sive ex Aegyptiis harenis sive ex Africae solitudinibus
Piazzesi 1989: 167. advectae porticum aliquam vel capacem populi cenationem
20. Amici 1982: 12. ferunt. Miramur parietes tenui marmore inductos, cum sci-
21. Packer 1997: 217–19. amus, quale sit quod absconditur. Oculis nostris inponimus,
22. Inv. 2780 = Packer 1997: Cat. 167. Amici 1982: figs. 32–35 et cum auro tecta perfudimus, quid aliud quam mendacio
and Piazzesi 1989: 152–3 put it in a porch of the south façade, gaudemus? Scimus enim sub illo auro foeda ligna latitare.
which they restore with a small vault, whereas Packer 1997:
219 argues that this cornice piece was part of the south façade
itself. chapter 7. vault behavior
23. Amici 1982: fig. 53. and buttressing
24. Packer 1997: 441, folio 33.
25. A recent excavation unearthed more large chunks of vaulting 1. For the concept of Vitruvius’s symmetria, see Wilson Jones
from the east aisle of the Basilica Ulpia (personal observation 2000: 40–3.
December 2003), but no evidence for tie bars was found as 2. Wilson Jones 2000: 5–6.
far as I know. 3. Wilson Jones 2000: 102–3.
26. The blocks can be seen in the photographs in Giuliano et al. 4. Esposito and Michetti 1996: 61–84.
1988: 109. 5. For a discussion of the specifics of the design of a large build-
27. Hoffmann 1980: 26. ing project, see DeLaine 1997: 45–68.
28. This is based on my own measurements. Olivier 1983: 940–51 6. O’Connor 1993: 166–71.
gives a somewhat larger estimate. He notes that the cuttings 7. O’Connor 1993: 89, 172. Blake 1947: 217 notes that the
for the bars vary between 5 and 7 cm and estimates that the bridge may have been originally constructed in the second
bars themselves would have been about 4–5 cm in section. half of the second century b.c. but that the upper part appears
29. Olivier 1983: 947 n. 9 notes that the cuttings at the edges of to have been reconstructed, possibly under Augustus. Aosta
the panels were for forceps, but they do not align on either was founded as the military colony of Augusta Praetoria in
side of the slab, which suggests that they were in fact pry 24 b.c., and the road system was upgraded at this time, making
holes for levers. an Augustan date likely.
30. Hoffmann 1980: 26. 8. O’Connor 1993: 180–2.
31. Bloch 1947: 175–82 shows that the major figlina found at the 9. Vitr., De arch. 5.1.3, 5.1.5.
villa was the opus Salarese with many others that were listed 10. Vitr., De arch. 6.8.4.
240
P1: NDZ
11. The construction sequence of this room consists of two transmittentium lumen, ut suspensuras balneorum et inpres-
phases with the vault added in the second phase. The se- sos parietibus tubos, per quos circumfunderetur calor, qui
quence has been interpreted in various ways. Both Fabbrini ima simul ac summa foveret aequaliter.
1995: 58 and Ball 2003: 157–65 see the first phase wall as 28. Pompeii: Verità 1999: 108–10. A mid-first-century a.d. pro-
belonging to a Neronian project begun before the fire of duction center that made window glass has been excavated
a.d. 64 and the second phase representing a modification to at Sentium (Taborelli 1980: 147–9).
the design. I argue in my dissertation that that both phases 29. Sen., Ep. 86.8, 11 . . . at nunc blattaria vocant balnea, si qua
belong to the period after 64 and that the project was in- non ita aptata sunt, ut totius diei solem fenestris amplissimis
tended from the beginning to be built in two phases as a way recipient, nisi et lavantur simul et colorantur, nisi ex solio
of leaving an opening on the construction site that could agros ac maria prospiciunt. . . . Quantae nunc aliqui rustici-
be closed later with the large vault (Lancaster 1995a: 22–3, tatis damnant Scipionem, quod non in caldarium suum latis
29–32). A similar procedure was used to construct the Aula specularibus diem admiserat, quod non in multa luce deco-
at Trajan’s Markets, which is a building in which the dating quebatur et expectabat, ut in balneo concoqueret.
of construction joints is less controversial (Lancaster 2000: 30. Mart., Ep. 7.34.
776–9). 31. On the Baths of Nero, see Yegül 1992: 137–9; Ball 2003:
12. Alexander et al. 1977: 241–9; Heyman 1995: 59–61. The 238–49.
controversy arises in part from the different approaches taken, 32. Champlin 2003: 112–44.
elastic theory versus plastic theory, which is explained further 33. Sen, Ep. 115.13. Nihil illis melius nec dare videntur di im-
in Chapter 8. mortales posse nec habere. “Regia Solis erat sublimibus alta
13. For further analysis of the structure of this building, see columnis, Clara micante auro” (Ov., Met. 2.1). Eiusdem cur-
Lancaster 2000: 772–9. rum aspice: “Aureus axis erat, temo aureus, aurea summae,
14. Yegül 1992: fig. 157; Zorzi 1959: 67–8, figs. 106–109. The Curvatura rotae, radiorum argenteus ordo” (Ov., Met. 2.107).
original plan of the Baths of Nero remains an enigma. They Denique quod optimum videri volunt saeculum, aurem
were renovated by Alexander Severus in the third century, appellant.
but the extent of the change is unclear given the fragmentary
nature of the remains. Palladio’s drawings of the remains in
the sixteenth century indicate a symmetrical plan with the chapter 8. structural analysis:
standard triple cross vaulted frigidarium, but whether this re- history and case studies
flects the original Neronian project is unknown (Yegül 1992:
137–9, fig. 150). 1. Zalewski and Allen 1998: 215.
15. Amici 2003: 18, 23, fig. 5; Calabresi and Fattorini 2003: 81–9. 2. Timoshenko 1983: 3.
16. Heyman 1995: 41. This is based on a theoretical model of 3. For a good explanation of material properties, see Gordon
a dome shell with a thickness approaching zero. In a real 1976.
dome, this angle will vary somewhat depending on its form 4. Timoshenko 1983: 65.
and thickness. 5. Hooke 1679.
17. Middleton 1892b: 131. Followed by Robertson 1988: 233; 6. Heyman 1988: 739; Heyman 1995: 7.
Ward-Perkins 1981: 101; Adam 1994: 177; Blake 1959: 163. 7. Timoshenko 1983: 63, 323; Zalewski and Allen 1998: 215.
18. Terenzio 1932: 54. 8. Giovannoni 1904: 3–39.
19. Mainstone 1988: 87–9. 9. Milani 1923: 44–5.
20. MacDonald 1982: 110 n. 42; Mainstone 1975: 116. 10. For a brief explanation of the historical development of the
21. Lancaster 2000: 766–7. two theories, see Foce and Aita 2003: 895–900.
22. Rakob 1988: 289–91 discusses the proposals for the dating of 11. Mainstone 1992: 158–76.
the building and argues for an Augustan date. 12. Williams and Todd 2000: 312.
23. Rakob 1988: 262–5, Abb. 2–4. 13. Vitr., De arch. 6.8.4.
24. Rasch 1991: 333. 14. Gerhardt et al. 2003: 1000–3.
25. Rivoira 1925: 147–8. 15. Milani 1923: 44–5; Thode 1975: 135–46; Mark and Hutchin-
26. The monument has not been published in detail since the son 1986: 24–34.
recent cleaning of the vegetation that covered it and the sub- 16. Both Milani and Thode were working with the assumptions
sequent restoration. The main publications are Colini 1955: of elastic theory and then applying the middle third rule as
164–7 and Colli 1996: 782–9. a factor of safety, so the limit state was not the focus of their
27. Sen., Ep. 90.25. Quaedam nostra demum prodisse memo- investigations.
ria scimus, ut speculariorum usum perlucente testa clarum 17. Rankine 1858.
241
P1: NDZ
18. Ochsendorf 2002: 68–72. 16. RG 21; Alf öldy 1995: 195–226.
19. Adam 1994: 166; Milani 1923: 45. 17. Gell., NA 13.25.
20. This is same assumption that Milani 1923: 44–5 used in his 18. CIL III, 1312.
thrust line analysis. 19. Birley 1988: 128; Jones 1950: 28–9; SHA, Sev. 12.
21. Mark and Hutchinson 1986: 32 n. 19; Nilson and Darwin 20. BMC 5: pp. 221–2, nn. 358–62 (aurei/denarii, Rome mint),
1997: 47 give the formula for the tensile strength of con- pl. 36.12–13.
crete as three to five times the square root of the compressive 21. Duncan-Jones 1994: 16; Dio Cass. 77.9.4–5.
strength. Recent tests on Roman concrete give a compres- 22. Cod. Theod. 14.6.1–3; Symmachus, Relat. 40.
sive strength in the range of 50–62 kg /cm2 (Ferretti 1997: 72; 23. Yegül 1992: 2.
Tosi 1997: 124). 24. Strabo 5.3.8.
22. Mark and Hutchinson 1986: 24–34. 25. Clayton and Price 1988: 1–12.
23. Penta 1956: 88. 26. Mart., Spect. 1.70.7–8.
24. Ochsendorf 2002: 53. 27. Diod. Sic. 2.10 (translation from Loeb edition by C. H. Old-
25. Alexander et al. 1977: 241–51. father). This is the only description that discusses the method
26. Mainstone 1975: 118. of waterproofing used to protect the substructures, although
27. Heyman 1995: 141. other authors gives less detailed and sometimes conflicting
accounts of the structure itself: Curt. 5.1.35; Strabo 16.1.5;
Philo of Byzantium (see Finkel 1988: 45–6). A similar cov-
chapter 9. innovations in context ering of bitumin and lead was used to waterproof the hulls
of the Caligulan ships excavated at Lago di Nemi (Ucelli
1. Cameron 2003: 10. 1950: 149, 257). Lead sheeting was sometimes used for water-
2. The political aspect of the fourth category is my own addition proofing roofs as can be seen at a fragments of vaulting at
to the four categories proposed by Schlebeker 1977: 641–55 Hadrian’s Villa, where it was used between layers of opus
and used by White 1984: 21 in his study of Greek and Roman spicatum and the cocciopesto (Salza Prina Ricotti 2001: 235,
technology. fig. 79).
3. Vitr., De arch. 2.4.1. 28. Krause 1986: figs. 160–161; Krause 1994: 213–15, fig. 258.
4. Dio Cass. 48.51. 29. Sen., Ep. 122.8 (translation from the Loeb edition by R.
5. On the population increase in Rome during the late Repub- M. Gummere). Non vivunt contra naturam qui pomaria in
lic, see Purcell 1994: 644–53. summis turribus serunt? Quorum silvae in tectis domuum
6. Juv. 10.77–80. ac fastigiis nutant, inde ortis radicibus quo inprobe cacumina
7. Sen. Ep. 86.6, 90.15, 90. 26, 115.8. egissent?
8. Fire in 14 b.c.: Dio Cass. 54.24.2; in a.d. 64: Tac., Ann. 15.38– 30. See Purcell 1987: 190–203.
43; Suet., Ner. 31; in a.d. 80: Dio Cass. 66.24.1–3, Suet., Tit. 31. Tac., Ann. 15.42 (translation from Purcell 1987: 199) . . .
8.4; in a.d. 104: Jer., Ab Abr. 2120; in 110(?): Oros., 7.12.5; magistris et machinatoribus Severo et Celere, quibus inge-
Jer., Ab Abr. 2126; in a.d. 283: Chron. a. 354, 146 m; in a.d. nium et audacia erat etiam, quae natura denegavisset, per
307: Chron. a. 354, 148; Aur. Vict., Caes. 40.26. artem temptare et viribus principis inludere.
9. Tac., Ann. 15.43. 32. Aviary: Varro, Rust. 3.5.17; Ennius: Varro, Ling. 5.19.
10. On Augustus’s sources of wealth, see RG 15 (Caesar’s will, 33. Dio Cass. 53.27.2.
spoils of war, patrimonium); Suet., Aug. 101 (legacies); Dio 34. DeLaine 1997: 220–4.
Cass. 51.17.7 (bona damnatorum). On funding of building 35. Domitius brothers: Levick 1999: 110, 156, 175; ILS 990–1.
projects with his money, see RG 19–21; Suet., Aug. 29. On 36. A boundary marker with the name of the Domitii has been
funding of building projects by friends, see Suet., Aug. 29. found near Bomarzo (Graham 2002: 76). I thank Shawn
On contribution to the aerarium, see RG 17. Graham for bringing this recent find to my attention.
11. On iron from Noricum, see Strabo 5.1.8; Hor., Carm. 1.16.9, 37. Plin., Ep. 8.18.
Epod. 17.71. On the administration of marble quarries with 38. On the slaves of the Domitii, see Steinby 1974–1975: 49–57.
bibliography, see Pensabene 2002: 15–20. 39. Steinby 1974–1975: 92. The officinator in question is Vis-
12. On bona damnatorum, see Suet., Ner. 32; Dio Cass. 61.5.3–6, matius Successus who appears as officinator of Calpernia Se-
62.18.5. On Nero’s confiscations and monetary reforms, see cunda (CIL XV, 611–12) and then later as the patronus of
Duncan-Jones 1994: 6, 12, 219–24. Vismatius Felix (CIL XV, 613a–b).
13. Dio Cass. 66.8.3–4. 40. Helen 1975: 112–13.
14. Jones 1950: 22–9; Millar 1963: 28–42. 41. Steinby 1978: col. 1494.
15. Jones 1950: 27. 42. Steinby 1974–1975: 110–11.
242
P1: NDZ
43. Plin., Ep. 7.19; Bloch 1947: 338. Rome, Helena herself may have had direct control over the
44. On the disappearance of stamped bricks after Caracalla, see construction on her property.
Steinby 2001: 128–32. On the decline of slaves in the brick 63. Deichmann and Tschira 1957: 74; Krautheimer 1993: 529–
industry, see Steinby 1974–1975: 58. 30; Guyon 1986: 321.
45. CIL VI, 1130 = ILS 646. 64. Guyon 1987: 259.
46. Bloch 1947: 311–16. 65. Frazer 1966: 387; Ward-Perkins 1966: 297–9.
47. Barnes 1982: 12–13. 66. Seston 1946: 206–10, 215.
48. Lactant., De mort. pers. 26.1–3; Aur. Vict., Caes. 40. 67. MacCormack 1981: 105–10, 133.
49. RIC 6: p. 372, n. 166 (aureus, Rome mint); p. 374, nn. 177– 68. On the changes in imperial funerary rites from the early to
8 (aureus, Rome mint); p. 375, n. 187 (silver, Rome mint); the late empire, see Price 1987: 91–105.
pp. 376–7, nn. 194–205 (aes, Rome mint); p. 378, nn. 208–13 69. Liber Pontificalis 34.26; Guyon 1987: 258–9.
(aes, Rome mint); pp. 382–3, nn. 258–63 (aes, Rome mint); 70. For the merging of imperial funeral celebrations with Chris-
p. 400, n. 2 (aureus, Ostia mint). tian ceremony, see MacCormack 1981: 131.
50. Lactantius (De mort. pers. 7.8–10) described Diocletian’s 71. Euseb., Vit. Const. 1.2.2. (translation from Cameron and Hall
building activities in the provinces and the burdens it caused 1999: 67).
to the local inhabitants. 72. Ward-Perkins 1994: 495–515.
51. Steinby 1986: 157. After the creation of Constantinople, some 73. Deichmann and Tschira 1957: 58–9; Vendittelli 2002: 780–2.
private figlinae reappear, and the owners are typically members 74. Liber Pontificalis 34.26.
of the highest order of society, that is, that from which the 75. Pan. Lat. 7.7.1 (translation from MacCormack 1981: 108) . . .
praefectus urbi typically comes. iturus ad deos genitorem illum deorum ignea caeli astra re-
52. Bloch 1947: 311–16 argued that the organization was struc- fouentem prospexit Oceanum, ut fruiturus exinde luce per-
tured differently than before because the names of officinatores petua iam uideret illic diem paene continuum.
were substituted by numbered stations (stationes) in those pro- 76. Krautheimer 1964: 224–9.
duction units owned by the res privata. Steinby 2001: 129–30, 77. For tubi fittili, see Bovini 1960: 78–99. For pitched brick
by contrast, sees the stamps as continuing to refer to contracts vaulting in Byzantine architecture, see Ousterhout 1999:
of locatio conductio whereby the state contracted with private 201–33.
individuals. 78. Mazzucato 1970: 339–70.
53. Constantine was in Rome only three times: from October
312 to January 313; from July 315 to September 315; and for a
few months in 326. appendix 3. scoria analysis
54. An inscription (CIL VI, 1136) found at the Sessorian palace
records her as the grandmother of at least two grand sons but 1. I thank Mario Gaeta of the Dipartimento di Scienze della
not yet as Augusta, which dates it to between 317 and 324 Terra, Università “La Sapienza” Rome, Italy, who prepared
(Drijvers 1992: 47–8). On Helena’s the land holdings “ad and interpreted the thin sections for me.
duas lauros,” see Guyon 1986: 300–3. 2. De Angelis d’Ossat 1930: 211–15.
55. LTUR IV: 304–8 (Guidobaldi); Guidobaldi 1999: 62; Cima 3. To my knowledge, the scientific basis for these ascertains have
1995: 64–5. not been published.
56. Steinby 1986: 141, n. 93; Heres 1983: 356–60. 4. Triglia et al. 1995: 127; De Casa et al. 1999: 8.
57. Biasci 2000: 79–87; Cima 1995: 62; LTUR IV: 307. 5. Scherillo 1941: 387.
58. For a discussion of the changing functions of the Roman 6. Ranieri and Yokoyama 1997: 33–50; Yokoyama and
aristocratic house in late antiquity, see Ellis 1988: 565– Marturano 1997: 1–32.
76. 7. Nappi et al. 1994: 502–15. Of the deposits discussed, the
59. Steinby 1986: 142; Steinby 2001: 134. Aurelius Victor (Caes. one called “Ospedaletto pumices (P4)” has the mineralogical
40) credits it to Constantine. profile that most closely matches the Colosseum sample. It
60. Krautheimer 1993: 542. contains phenocrysts of sanidine and plagioclase with smaller
61. Deichmann and Tschira 1957: 64. amounts of biotite, augite, leucite, magnetite, and apatite.
62. Eusebius (Vit. Const. 3.47.3) notes that Helena had full ac- However, further thin section analyses on samples from the
cess to the imperial treasury. As Constantine was rarely in actual deposits are required to confirm the correspondence.
243
P1: NDZ
244
P1: NDZ
0521842026gsy.xml CB885B/Lancaster 0 521 84202 6 April 18, 2005 12:21
GLOSSARY
The following glossary is intended to provide definitions for CENTER OF GRAVITY – the point at which the object will
technical and culture specific words. The key words are de- balance as if the whole weight of the object is concentrated
noted in capital letters. The use of capital letters within each at that point, as on a fulcrum.
entry signifies a cross-reference to another keyword in the list. CLIENTELA – Roman institution by which two people of
different status are informally bonded, the person of lower
ABUTMENT – the part of a structure that bears the weight or status, the client, typically providing services and political
pressure of an arch or vault. support to the person of higher status, the patron, who in
ANATHYROSIS – the smooth dressing along the margins of turn provides financial support, food, and other aid.
a block that form a joint surface. COCCIOPESTO – Italian term for a hydraulic mortar made
ANIENE TUFA (= TUFO LIONATO) – traditional archaeo- with crushed terracotta.
logical term for reddish brown tuff from Colli Albani vol- COFFER – a recessed decorative feature used in vaults.
canic district (Map 3, p. 14 and Pl. I). COLLEGIUM FABRUM TIGNARIORUM – builder’s associa-
ARCUATED LINTEL – a LINTEL that takes the form of an tion (literally, “woodworker’s guild”). The association re-
arch (instead of a flat beam). quired an entry fee and consisted of freeborn and freed-
BESSALIS – Roman brick 2/3 RF square (c. 20 cm). men builders (not limited to wood workers in spite of the
BIPEDALIS – Roman brick 2 RF square (c. 58 cm). name). It did not have the role of controlling artisans as in
BUTTRESS – a structure built up against a wall for support or the medieval period nor did it negotiate fees as do modern
reinforcement. unions. It was a social organization focusing on the well-
CANTILEVER – a projecting horizontal beam supported at to-do builders (as opposed to simple laborers), although it
one end only. may have been a way of establishing contacts and gathering
CARBURIZATION – the process of increasing the carbon labor.
content of iron so that it can be heat treated to form steel. COMPRESSION – the state of a material that is being com-
The Romans typically accomplished this by heating iron pacted together.
with charcoal, which has a high carbon content, and by COMPRESSIVE STRESS – internal distribution of forces
hammering it. within a material in COMPRESSION.
CEMENT – a powdered substance made from lime and clay, CORBEL – a bracket of stone or wood projecting from a pier
which when mixed with water can be used to make mor- or wall face and supporting something above, such as a
tar or concrete (see PORTLAND CEMENT). The term cornice, arch, vault, or beam. It is based on the structural
refers to modern material and has no relevance in ancient principle of the CANTILEVER.
construction. CREEP – the slow permanent deformation of a material under
CENTERING – a temporary structure used to support an arch stress.
or vault during construction. CUBIT – unit of linear measure equal to 1 12 Roman feet.
245
P1: NDZ
GLOSSARY
CROWN – the uppermost zone of an arch or vault. GROTTA OSCURA TUFA (= TUFO GIALLO DELLA VIA
DEAD LOAD – weight of a permanent part of the structure. T IBERINA) – traditional archaeological name for a yellow
DECENTER – to remove the centering from an arch or vault. tuff found north of Rome. Product of Sabatini volcanic
1
DIGITUS – unit of linear measure equal to 16 Roman foot. district. Ancient quarries were found near an area called
DOMINUS – in the brick industry, signifies the owner of land Grotta Oscura (Map 3, p. 14 and Pl. I).
from where the clay was taken. HAUNCH – the lower part of a vault or arch just above its
EASING – the process of slowly lowering the CENTERING springing point.
from a hardened concrete vault. HOOP TENSION – refers to the circumferential tensile stresses
ELASTIC THEORY – Structural design theory in which the that occur in the lower part of a dome.
structure is assumed to be able to deform without failing HORTI PENSILES – a term that literally means hanging gar-
under moderate loads. Stresses are determined under certain dens and is used to refer to gardens that are elevated above
loading conditions, and then the members are designed their surroundings.
with a safety factor based on the YIELD STRENGTH of IMPOST – the point from which an arch or vault springs.
the material in relation to that of the structure. IMPOST BLOCK – a stone block trapezoidal in form that
ENTABLATURE – the superstructure (consisting of the archi- occurs directly above a column or pier and serves to accept
trave, frieze, and cornice) located above columns in classical the ends of arches.
architecture. IMBREX (pl. imbrices) – rounded terracotta roof tile used to
EXTRADOS – the exterior surface of an arch or vault. cover the joints between TEGULAE.
FELDSPAR – a group of minerals consisting of aluminum INTRADOS – the inner surface of an arch or vault.
silicates of potassium, sodium, calcium, or barium. Includes LEUCITE – a mineral characterized by translucent or opaque
the minerals sanidine and orthoclase as well as those of the whitish color and spherical tetragonal form and consist-
PLAGIOCLASE series. ing of potassium aluminum silicate. It is common to the
FELDSPATHOID – group of minerals chemically related to volcanic products of Colli Albani and Vesuvius.
feldspars but with less silica; includes leucitite, nepheline, LEUCITITIC LAVA – a dense gray lava containing much
and sodalite among others. leucite. Product of Colli Albani volcanic system and typi-
FIDENAE TUFA (= TUFO ROSSO A SCORIE NERE) – cally used for paving stones. Often called “selce.”
traditional archaeological name for the tuff with large black LINE OF THRUST – the internal line of forces within a struc-
scoriae that is found near the ancient town of Fidenae north ture due to the applied loads (including the weight of the
of Rome (Map 3, p. 14 and Pl. I). Product of the Sabatini structure itself ).
volcanic district. LINTEL – a flat beam that spans an opening between two walls
FIGLINA – on brick stamps, signifies the name of the place of or columns.
production of the bricks, a production unit; alternatively, LOCATIO CONDUCTIO – type of contract often used in
some scholars argue that it signifies the name of clay beds. building industry in ancient Rome. Could be used in three
FINITE ELEMENT MODELING – a type of structural analy- different ways: locatio conductio operis (the locator lets out a job
sis based on ELASTIC THEORY in which the structure is to be completed by the conductor), locatio conductio operarum
modeled with computer software by creating a mesh of dis- (the locator lets out his own labor to the conductor, i.e., day
crete elements joined together at nodes. The stresses that labor), and locatio conductio rei (the locator lets out the object
develop at the nodes under particular assigned loads can of the contract to the conductor). A contract of locatio conductio
then be calculated. Once the maximum stresses are deter- included a final inspection (probatio) and an agreed-on price
mined in the model, they can be compared to the tested (merces).
strengths of the materials employed. LUNETTE – the semicircular space underneath an arch.
FORCE – an application of pressure that produces the accel- MICROCRYSTS – very small crystals.
eration of a body in the direction of pressure. A force is MIDDLE THIRD RULE – states that the curve of the line of
measured in Newtons (N) or kilonewtons (kN) and can be thrust should remain within the middle third of the section
drawn as a vector showing magnitude and direction. of the arch and its abutment to be safe. This rule is simply
FORMWORK – the covering on the CENTERING of a vault a geometrical safety factor of three based on the line of
to create the form on which the concrete is placed. thrust.
246
P1: NDZ
GLOSSARY
MOMENT – a tendency to produce rotation about a point or PLAGIOCLASE – a subset of FELDSPAR minerals consisting
axis. Measured in terms of the product of force or mass and of sodium and calcium aluminum silicates. Includes albite,
its distance from the point of rotation (Fig. 134, p. 151). oligoclase, andesine, bytownite, anorthite, and labradorite.
MONTEVERDE TUFA (= TUFO LIONATO) – traditional PLASTIC THEORY – theory of structural design, which
archaeological name for the reddish brown tuff quarried when applied to masonry structures is based on the as-
in the Trastevere area of Rome along the slopes of Mon- sumptions (1) that because the compressive yield strength
teverde. Geologically it was produced by the same event as of the materials is so much greater than the compressive
ANIENE TUFA (TUFO LIONATO). stresses that could ever develop in the structure, the ma-
OFFICINATOR – on brick stamps indicates the person in terials can effectively be considered to be infinitely strong
charge of the fabrication of the brick. in compression, (2) that the material has no resistance to
OPUS CAEMENTICIUM – term referring to ancient concrete tension, and (3) that slippage cannot occur between stones.
(meaning mortared rubble work). In plastic theory the strengths of the materials are irrelevant
OPUS QUADRATUM – term referring to technique of cut since they are accounted for in assumptions 1 and 2.
stone construction using four-sided blocks. PORPHYRIC – the groundmass contains PHENOCRYSTS.
OPUS RETICULATUM – term referring to a type of facing PORPHYRICITY INDEX – indicates the amount of PHE-
for concrete walls in which small pyramidal shaped stones NOCRYSTS (and their dimensions) with respect to the
are set so that the joints between them form a diagonal grid groundmass.
on the face of the wall. PORTLAND CEMENT – a hydraulic cement (i.e., when
OPUS SIGNINUM – ancient term used to refer to method mixed with water it hardens under water) made from lime-
of building waterproof structures but often used synony- stone providing lime (CaO) and clays or shales providing
mously with COCCIOPESTO. silica (SiO2 ) and alumina (Al2 O3 ). These components are
OPUS SPICATUM – type of paving employing small bricks ground together and fused in a kiln; the resulting clinker is
set in herringbone fashion. then ground into cement powder. It takes its name from its
OPUS TESTACEUM – term referring to a wall facing of similarity in appearance to Portland stone quarried on the
mortared brickwork. Isle of Portland off the southern coast of England.
PECULIUM – sum of money given to a slave to use as his own POZZOLANA – a volcanic ash rich in silica and alumina used
but which reverts to the owner upon the death of the slave. combined with lime to create a very strong hydraulic mortar
PEDES (= ROMAN FOOT = RF) – ancient unit of linear (Pl. III).
measure equal to 29.5 cm. PUMICE – a volcanic material consisting of glass and various
PEPERINO – a dense consolidated gray tuff dappled with types of mineral crystals. Created by violent airborne ex-
black and white lithic fragments. Product of the Colli Al- plosion resulting in a very light vesicular material that can
bani volcanic district (Map 3, p. 14 and Pl. I). appear in a variety of colors.
PHENOCRYSTS – large crystals. PUTLOG HOLE – the holes in a wall in which the ends of
PHONOLITE – group of rocks composed primarily of al- scaffolding beams were lodged during construction. In
kali FELDSPAR (especially anorthoclase or sanidine) with Roman buildings they were usually covered by the final
nepheline as the main FELDSPATHOID. wall facing.
PHOTOGRAMMETRY – a method of measuring an object or QUENCHING – the process of hardening metal through heat-
structure by means of multiple photographs that have coming it and then rapidly cooling it by plunging it into
mon control points between them, which can be recreated liquid.
in 3D space using principles of trigonometry. The control QUICKLIME – calcium oxide (CaO) or unslaked lime. It re-
points represent the nodes of triangles used in the calcula- sults from kiln fired stones that are high in calcium carbon-
tions. By taking photographs, which include at least three ate (CaCO3 ) such as lime stone, marble, and travertine. It
of the control points, from different perspectives, the points comes out of the kiln in the form of a very lightweight rock
can be placed in an x-,y-,z-coordinate system to create a (Fig. 40, p. 54).
3D model of the structure. REDEMPTOR (pl. redemptores) – Latin term for one who con-
PI CLAMP – a clamp formed like the letter pi, “π ,” with the tracts to provide goods or services. Often used to describe
two ends bent down so that it looks like a staple. building contractors.
247
P1: NDZ
GLOSSARY
RELIEVING ARCH – an arch built into a wall usually to relieve floors for warehouses, hot rooms in bath buildings, and
the pressure from an opening below. Differs from an arch elevated gardens.
in that there is walling in the LUNETTE. SURCHARGE – the added weight above the haunch of a
RF (= ROMAN FOOT = PEDES) – ancient unit of linear vault that counters the horizontal thrust of the vault on
measurement that equals 29.5 cm. its support. In domes it often the takes the form of STEP-
SANIDINE – a mineral of the FELDSPAR group character- RINGS.
ized by transparent or whitish translucent color and tab- TEGULA (pl. tegulae) – flat, terracotta roof tile with upturned
ular form and consisting of potassium aluminum silicate. flanges along two sides.
Common to products of the Sabatini and Vulsini volcanic TEPHRITE – group of rocks characterized by a combination
districts. of plagioclase, augite, and leucite or nepheline.
SCORIA – a light vesicular, vitreous volcanic product that TENSILE STRESS – internal distribution of forces within a
often occurs as the top layer of a lava flow. Formed as the material in TENSION.
gases escaped as the lava cooled. Similar to PUMICE but TENSION – the state of a material that is being stretched or
somewhat denser (Pl. VIII). pulled apart.
SCORIACEOUS – texture consisting of coarsely vesicular rock THIN SECTION PETROGRAPHY – a technique for deter-
in which the vesicles can be either smooth or jagged. mining the mineral contents of a rock or other material
SELCE – see LEUCITITIC LAVA. containing minerals such as terracotta or mortar. A section
SEMILATERES – a half brick usually cut along the diagonal of the material is sliced to a standard thickness of 0.03 mm
to produce a triangle; a diagonally cut BESSALIS. and glued to a glass slide. When examined under a polar-
SESQUIPEDALIS – Roman brick 1 12 RF square (c. 44 cm). izing microscope, the light penetrates the thin section and
SHRINKAGE – a phenomenon exhibited by concrete as it interacts with certain minerals in defined ways. The miner-
cures. Shrinkage in modern concrete averages 0.15–0.30 als are distinguished by their known optical properties such
percent. The rate of shrinkage increases with the amount as color, transparency, or refractivity among others. Because
of water used to mix the mortar. both rocks and clays can have distinctive mineralogical sig-
SLAKING – the process of adding water to QUICK- natures, this method can be very useful for determining
LIME (CaO) to create slaked lime (CaO2 H2 ), which provenance.
can then be combined with sand or pozzolana to make THREE-HINGE ARCH – an arch that has three possible rota-
mortar. tion points (e.g., at cracks). It is inherently stable in spite of
SOFFIT – the exposed lower surface of a lintel or architrave. the rotation points, but the development of a fourth hinge
SPANDREL – the roughly triangular space between the would cause failure.
haunches of two adjacent arches. THRUST LINE ANALYSIS – a type of analysis used to de-
STEP-RING – refers to the steps often built along the outer termine the LINE OF THRUST through a structure (as
haunch of domes and semidomes to provide a SUR- opposed to STRESS ANALYSIS).
CHARGE to counter horizontal thrusts. TRABEATED – refers to a structure with horizontal lintels
STIFFNESS – property of a material that is measured in terms (derived from the Latin trabs meaning “beam”).
of strain values. TRACHYTE – group of rocks characterized by much
STRENGTH – property of a material that is measured in terms FELDSPAR, especially SANIDINE.
of STRESS values. TRAVERTINE – a sedimentary stone formed by the precipi-
STRESS – internal distribution of forces within a material. tation of calcium carbonate (CaCO3 ) from water. Usually
Measured as force per unit area, usually expressed in terms characterized by strata of small holes created by the presence
of kg/cm2 . of plant life during the formation process (Pl. I).
STRESS ANALYSIS – a type of structural analysis the goal of TRUSS (TRIANGULAR) – an assemblage of wooden beams
which is to determine levels of stress within a structure (as into a triangular form (or a series of triangular forms), sup-
opposed to THRUST LINE ANALYIS). ported at the two endpoints. The beams are arranged so
STRIKING – the act of removing the centering from a that the lower member (tie beam) is in TENSION and
vault. the oblique side members (rafters) are in COMPRESSION
SUSPENSURAE – posts used to elevate a floor. Typically used (Fig. 9, p. 23). The assemblage into the triangular form cre-
by Roman builders to create air spaces by means of double ates a structure that can span much greater distances than
248
P1: NDZ
GLOSSARY
a simple beam and that can be built of smaller beams of fact, a sedimentary rock formed of calcium carbonate
wood. (CaCO3 ).
1
TUBI FITTILI – terracotta tubes made with one end open UNCIA – unit of linear measurement equal to 12 of a Roman
and the other closed in a point so that they a can be joined foot.
together end to end. Used to create permanent centerings VITROPHYRIC – term describing rock in which the ground-
for concrete vaults. mass is composed largely of glass with a small presence of
TUFF – volcanic rock formed of consolidated ash ejected MICROCRYSTS.
during volcanic explosion. Consists of SCORIA, PUMICE, YIELD STRENGTH – the point at which a material loses
and lithic fragments within an ash matrix. In archae- its elasticity and will no longer return to its original
ological literature, often called “tufa,” which is, in form.
249
P1: NDZ
250
P1: KIC
0521842026cit.xml CB885B/Lancaster 0 521 84202 6 April 18, 2005 12:27
WORKS CITED
For periodicals, the abbreviations listed in American Journal of Amici, C. M. 2005. “Dal progetto al monumento,” “Le tec-
Archaeology 95 (1991), 4–16 are followed; for ancient authors niche di cantiere ed il procedimento costruttivo.” In C.
and reference works, those in the Oxford Classical Dictionary Giavarini, ed. La Basilica di Massenzio: il monumento le strut-
(2003). ture, la stabilità (Rome): 21–74, 125–60.
Anderson, J. C. 1997. Roman Architecture and Society.
Adam, J.-P. 1994. Roman Building: Materials and Techniques. Baltimore: Johns Hopkins University Press.
Translated by A. Mathews. Indianapolis: Indiana University Ashby, T. 1907. “The Classical Topography of the Roman
Press. Campagna III.” PBSR 4.1: 3–159.
Alexander, K. D., R. Mark, and J. F. Abel 1977. “The Struc- Ashby, T. 1924. “La via Tiberina e i territori di Capena e del
tural Behavior of Medieval Ribbed Vaulting.” JSAH 36: Soratte nel periodo romano.” MemPontAcc 1.2: 129–75.
241–51. Ashby, T. 1935. The Aqueducts of Ancient Rome. Oxford:
Alföldy, G. 1995. “Ein Bauinschrift aus dem Colosseum.” Clarendon Press.
ZPE 109: 195–226. Ashby, T., and G. Lugli 1928. “La villa dei Flavi cristiani ‘ad
Alföldy, G. 2001. “Difficillima Tempora: Urban Life, Inscrip- duas lauros’ e il suburbano imperiale ad oriente di Roma.”
tions, and Mentality in Late Antique Rome.” In T. S. Burns MemPontAcc (ser. 3) 2: 157–92.
and J. W. Eadie, eds. Urban Centers and Rural Contexts in Late Astolfi, A., F. Guidobaldi, and A. Pronti 1978. “Horrea Agrip-
Antiquity (East Lansing, Mich.): 3–24. piana.” ArchCl 30: 31–100.
Amici, C. M. 1982. Foro di Traiano: Basilica Ulpia e Biblioteche. Ball, L. F. 1994. “A Reappraisal of Nero’s Domus Aurea.” In
Rome: Comune di Roma: X Ripartizione antichità belle Rome Papers ( JRA Supp. 11, Portsmouth, R. I.): 183–254.
arti e problemi di cultura. Ball, L. F. 2003. The Domus Aurea and the Roman Ar-
Amici, C. M. 1991a. “Elementi fittili nella chiesa di S. An- chitectural Revolution. Cambridge: Cambridge University
tonio in Archi a Reggio Calabria: vasi echitici o artifici Press.
strutturali?” Quaderni del Dipartimento Patrimonio Architet- Barnes, T. D. 1982. The New Empire of Diocletian and Constan-
tonico e Urbanistico, Università di Lecce 1.2: 46–7. tine. Cambridge, Mass.: Harvard University Press.
Amici, C. M. 1991b. Il Foro di Cesare. Florence: Leo S. Bartolini, F., and R. Turchetti 1999. “Magazzini cosiddetto
Olschki. Traianei.” Meded 58: 26–8.
Amici, C. M. 1997. “L’uso del ferro nelle strutture romane.” Bauer, H. 1978. “Un tentativo di ricostruzione degli Horrea
Materiali e strutture 7.2–3: 85–95. Agrippiana.” ArchCl 30: 132–46.
Amici, C. M. 2003. “L’iter progettuale e costruttivo nel con- Bauer, H. 1988. “Basilica Aemilia.” In Kaiser Augustus und die
testo storico topografico,” La Basilica di Massenzio: ricerca Verlorene Republik (Mainz): 200–12.
interdisciplinare applicata allo studio e alla conservazione di un Bauer, H., and A. Pronti 1978. “Elementi architettonici degli
monumento: Atti del Convegno (Rome): 17–29. Horrea Agrippiana.” ArchCl 30: 107–31.
251
P1: KIC
WORKS CITED
Bazant, Z. P., and F. H. Wittmann 1982. Creep and Shrinkage tura, Dirección General de Bellas Artes y Archivos, Instituto
in Concrete Structures. New York: John Wiley & Sons. de Conservación y Restauración de Bienes Culturales.
Beltrami, L. 1892. “Roma. Nuove scoperte nella città e nel Bloch, H. 1947. I bolli laterizi e la storia edilizia romana. Rome:
suburbio. Regione IX.” NSc: 88–90. Comune di Roma, Ripartizione antichità e belle arti.
Beltrami, L. 1898. Il Pantheon. Milan: U. Allegretti. Boersma, J., D. Yntema, and J. van der Werff 1986. “Ex-
Beste, H.-J. 1998. “Relazione sulle indagine in corso nei sot- cavations in the House of the Porch (V.ii.4–5) at Ostia.”
terranei, i cosiddetto ipogei.” RM 105: 106–18. BABesch 61: 77–137.
Beste, H.-J. 1999. “Neue Forschungsergebnisse zu einem Boethius, A. 1987. Etruscan and Early Roman Architecture.
Aufzugssystem im Untergeschoss des Kolosseums.” RM Harmondsworth: Penguin.
106: 249–76. Boethius, A., and N. Carlgren 1932. “Die spätre-
Beste, H.-J. 2000. “The Construction and Phases of Develop- publikanischen Warenhäuser in Ferentino und Tivoli.”
ment of the Wooden Arena Flooring of the Colosseum.” ActaArch 3.3: 181–208.
JRA 13.1: 79–92. Bost, J.-P., M. Campo, D. Colls, V. Guerrero, and F. Mayet
Beste, H.-J. 2001. “I sotterranei del Colosseo: impianto, 1992. L’épave Cabrera III (Majorque): exchanges commerciaux
trasformazioni e funzionamento.” In A. La Regina, ed. et circuits monétaires au milieu du III e siècle après Jèsus-Christ.
Sangue e arena (Milan): 277–99. Paris: De Boccard.
Bianchi, E. 2000. “Le nervature nelle volte massive di età Bovini, G. 1960. “L’impiego dei tubi fittili nelle volte degli
romana.” BullCom 101: 105–62. edifici di culto ravennati.” Felix Ravenna 30: 78–99.
Bianchi, E. 2001. “I bolli laterizi del Foro di Traiano: il cat- Boynton, R. S. 1980. Chemistry and Technology of Lime and
alogo del Bloch e i rinvenimenti delle campagne di scavo Limestones. 2nd ed. rev. New York: John Wiley & Sons.
1991–1997 e 1998–2000.” BullCom 102: 82–120. Brandon, C. 1999. “Pozzolana, Lime, and Single Mission
Bianchi, E., and R. Meneghini 2002. “Il cantiere costruttivo Barges.” In K. G. Holum, A. Raban, and J. Patrich, eds.
del Foro di Traiano.” RM 109: 395–417. Caesarea Papers ( JRA Supp. 35, Portsmouth, R. I.): 169–78.
Bianchini, F. 1738. Il palazzo dei Cesari. Verona: P. Berno. Brödner, E. 1951. Untersuchungen an den Caracallathermen.
Bianchini, M. 1991. “I ‘Mercati di Traiano’.” Bollettino di Berlin: Walter de Gruyter & Co.
archeologia 8: 102–21. Brown, F. E. 1964. “Hadrianic Architecture.” In L. F. Sandler,
Biasci, A. 2000. “Il padiglione del ‘Tempio di Minerva Medicà ed. Essays in Memory of Karl Lehmann (New York): 55–8.
a Roma: struttura, tecniche di costruzione e particulari Bruni, S. 2000. “Il porto urbano di Pisa e i relitti del complesso
inediti.” Science and Technology for Cultural Heritage 9. 1–2: ferroviario di ‘Pisa-San Rosore’.” In S. Bruni, ed. Le navi
67–88. antiche di Pisa: ad un anno dall’inizio delle ricerche (Florence):
Birley, A. R. 1988. The African Emperor Septimius Severus. New 21–79.
Haven: Yale University Press. Brunt, P. A. 1980. “Free Labour and Public Works at Rome.”
Blake, M. E. 1947. Ancient Roman Construction in Italy from the JRS 70: 81–100.
Prehistoric Period to Augustus. Washington, D.C.: Carnegie Bruun, C. 1996. Review of Die kaiserliche Bauverwaltung in der
Institution of Washington. Stadt Rom: Geschichte und Aufbau der cura operum publicorum
Blake, M. E. 1959. Roman Construction in Italy from Tiberius unter dem Prinzipat (1993), by A. Kolb. BJb 196: 735–40.
Through the Flavians. Washington, D.C.: Carnegie Institu- Burford, A. M. 1969. The Greek Temple Builders at Epidauros.
tion of Washington. Liverpool: Liverpool University Press.
Blake, M. E. 1973. Roman Construction in Italy from Nerva Burnell, G. R. 1850. A Rudimentary Treatise on Limes, Cements,
Through the Antonines. Edited by D. T. Bishop. Philadelphia: Mortars, Concretes, Mastics, Plastering, etc. London: J. Weale.
American Philosophical Society. Calabresi, G., and M. Fattorini 2003. “Il terreno e le fon-
Blázquez Martı́nez, J. M. 1992. “The Latest Work on the dazioni.” La Basilica di Massenzio: ricerca interdisciplinare ap-
Export of Baetican Olive Oil to Rome and the Army.” plicata allo studio e alla conservazione di un monumento: Atti del
GaR 39: 173–88. Convegno (Rome): 81–9.
Blázquez Martı́nez, J. M., J. R. Rodrı́guez, and E. Rodrı́guez- Calza Bini, A. 1953. “Il teatro di Marcello. Forma e strutture.”
Almeida 1994. Excavaciones arqueologı́cas en el Monte Testaccio BCSSA 7: 1–44.
(Roma). Memoria campaña 1989. Madrid: Ministerio de Cul- Calza, G. 1953. Scavi di Ostia 1. Rome: Libreria dello Stato.
252
P1: KIC
WORKS CITED
Cameron, A. 2003. “Ideologies and Agendas in Late Antique Casson, L. 1994. Ships and Seafaring in Ancient Times. Austin:
Studies.” In L. Lavan and W. Bowden, eds. Theory and Prac- Texas University Press.
tice in Late Antique Archaeology (Leiden): 2–21. Castagnoli, F. 1949–1950. “Documenti di scavi eseguiti in
Cameron, A., and S. G. Hall 1999. Eusebius, Life of Constantine. Roma negli anni 1860–1870.” BullCom 73: 167–73.
Oxford: Oxford University Press. Champlin, E. 2003. Nero. Cambridge, Mass.: Harvard Uni-
Candilio, D. 1985. “Terme di Diocleziano – Museo nazionale versity Press.
romano: scavo nella palaestra nord-occidentale.” In Roma. Chastagnol, A. 1960. La préfecture urbaine à Rome sous le Bas-
Archeologia nel Centro II. La ‘Città Murata’ (LSA 6.2, Rome): Empire. Paris: Presses universitaires de France.
525–32. Chevallier, R. 1976. Roman Roads. Translated by N. H. Field.
Canina, L. 1860. “Sulla scoperta della Basilica Giulia al Foro London: Batsford.
Romano.” DissPontAcc 14: 179–94. Chiari, G., L. Cimitàn, G. D. Ventura, M. G. Filetici, M. L.
Cantasini, E., F. Fratini, P. Pallecchi, E. Pecchioni, and S. Santarelli, and G. Torraca 1996. “Le malte pozzolaniche del
Rescic 2003. Abstract of “Characterisation of Stones used as mausoleo di Sant’Elena e le pozzolane di Torpignattara.”
Stowage and Ballast in the Ships of the Ancient San Rossore Materiali e Strutture 4.1: 1–36.
Harbour (Pisa).” ASMOSIA VII, 15–20 September 2003 Chiari, G., M. L. Santarelli, and G. Torraca 1992. “Caratter-
(Thasos, Greece). izzazione delle malte antiche mediante l’analisi di campioni
Caraffa, G. 1942. L’amplimento della Piazza dei Cinquecento e le non frazionati.” Materiali e Strutture 2.3: 111–37.
Terme di Diocleziano. Rome: V. Ferri. Choisy, A. 1873. L’art de bâtir chez les romaines. Paris: Ducher
Caraffa, G. 1944. La cupola della sala decagona degli Horti & Cie.
Liciniani – restauri 1942. Rome: V. Ferri. Ciarallo, A., and E. De Carolis 1999. Homo faber: natura, scienza
Carandini, A., M. Medri, M. L. Gualandi, and E. Papi e tecnica nell’antica Pompeii. Milan: Electa.
1986. “Pendici settentrionali del Palatino.” BullCom 91.2: Cifani, G. 1994. “Aspetti dell’edilizia romana arcaica.” StEtr
429–38. 60: 185–226.
Carettoni, G. 1949. “Costruzioni sotto l’angolo sud- Cima, M. 1995. “Gli horti Liciniani e le statue dei Magistrati,”
occidentale della Domus Flavia (triclinio e ninfeo occiden- In M. Cima, ed. Restauri nei Musei Capitolini (Venice): 53–
tale).” NSc 74: 48–79. 69.
Carettoni, G. 1956–1959. “Le gallerie e ipogee del foro Ro- Clark, G., S. Coccia, J. A. Giorgi, L. Romeo, F. Scoppola, and
mano e i ludi gladiatori forensi.” BullCom 76: 23–44. D. N. Smith 1989. “L’indagine archeologica nell’elevato –
Carettoni, G. 1971. “Scoperte avvenute in occasione di lavori Palazzo Altemps (Roma).” Archeologia medievale 16: 289–
di restauro al palazzo imperiale.” NSc 96: 300–26. 328.
Carettoni, G. 1972. “Terme di Settimo Severo e terme di Clayton, P. A., and M. J. Price, eds. 1988. The Seven Wonders
Massenzio ‘in Palatio’.” ArchCl 24: 96–104. of the Ancient World. London: Routledge.
Carreras Monfort, C., and P. P. A. Funari 1998. Britannia y Coarelli, F. 1977. “Public Building in Rome between the
el Mediterráneo. Estudios sobre el abastecimiento de aceite bético Second Punic War and Sulla.” PBSR 45: 1–23.
y africano en Britannia. Barcelona: Publicions Universitat de Coarelli, F. 1981. Dintorni di Roma. Bari: Laterza.
Barcelona. Coarelli, F. 2001. “Gli anfiteatri a Roma prima del Colosseo.”
Cassatella, A. 1990. “Edifici palatini nella Domus Flavia.” Bol- In A. La Regina, ed. Sangue e arena (Milan): 43–7.
lettino di archeologia 3: 91–104. Coates-Stephens, R. 2003. “Attitudes to Spolia in Some
Cassatella, A., and I. Iacopi 1991. “Il balneum presso le Scalae Late Antique Texts.” In L. Lavan and W. Bowden, eds.
Caci sul Palatino.” In Les thermes romaines. Actes de la table Theory and Practice in Late Antique Archaeology (Leiden):
rondes organisée par l’Ecole francaise de Rome (Collection de 341–58.
l’Ecole francaise de Rome 142, Rome): 129–38. Coccia, S., and A. G. Fabiani 2003. “Le indagini archeo-
Cassieri, N., and G. Ghini 1990. “La cosiddetta Villa degli An- logiche recenti,” La Basilica di Massenzio: ricerca interdisci-
tonini al XVIII miglio della via Appia.” Archeologia Laziale plinare applicata allo studio e alla conservazione di un monumento:
10.1: 168–78. Atti del Convegno (Rome): 30–51.
Casson, L. 1965. “Harbour and River Boats of Ancient Colini, A. M. 1944. Storia e topografia del Celio nell’antichità.
Rome.” JRS 55: 31–9. Rome: MemPontAcc (ser. 3) 7.
253
P1: KIC
WORKS CITED
Colini, A. M. 1947. “Officina di fabri tignuarii nei frammenti De Angelis d’Ossat, Gu. 1933. “L’Aula del Planetario ed un
di un’ara monumentale rinvenuti fra il Campidoglio e il disegno di Baldassarre Peruzzi.” Capitolium 11: 12.
Tevere.” Capitolium 22: 21–8. De Angelis d’Ossat, Gu. 1938. “Le origini romane della cupola
Colini, A. M. 1955. “Horti Spei Veteris, Palatium Sessori- bizantina.” AttiCStR 4: 136–46.
anum.” MemPontAcc (ser. 3) 8: 137–77. De Angelis d’Ossat, Gu. 1938 (pub. 1940). “La forma e la
Colini, A. M., and I. Gismondi 1926. “Contributi allo studio costruzione delle cupole nell’architettura romana.” Atti del
del Pantheon.” BullCom 54: 67–92. Congresso nazionale di storia dell’architettura 3: 223–50.
Colli, D. 1996. “Il palazzo Sessoriano nell’area archeologica di De Angelis d’Ossat, Gu. 1946. Romanità delle cupole paleocris-
S. Croce in Gerusalemme: ultime sede imperiale a Roma?” tiane 5. Spoleto: Istituto di studi romani.
MÉFRA 108: 771–815. De Casa, G., G. Lombardi, C. Meucci, R. Galloni, and P.
Colosimo, R. 1938 (pub. 1940). “Il serbatoio romano di Chi- Vitali 1999. “Il tufo lionato dei monumenti romani: carat-
eti.” Atti del Congresso nazionale di storia dell’architettura 3: teri petrografici, geomeccanici e trattamenti conservati.”
251–62. Geologica romana 35: 1–25.
Conforto, M. L. 1991. “Terme di Caracalla. Dati acquisiti ed de Fine Licht, K. 1968. The Rotunda in Rome. A Study of
ipotesi di ricerca.” In Les thermes romaines. Actes de la table Hadrian’s Pantheon. Jutland Archeological Society Publica-
ronde organisée par l’Ecole francaise de Rome, 11–12 novembre tions 8. Copenhagen: Gyldendal.
1988 (Rome): 43–8. de Fine Licht, K. 1974. Untersuchungen an den Trajansthermen
Coulton, J. J. 1977. Ancient Greek Architects at Work. Ithaca, zu Rom. Copenhagen: E. Munksgaard.
N.Y.: Cornell University. de Fine Licht, K. 1990. Sette Sale. Rome: AnalRom Supp. 19.
Cozza, L. 1974–1975. “I recenti scavi delle Sette Sale.” Rend- De Magistris, E. 1995. “Appunti per una lettura della Porta
PontAcc 47: 79–101. Rosa a Velia.” Quaderni del Dipartimento della Scienze
Cozza, L. 1982. Tempio di Adriano. LSA 1. Rome: De Luca. dell’Antichità, Università di Salerno 16: 87–93.
Cozza, L. 1983. “Le tegole di marmo del Pantheon.” In K. de Robertis, F. M. 1938. Il diritto associative romano dei collegi
de Fine Licht, ed. Città e architettura nella Roma imperiale della repubblica alle corporazioni del basso impero. Bari: Laterza.
(AnalRom Supp. 10, Odense): 109–18. De Romanis, A. 1822. Le antiche camere esquiline dette comune-
Cozza, L. 1986. “Mura Aureliane, 1 – Trastevere, il braccio mente delle Terme di Tito. Rome: n.p.
settentrionale: dal Tevere a porta Aurelia – San Pancrazio.” de’Ficoroni, F. 1744. Le vestigia e rarità di Roma antica. Rome:
BullCom 91.1: 101–30. G. Mainardi.
Cozza, L. 1987. “Osservazioni sulle mura aureliane a Roma.” Degrassi, A. 1969. “Epigraphica IV: 1. Quando fu costruito il
AnalRom 16: 25–52. santuario della Fortuna Primigenia di Palestrina.” MemLinc
Cozzo, G. 1928. Ingegneria romana. Rome: Libreria Editrice (ser. 8) 14: 111–27.
Mantegazza di P. Cremonese. Deichmann, F. W., and A. Tschira 1957. “Das Mausoleum der
Cuomo, S. 2000. Pappus of Alexandria and the Mathematics of Kaiserin Helena und die Basilika der Heiligen Marcellinus
Late Antiquity. Cambridge: Cambridge University Press. und Petrus an der Via Labicana vor Rom.” JdI 72: 44–110.
Curle, J. 1911. A Roman Frontier Post and Its People: The Fort of DeLaine, J. 1985. “An Engineering Approach to Roman
Newstead in the Parrish of Melrose. Glasgow: James Maclehose Building Techniques: The Baths of Caracalla in Rome.”
and Sons. In C. Malone and S. Stoddart, eds. Papers in Italian Archae-
Daguet-Gagey, A. 1997. Les opera publica à Rome (180–305 ap. ology IV, Part iv: Classical and Medieval Archaeology (BAR IS
J.C.). Paris: Institut d’Études Augustiniennes. 246, Oxford): 195–206.
D’Arms, J. H. 1981. Commerce and Social Standing in Ancient DeLaine, J. 1987. “The ‘Cella Solearis’ of the Baths of Cara-
Rome. Cambridge, Mass.: Harvard University Press. calla: A Reappraisal.” PBSR 55: 147–56.
Davey, N. 1974. “Roman Concrete and Mortar.” The Struc- DeLaine, J. 1990. “Structural Experimentation: The Lintel
tural Engineer 52.6: 193–5. Arch, Corbel, and Tie in Western Roman Architecture.”
De Angelis d’Ossat, Gio. 1930. “Roccie adoperate nella WorldArch 21.3: 407–24.
cupola del Pantheon.” Atti della Pontificia accademia della DeLaine, J. 1995. “The Supply of Building Materials to the
scienze nuovi Lincei 83: 211–15. City of Rome.” In N. Christie, ed. Settlement and Economy in
De Angelis d’Ossat, Gio. 1945. “Rocce messe in opera nella Italy 1500 BC to AD 1500 (Oxbow Monograph 41, Oxford):
cupola di ‘Minerva Medica’.” BCSSA 4: 3–6. 555–62.
254
P1: KIC
WORKS CITED
DeLaine, J. 1997. The Baths of Caracalla: A Study in the Design, Methodik und Erkenntnismöglichkeit der Kaiserzeitlichen
Construction, and Economics of Large-Scale Building Projects in Prosopographie: Kolloquium Köln 24, Vienna): 365–96.
Imperial Rome. Portsmouth, R. I.: JRA Supp. 25. Ellis, S. P. 1988. “The End of the Roman House.” AJA 92:
DeLaine, J. 2000. “Building the Eternal City: The Construc- 565–76.
tion Industry of Imperial Rome.” In J. Coulston and H. Esposito, F., and A. Michetti 1996. “I criteri di dimension-
Dodge, eds. Ancient Rome: The Archaeology of the Eternal City amento degli organismi a cupola presso i romani (III).”
(Oxford): 119–41. Materiali e Strutture 6.2: 61–84.
DeLaine, J. 2003. “The Builders of Roman Ostia: Organiza- Fabbrini, L. 1982. “Domus Aurea – Il piano superiore del
tion, Status and Society.” In S. Huerta, ed. Proceedings of the quartiere orientale.” MemPontAcc (ser. 3) 14: 5–24.
First International Congress on Construction History 2 (Madrid): Fabbrini, L. 1985–1986. “I corpi edilizi che condizionarono
723–32. l’attuazione del progetto del palazzo esquilino di Nerone.”
DeLaine, J., S. Gibson, and A. Claridge 1994. “The Triclin- RendPontAcc 58: 129–79.
ium of the Domus Flavia: A New Reconstruction.” PBSR Fabbrini, L. 1995. “Domus Aurea: il palazzo sull’Esquilino.”
62: 67–97. In E. M. Steinby, ed. Lexicon Topographicum Urbis Romae 2:
Delbrück, R. 1907. Hellenische Bauten in Latium 1. Strassburg: 56–63.
K. J. Trübner. Fant, J. C. 1989. Cavum Antrum Phrygiae: The Organization and
Delbrück, R. 1921. “Der Südostbau am Forum Romanum.” Operations of the Roman Imperial Marble Quarries in Phrygia.
JdI 36: 8–33. Oxford: BAR IS 482.
D’Elia, L., and S. L. P. Buranelli 1985. “Relievo del peristilio Fasolo, F., and G. Gullini 1953. Il santuario della Fortuna Primi-
inferiore della Domus Flavia.” In Roma: Archeologia nel centro genia a Palestrina. Rome: Istituto di archeologia, Università
I, L’area archeologica centrale (LSA 6.1, Rome): 176–8. di Roma.
Dinsmoor, W. B. 1922. “Structural Iron in Greek Architec- Ferretti, A. S. 1997. “Proposte per lo studio teoritico-
ture.” AJA 26: 148–58. sperimentale della statica dei monumenti in opus caemen-
Dobson, E. 1849. Rudiments of the Art of Building. London: ticium.” Materiali e strutture 7.2–3: 63–84.
John Weale. Fidenzoni, P. 1970. Il teatro di Marcello. Rome: Liber.
Donati, A. 1998. Romana Pictura: la pittura romana dalle origini Finkel, I. L. 1988. “The Hanging Gardens of Babylon.” In
all’età bizantina. Milan: Electa. P. A. Clayton and M. J. Price, eds. The Seven Wonders of the
Downey, G. 1948. “Byzantine Architects: Their Training and Ancient World (London): 38–58.
Methods.” Byzantion 18: 99–118. Finsen, H. 1962. Domus Flavia sur le Palatin: Aula Regia-
Drachmann, A. G. 1963. The Mechanical Technology of Greek and Basilica. Copenhagen: Munksgaard.
Roman Antiquity: A Study of the Literary Sources. Madison, Finsen, H. 1969. La résidence de Domitien sur le Palatin.
Wisconsin: University of Wisconsin Press. Copenhagen: Munksgaard.
Dressel, H. 1879. “Di un grande deposito di anfore rinvenuto Fitchen, J. 1961. The Construction of Gothic Cathedrals: A Study
nel nuovo quartiere del Castro Pretorio.” BullCom 7: 36– of Medieval Vault Erection. Oxford: Clarendon Press.
112. Foce, F., and D. Aita 2003. “The Masonry Arch between
Drijvers, J. W. 1992. Helena Augusta: The Mother of Constantine ‘limit’ and ‘elastic’ analysis. A Critical Re-Examination of
the Great and the Legend of Her Finding of the True Cross. New Durand-Claye’s Method.” In S. Huerta, ed. Proceedings of the
York: E. J. Brill. First International Congress on Construction History 2 (Madrid):
Duncan-Jones, R. P. 1994. Money and Government in the Roman 895–908.
Empire. Cambridge: Cambridge University Press. Fontana, C. 1694. Templum Vaticanum et ipsius origo. Rome:
Durm, J. 1905. Handbuch der Architektur: Die Baukunst der Francisci Buagni.
Etrusker und der Römer. Stuttgart: A. Kröner. Fontana, S. 1995. “Un impianto per la produzione di calce
Eck, W. 1992. “Cura viarum und cura operum publico- presso Lucus Feroniae.” In N. Christie, ed. Settlement
rum als kollegiale Ämter im frühen Prinzipat.” Klio 74: and Economy in Italy 1500 BC to AD 1500 (Oxford):
237–45. 563–70.
Eck, W. 1993. “Überlieferung und historische Realität: Foss, J. E. 1988. “Paleosols of Pompeii and Oplontis.” In R. I.
Ein Grundproblem prosopographischer Forschung.” In W. Curtis, ed. Studia Pompeiana & Classica in Honor of Wil-
Eck, ed. Prosopographie und Sozialgeschichte (Studien zur helmina F. Jashemski 1 (New York): 127–44.
255
P1: KIC
WORKS CITED
Frank, T. 1924. Roman Buildings of the Republic: An Attempt to Gerkan, A. von 1951. “St. Gereon in Köln.” Germania 29:
Date Them from Their Materials. Rome: American Academy 215–18.
in Rome. Ghini, G. 1988. “Prime indagini archeologiche.” In Anfiteatro
Frazer, A. 1969. “The Porch of the Tor de’Schiavi at Rome.” Flavio: immagine, testimonianze, spettacoli (Rome): 101–5.
AJA 73: 45–8. Giachi, G., and P. Pallecchi 2000. “Analisi preliminari sui ma-
Frazer, A. K. 1966. “The Iconography of the Emperor Max- teriali.” In S. Bruni, ed. Le navi antiche di Pisa: ad un anno
entius’ Buildings in Via Appia.” ArtB 47: 385–92. dall’inizio delle ricerche (Florence): 348–51.
Gaitzsch, W. 1980. Eiserne römische Werkzeuge: Studien zur Giovannoni, G. 1904. “La sala termale della Villa Liciniana e
römischen Werkzeugkunde in Italien und den nördlichen Prov- le cupole romane.” Annali della Società degli Ingegneri e degli
inzen des Imperium Romanum. Oxford: BAR IS 78. Architetti Italiani 3: 3–39.
Galliazzo, V. 1995. I ponti romani 1. Treviso: Canova. Giovannoni, G. 1925. La tecnica della costruzione presso i romani.
Ganzert, J. 1985. “Der Mars-Ultor Tempel auf dem Augus- Rome: Società editrice d’arte illustrata.
tusforum in Rom. Vorläufiger Arbeits bericht.” RM 92: Giovannoni, G. 1931. “Contributi allo studio della technica
201–19. nelle costruzione romane.” AttiCStR 2: 281–94.
Ganzert, J., and V. Kockel 1988. “Augustusforum und Mars- Giovannoni, G. 1938. “La cupola di S. Costanza e le volte
Ultor Tempel.” In Kaiser Augustus und die verlorene Republik romane a struttura leggera.” AttiCStR 4.2: 211–16.
(Rome): 149–99. Giovannoni, G. 1938 (pub. 1940). “La basilica dei Flavi sul
Garello, F. 1991. “Gli ipogei dell’Anfiteatro Flavio nell’analisi Palatino.” Atti del Congresso nazionale di storia dell’architettura
delle strutture murarie: l’impianto di etá flavia.” Meded 50: 3: 85–94.
173–95. Giuliani, C. F. 1970. Tibur I (Forma Italiae I.7). Rome: Danesi.
Garnsey, P. 1981. “Independent Freedmen and the Economy Giuliani, C. F. 1975. “Volte e cupole a doppia calotta in età
of Roman Italy under the Principate.” Klio 63: 359–71. adriana.” RM 82: 329–42.
Gatti, G. 1934. “Saepta Julia e Porticus Aemilia nella Forma Giuliani, C. F. 1977. “Domus Flavia: una nuova lettura.” RM
Severiana.” BullCom 62: 123–49. 84: 91–106.
Gatti, G. 1938 (pub. 1940). “Il Portico degli Argonauti e la Giuliani, C. F. 1990. L’edilizia nell’antichità. Rome: La nuova
Basilica di Nettuno.” Atti del Congresso nazionale di storia Italia scientifica.
dell’architettura 3: 61–74. Giuliani, C. F. 1992. “Opus signinum e cocciopesto.” Segni
Gazda, E. K., and A. M. McCann 1987. “Chapter VII. Re- (Serie storia antica e archeologia): 89–94.
construction and Function: Port, Fishery, and Villa.” In Giuliano, A., C. F. Giuliani, M. L. Veloccia Rinaldi, M. Lollo-
A. M. McCann, ed. The Roman Port and Fishery at Cosa: A Ghetti, and C. Angelini 1988. Villa Adriana. Rome: Silvana
Center of Roman Trade (Princeton): 137–59. Editoriale.
Gazda, E. K. 1987. “Chapter 4. The Port and Fishery: De- Goodman, W. L. 1964. The History of Woodworking Tools.
scription of the Extant Remains and Sequence of Construc- London: G. Bell.
tion.” In A. M. McCann, ed. The Roman Port and Fishery Gordon, A. E. 1952. “Quintus Veranius, Consul AD 49.”
of Cosa (Princeton): 74–97. University of California Publications in Classical Archaeology 2.5:
Gazda, E. K. 2001. “Cosa’s Contribution to the Study of Ro- 231–341.
man Hydraulic Concrete: An Historiographical Commen- Gordon, J. E. 1976. The New Science of Strong Materials or Why
tary.” In N. W. Goldman, ed. New Light from Ancient Cosa You Don’t Fall Through the Floor. 2nd ed. Harmondsworth:
(New York): 145–77. Penguin.
Gerding, H. 2002. The Tomb of Caecilia Metella: Tumulus, Graham, S. 2002. “Ex Figlinis: The Complex Dynamics of the
Tropaeum and Trophy. Lund: Lund University. Roman Brick Industry in the Tiber Valley during the 1st
Gerhardt, R., K. E. Kurrer, and G. Pichler 2003. “The Meth- to 3rd Centuries AD.” Ph.D. diss., University of Reading.
ods of Graphical Statics and Their Relation to the Struc- Guidobaldi, F. 1992. San Clemente: gli edifici romani, la basil-
tural Form.” In S. Huerta, ed. Proceedings of the First Inter- ica paleocristiana e le fasi alto medievali. Rome: Apud S.
national Congress on Construction History 2 (Madrid): 997– Clementem.
1006. Guidobaldi, F. 1995. “Sull’originalità dell’architettura di età
Gerkan, A. von 1925. “Das Obergeschoss des flavischen Am- costantiniana.” Corsi di Cultura sull’arte ravennate e bizantina
phitheaters.” RM 40: 11–50. 42: 419–41.
256
P1: KIC
WORKS CITED
Guidobaldi, F. 1999. “Le domus tardoantiche di Roma come Heyman, J. 1995. The Stone Skeleton: Structural Engineering
‘sensori’ delle trasformazioni culturali e sociali.” In W. V. of Masonry Architecture. Cambridge: Cambridge University
Harris, ed. The Transformations of Urbs Roma in Late Antiquity Press.
( JRA Supp. 33, Portsmouth, R. I.): 53–68. Hoffmann, A. 1980. Das Gartenstadion in der Villa Hadriana.
Gullini, G. 1954. “I monumenti dell’acropoli di Ferentino.” Deutsches Archaeologisches Institut Rom, Sonderschriften
ArchCl 6: 185–216. IV. Mainz: Philipp von Zabern.
Gullini, G. 1973. “La datazione e l’inquadramento stilistico del Hooke, R. 1679. Lectiones Cutlerinae, or, A collection of Lectures:
santuario della Fortuna Premigenia.” ANRW 1.4: 746–99. Physical, Mechanical, Geographical,and astronomical. London:
Guyon, J. 1986. “Dal praedium imperiale al santuario dei mar- John Martyn.
tiri. Il territorio ‘ad duas lauros’.” In A. Giardina, ed. Società Hopkins, K. 1983. “Models, Ships, and Staples.” In P. Gar-
romana e impero tardo antico: Roma politica, economia, paessagio nsey and C. R. Whittaker, eds. Trade and Famine in Classical
urbano 2 (Bari): 300–32. Antiquity (Cambridge): 84–109.
Guyon, J. 1987. Le cimetière aux deux lauriers. Vatican City: Hülsen, C. 1889. “Jahresbericht über neue Funde und
Pontificio Istituto di archeologia cristiana. Forschungen zur Topographie der Stadt Rom 1887–1889.”
Harrison, T. A. 1995. Formwork Striking Times: Criteria, Pre- RM 4: 227–91.
diction and Methods of Assessment. London: Construction In- Hülsen, C. 1910. Die Thermen des Agrippa: Ein Beitrag zur To-
dustry Research and Information Association, Report 136. pographie des Marsfeldes in Rom. Rome: Verlag von Loescher.
Haselberger, L. 1994. “Ein Giebelriss der Vorhalle des Hurst, H. 1986. “Area di S. Maria Antiqua: la ricerca arche-
Pantheon: Die Werkrisse vor dem Augustusmausoleum.” ologica.” BullCom 91.2: 470–8.
RM 101: 279–308. Ioppolo, G., and G. Pisani Sartorio 1999. La Villa di Massenzio
Healy, J. F. 1978. Mining and Metallurgy in the Greek and Roman sulla Via Appia: Il Circo. Rome: Colombo.
World. London: Thames and Hudson. Jashemski, W. F. 1979. The Gardens of Pompeii: Herculaneum
Heath, T. 1921. A History of Greek Mathematics. Oxford: and the Villas Destroyed by Vesuvius 1. New Rochelle, N.Y.:
Clarendon Press. Caratzas Brothers.
Heene, G. 2004. Baustelle Pantheon. Dusseldorf: Verlag Jones, A. H. M. 1950. “The Aerarium and the Fiscus.” JRS
Bau+Technik. 40: 22–9.
Heilmeyer, W.-D. 1975. “Apollodorus von Damaskus, der Ar- Jones, A. H. M. 1964. The Later Roman Empire 284–602: A So-
chitekt des Pantheon.” RM 90: 316–47. cial, Economic, and Administrative Survey. Oxford: Blackwell.
Helen, T. 1975. The Organization of Roman Brick Production in Kawamoto, T., and Y. Tatsumi 1992. “Classification and Re-
the 1st and 2nd centuries AD: An Interpretation of Roman Brick gional Distribution of Lava Blocks in Pompeii.” Opuscula
Stamps. Helsinki: Suomalainen Tiedeakatemia. Pompeiana 2: 92–7.
Hemsoll, D. 1990. “The Architecture of Nero’s Golden Keay, S. 1984. Late Roman Amphorae in the Western Mediter-
House.” In M. Henig, ed. Architecture and Architectural Sculp- ranean. A Typology and Economic Study: The Catalan Evidence.
ture (Oxford): 10–38. Oxford: BAR IS 136.
Heres, T. L. 1983. Paries. A Proposal for a Dating System of Late- Köhler, J. 1999. “Die Terme Taurine bei Civitavecchia. Pub-
Antique Masonry Structures in Rome and Ostia. Amsterdam: likationsstand – Chronologie – Bibliothek.” RM 106: 365–
Rodopi. 76.
Herrmann, J. J. 1976. “Observations on the Baths of Maxen- Kolb, A. 1993. Die kaiserliche Bauverwaltung in der Stadt Rom:
tius in the Palace.” RM 83: 403–24. Geschichte und Aufbau der cura operum publicorum unter
Hesberg, H. von 1978–1980. “Zur datierung des Theaters dem PrinzipatfRapp. Stuttgart: F. Steiner.
in der Domitians villa von Castel Gandolfo.” RendPontAcc Krause, C. 1985. “Ricerche sulla storia edilizia: rapporto pre-
51–2: 305–24. liminare.” In Domus Tiberiana: Nuove ricerche – studi di
Hesnard, A. 1980. “Un dépot augustéen d’amphores à la Lon- restauro (Zurich): 73–136.
garina, Ostia.” In J. H. D’Arms and E. C. Kopff, eds. The Krause, C. 1986. “Domus Tiberiana – Resulti degli scavi
Seaborne Commerce of Ancient Rome: Sudies in Archaeology and 1984–1985.” BullCom 91.2: 442–63.
History (MAAR 36, Rome): 141–56. Krause, C. 1994. Domus Tiberiana 1: gli scavi. Bollettino di arche-
Heyman, J. 1988. “Poleni’s Problem.” Proceedings of the Insti- ologia 25–27. Rome: Libreria dello Stato.
tution of Civil Engineers 84: 737–59.
257
P1: KIC
WORKS CITED
Krautheimer, R. 1964. “Zu Konstantins Apostelkirche in Levick, B. 1999. Vespasian. London: Routledge.
Konstantinopel.” In A. Stuiber and A. Hermann, eds. Lombardi, L., and A. Corazza 1995. Le Terme di Caracalla.
Mullus: Festscrift Theodor Klauser (Münster): 224–9. Rome: Fratelli Palombi.
Krautheimer, R. 1993. “The Ecclesiastical Building Policy Luciani, R. 1985. “Il criptoportico del Palatino.” In R. Lu-
of Constantine.” In G. Bonamente and F. Fusce, eds. ciani, ed. Roma Sotterranea (Rome): 148–55.
Costantino il Grande dall’antichità all’Umanismo. Colloquio sul Luciani, R. 1993. Il Colosseo. Milan: Fenice 2000.
Cristianismo nel mondo antico 2 (Macerata): 509–52. Lugli, G. 1915. “La villa dei Gordiani e i monumenti al III
Krautheimer, R., S. Corbett, and A. K. Frazer 1980. Corpus miglio della via Prenestina.” BullCom 43: 136–67.
Basilicarum Christianarum Romae 5. Vatican City: Pontificio Lugli, G. 1918. “La villa di Domiziano sui colli Albani.” Bull-
Istituto di archeologia cristiana. Com 46: 3–68.
La Follette, L. 1994. “The Baths of Trajan Decius on the Lugli, G. 1926. Ager Pomptinus-Anxur-Tarracina (Forma Italiae
Aventine.” In Rome Papers ( JRA Supp. 11, Portsmouth, I.1). Rome: Danesi.
R. I.): 6–88. Lugli, G. 1929–1930. “I Mercati Traianei.” Dedalo 10: 527–51.
La Regina, A., ed. 2001. Sangue e arena. Milan: Electa. Lugli, G. 1933. “Pantheon.” BullCom 61: 273–4.
Lamprecht, H.-O. 1987. Opus caementicium: Bautechnik der Lugli, G. 1938. Monumenti antichi di Roma e suburbio 3. Rome:
Römer. Düsseldorf: Beton-Verlag. G. Bardi.
Lancaster, L. C. 1995a. “Concrete Vaulted Construction: De- Lugli, G. 1946. Roma antica: il centro monumentale. Rome:
velopments in Rome from Nero to Trajan.” D.Phil. diss., G. Bardi.
Oxford University. Lugli, G. 1954. “Il santuario della Fortuna Primigenia in
Lancaster, L. C. 1995b. “The Date of Trajan’s Markets: An As- Praeneste e la sua datazione.” RendLinc (ser. 8) 9.3–4: 51–87.
sessment in the Light of Some Unpublished Brick Stamps.” Lugli, G. 1957. La tecnica edilizia romana con particolare riguardo
PBSR 63: 25–44. a Roma e Lazio. Rome: G. Bardi.
Lancaster, L. C. 1998a. “Building Trajan’s Markets.” AJA 102: Lugli, G., and G. Filibeck 1935. Il porto di Roma imperiale e
283–308. l’agro portuense. Rome: n.p.
Lancaster, L. C. 1998b. “Reconstructing the Restoration of Lupu, N. 1937. “La villa di Sette Bassi sulla via Latina.” EphDac
the Colosseum after the Fire of 217.” JRA 11: 146–74. 7: 117–88.
Lancaster, L. C. 1999. “Building Trajan’s Column.” AJA 103: Luschi, L. 1989–1990. “Gli stucchi della villa detta ‘dei Gor-
419–39. diani’ sulla via Prenestina.” BullCom 93: 407–46.
Lancaster, L. C. 2000. “Building Trajan’s Markets 2: The Con- MacCormack, S. G. 1981. Art and Ceremony in Late Antiquity.
struction Process.” AJA 104: 755–85. Berkeley: University of California Press.
Lanciani, R. 1897. The Ruins and Excavations of Ancient Rome. MacDonald, W. L. 1982. The Architecture of the Roman Em-
London: Macmillan. pire I: An Introductory Study. 2nd ed. rev. New Haven: Yale
Laubenheimer, F. 1998. “L’eau et les amphores. Les systèmes University Press.
d’assainissement en Gaule romaine.” In S. Pesavento Mat- MacDonald, W. L., and J. Pinto 1995. Hadrian’s Villa and Its
tioli, ed. Bonifiche e drenaggi con anfore in epoca romana: aspetti Legacy. New Haven: Yale University Press.
tecnici e topografici (Modena): 47–70. MacMullen, R. 1964. “Social Mobility and the Theodosian
Lea, F. M. 1970. The Chemistry of Cement and Concrete. 3rd ed. Code.” JRS 54: 49–53.
London: Edward Arnold. Maddin, R., A. Hauptmann, and D. Baatz 1991. “A Metal-
Leacroft, H., and R. Leacroft 1969. The Buildings of Ancient lographic Examination of Some Iron Tools from the Saal-
Rome. Reading, Mass.: Addison-Wesley. burgmuseum.” SaalbJb 46: 5–23.
Lechtman, H. N., and L. W. Hobbs 1987. “Roman Concrete Maggiani, A. 1981. “Nuove tendenze archeologiche all’isola
and the Roman Architectural Revolution.” Ceramics and d’Elba: i rinvenimenti di età classica e ellenistica.” In
Civilization 3: 81–128. L’Etruria mineria Atti del XII Convegno di Studi Etruschi e
Leger, A. 1875. Les travaux publics, les mines et mètallurgie aux Italici (Florence): 173–92.
temps des Romains. Paris: J. Dejey & Cie. Mainstone, R. J. 1975. Developments in Structural Form. Cam-
Lehmann, K. 1945. “The Dome of Heaven.” ArtB 27: 1–27. bridge, Mass.: M.I.T. Press.
Lehmann-Hartleben, K., and J. Lindros 1935. “Il palazzo degli Mainstone, R. J. 1986. “On the Structure of the Roman Pan-
Orti Sallustiani.” OpArch 1: 196–227. theon.” ArtB 68.4: 673–4.
258
P1: KIC
WORKS CITED
Mainstone, R. J. 1988. Hagia Sophia – Architecture, Structure Meighörmer-Schardt, W. 1990. “Zur Rekonstruktion eines
and Liturgy of Justinian’s Great Church. London: Thames and römischen Bockkranes.” Journal of Roman Military Equip-
Hudson. ment Studies 1: 43–60.
Mainstone, R. J. 1992. “Questioning Hagia Sophia.” In R. Meneghini, R. 1989. “Ricerche nel Foro di Traiano – Basilica
Mark and A. S. Çakmak, eds. Hagia Sophia From the Age of Ulpia: un esempio di sopravivvenza di strutture antiche in
Justinian to the Present (Cambridge): 158–76. età medievale.” Archeologia medievale 16: 541–57.
Malinowski, R. 1979. “Concretes and Mortars in Ancient Meneghini, R. 1993. “Nuovi dati sulle biblioteche e il Tem-
Aqueducts.” Concrete International ( Jan): 66–76. plum Divi Traiani nel Foro di Traiano.” Bollettino di arche-
Manacorda, D. 1977. “Anfore.” In A. Carandini and ologia 19–21: 13–21.
C. Panella, eds. Ostia I: Le terme del Nuotatore: Scavo Meneghini, R. 2001. “Il foro di Traiano. Ricostruzione ar-
dell’ambiente XVI e dell’area XXV (Studi Miscellanei 23, chitettonica e analisi strutturale.” RM 108: 245–69.
Rome): 277–83. Mercer, H. C. 1929. Ancient Carpenter’s Tools. Doylestown,
Mancioli, D., A. Ceccherelli, and R. Santangeli Valen- Pa.: The Bucks County Historical Society.
zani 1993. “Domus Parthorum.” Archeologia Laziale 11.2: Messineo, G. 1989. Malborghetto. LSA 15. Rome: De Luca.
53–8. Mezzena, R. M. 1981. “Augusta Praetoria ed il suo territo-
Mango, C. 1992. “Byzantine Writers on the Fabric of the rio.” In Archeologia in Valle d’Aosta dal neolitico alla caduta
Hagia Sophia.” In R. Mark and A. S. Çakmak, eds. Hagia dell’impero romano 3500 a.C.–V sec. d.C. (Aosta): 63–138.
Sophia from the Age of Justinian to the Present (Cambridge): Middleton, J. H. 1892a. The Remains of Ancient Rome 1.
41–56. London: A. and C. Black.
Manieri Elia, G. 1999. “Stabilità delle strutture e con- Middleton, J. H. 1892b. The Remains of Ancient Rome 2.
servazione.” In J. Charles-Gaffiot and H. Lavagne, eds. London: A. and C. Black.
Adriano: architettura e progetto (Milan): 175–81. Milani, G. B. 1923. L’ossatura murale. Parte I. La stabilità. Turin:
Mari, Z. 1983. Tibur III (Forma Italiae I.17). Rome: Danesi. C. Crudo.
Mari, Z., and M. G. Fiore 2005. “Per Ercole vincitore.” Archeo Milella, M., L. Messi, P. Pensabene, G. Piazzesi, B. M.
21.1: 32–41. Tummarello, and L. Ungaro 1989. “Foro Traiano. Con-
Mark, R., and A. S. Çakmak 1992. Hagia Sophia from the Age tribuiti per una ricostruzione storica e architettonica.”
of Justinian to the Present. Cambridge: Cambridge University ArchCl 41: 27–291.
Press. Millar, F. G. B. 1963. “The Fiscus in the First Two Centuries.”
Mark, R., and P. Hutchinson 1986. “On the Structure of the JRS 53: 28–42.
Pantheon.” ArtB 68: 24–34. Minoprio, A. 1932. “A Restoration of the Basilica of Con-
Martin, R. 1965. Manuel d’architecture grecque vol. I, Matériaux stantine.” PBSR 12: 1–25.
et techniques. Paris: J. Picard. Monneret de Villard, U. 1924. “Sull’impiego di vasi e tubi
Martin, S. D. 1989. The Roman Jurists and the Organization of fittili nella costruzione delle volte.” In C. Hülsen, C. Cec-
Private Building in the Late Republic and Early Empire. Brussels: chelli, G. Giovannoni, U. Monneret de Villard, and A.
Collection Latomus 204. Muñoz-Alonso, Sant’Agata dei Goti (Rome): 149–54.
Martines, G. 1991. “Argomenti di geometria antica a propos- More, J. H. 1971. “The Fabri Tignuarii of Rome.” HSCP
ito della cupola del Pantheon.” Quaderni dell’Istituto di storia 75: 202–5.
dell’architettura 13: 3–18. Napoli, M. 1965. “La attività archeologica nelle province di
Massaccesi, V. 1939. “I restauri di Settimio Severo e Caracalla Avellino, Benevento, e Salerno.” AttiTaranto 5: 193–211.
agli edifici Palatino.” BullCom 67: 117–33. Napoli, M. 1966. “La ricerca archeologica di Velia.” PP 21:
Mattingly, D. J. 1988. “Oil for Export? A Comparison of 191–226.
Libyan, Spanish, and Tunisian Olive Oil Production in the Nappi, G., B. Cappacioni, M. Mattioli, E. Mancini, and L.
Roman Empire.” JRA 1: 33–56. Valentini 1994. “Plinian Fall Deposits from Vulsini Volcanic
Mazzucato, O. 1970. “Ceramica medievali nell’edilizia District (Central Italy).” Bulletin of Volcanology 56: 502–15.
laziale.” In Atti del Convegno Internazionale della Ceramica Nielsen, I. 1993. “Castor, Aedes, Templum.” In E. M. Steinby,
(Albisole): 339–70. ed. Lexicon Topographicum Urbis Romae 1 (Rome): 242–5.
Meiggs, R. 1982. Trees and Timber in the Ancient Mediterranean Nielsen, I., and B. Poulsen, eds. 1992. The Temple of Castor
World. Oxford: Clarendon Press. and Pollux I. LSA 17. Rome: De Luca.
259
P1: KIC
WORKS CITED
Nilson, A. H., and D. Darwin 1997. Design of Concrete Struc- Peacock, D. P. S. 1994. “Mons Claudianus and the Problem
tures. 12th ed. rev. New York: McGraw-Hill. of ‘granito del foro’: A Geological and Geochemical Ap-
Ochsendorf, J. A. 2002. “Collapse of Masonry Structures.” proach.” Antiquity 68: 209–30.
Ph.D. diss., Cambridge University. Peacock, D. P. S., and V. A. Maxfield 1997. Mons Claudianus:
O’Connor, C. 1993. Roman Bridges. Cambridge: Cambridge Survey and Excavation, 1987–1993. Cairo: Institute francais
University Press. d’archeologique orientale du Caire.
Olcese, G. 1993. “Archeologia e archeometria dei laterizi Peacock, D. P. S., and D. F. Williams 1986. Amphorae and the
bollati urbani: primi resultati e prospettive di ricerca.” In Roman Economy: An Introductory Guide. London: Longman.
W. V. Harris, ed. The Inscribed Economy ( JRA Supp. 6, Pearse, J. L. D. 1974. “The Organization of Roman Pub-
Portsmouth, R. I.): 121–7. lic Building during the Late Republic and Early Empire.”
Olcese, G. 1994. “Laterizi bollati da Roma, Villa dei Quin- Ph.D diss., Cambridge University.
tilli. Analisi chimiche e mineralogiche preliminari.” In F. Peña, J. T. 1999. The Urban Economy during the Early Dominate:
Burragato, O. Grubessi, and L. Lazzarini, eds. 1st European Pottery Evidence from the Palatine Hill. Oxford: BAR IS
Workshop on Archaeological Ceramics (Modena): 519–25. 784.
Oleson, J. P., C. Brandon, S. M. Cramer, R. Cucitore, E. Pensabene, P. 1994. Le vie del marmo: i blocchi di cava di Roma
Gotti, and R. L. Hohlfelder 2004. “The ROMACONS e di Ostia: il fenomeno del marmo nella Roma antica. Itinerari
Project: A Contribution to the Historical and Engineer- ostiensi 7. Rome: Ministero per i beni culturali e ambien-
ing Analysis of Hydraulic Concrete in Roman Maritime tali, Soprintendenza Archeologica di Ostia.
Structures.” IJNA 33.2: 199–229. Pensabene, P. 2001. “Criteri di reimpiego e nuove mode ar-
Olivier, A. 1983. “Sommiers de plates-bandes appareillées et chitettoniche nella basilica paleocristiana di Roma.” In M.
armées a Conimbriga et á la Villa d’Hadrien á Tivoli.” Cecchelli, ed. Materiali e techniche dell’edilizia paleocristiana a
MÉFRA 95.2: 937–59. Roma (Rome): 103–26.
Orlandos, A. K. 1968. Les matériaux de construction et la technique Pensabene, P. 2002. “Il fenomeno del marmo nel mondo ro-
architectural des anciens grecs 2. Translated by V. Hadjimichali. mano.” In M. De Nuccio and L. Ungaro, eds. I marmi antichi
Paris: École Française d’Athènes. della Roma imperiale (Venice): 3–67.
Ousterhout, R. 1999. Master Builders of Byzantium. Princeton, Pensabene, P., and C. Panella 1994–1995 (pub. 1998). “Riem-
N.J.: Princeton University Press. piego e progettazione architettonica nei monumenti tardo-
Packer, J. E. 1968–1969. “La casa di via Giulio Romano.” antichi di Roma II.” RendPontAcc 67: 25–67.
BullCom 81: 127–48. Penta, F. 1956. I materiali da costruzione del Lazio. Spoleto: Arti
Packer, J. E. 1997. The Forum of Trajan in Rome. A Study of the Grafiche Panetto e Petrelli.
Monuments. Berkeley: University of California Press. Pesavento Mattioli, S., ed. 1998. Bonifiche e drenaggi con anfore in
Packer, J. E. 2003. “TEMPLUM DIVI TRAIANI epoca romana: aspetti tecnici e topografici. Modena: F. C. Panini.
PARTHICI ET PLOTINAE: A Debate with Roberto Piazzesi, G. 1989. “Foro Traiano. Contributi per una ri-
Meneghini.” JRA 16.1: 109–36. costruzione storica e architettonica. Gli edifici: ipotesi ri-
Pagano, M. 1995–1996 (pub. 1998). “Testimonianze post 79 costruttive.” ArchCl 41: 125–98.
nell’area vesuviana.” Rivista di Studi Pompeiani 7: 35–44. Pietrangeli, C. 1978. Otricoli: un lembo dell’Umbria alle porte di
Panella, C. 1983. “I contenitori oleari presenti ad Ostia in etá Roma. Narni: Cassa di risparmio di Narni.
antonina: analisi tipologica, epigrafica, quantitiva.” In J. M. Piraino, C. 1996. “Anfiteatro Flavio. Recenti indagini arche-
Blázquez Martı́nez and J. R. Rodrı́guez, eds. Produccion y ologiche lungo il piano del podio e nelle concamerazioni
comercio del aceite en la antigüedad (Madrid): 225–61. ipogee.” Bollettino di archeologia 41–2: 143–55.
Panella, C. 1999. “Rifornimenti urbani e cultura materiale tra Piraino, C. 1998. “Bolli laterizi dal Colosseo.” RM 105: 119–
Aureliano e Alarico.” In W. V. Harris, ed. The Transformation 25.
of Urbs Roma in Late Antiquity ( JRA Supp. 33, Portsmouth, Pollitt, J. J. 1986. Art in the Hellenistic Age. Cambridge: Cam-
R. I.): 183–215. bridge University Press.
Parker, A. J. 1992. Ancient Shipwrecks of the Mediterranean and Price, S. R. F. 1987. “From Noble Funerals to Divine Cult:
the Roman Provinces. Oxford: BAR IS 580. The Consecration of the Roman Emperors.” In D. Can-
Paulin, E. 1890. Restauration des Termes de Diocletien. Paris: nadine and S. R. F. Price, eds. Rituals of Royalty: Power and
Firmin-Didot et Cie. Ceremonial in Traditional Societies (Cambridge): 56–105.
260
P1: KIC
WORKS CITED
Promis, C. 1862. Le antichità di Aosta – Augusta Praetoria Salas- Rasch, J. J. 1993. Das Mausoleum bei Tor de’Schiavi in Rom.
sorum. Turin: Stamperia Reale. Mainz: Philipp von Zabern.
Purcell, N. 1987. “Town in Country and Country in Town.” Rasch, J. J. 1998. Das Mausoleum der Kaiserin Helena in Rom und
In E. B. MacDougall, ed. Ancient Roman Villa Gardens der ‘Tempio della Tosse’ in Tivoli. Mainz: Philipp von Zabern.
(Washington, D.C.): 185–203. Rea, R., ed. 2002. Rota Colisei. La valle del Colosseo attraverso
Purcell, N. 1994. “The City of Rome and the Plebs Urbana i secoli. Milan: Electa.
in the Late Republic.” In J. A. Crook, A. Lintott, and Rea, R., H.-J. Beste, and L. C. Lancaster 2002. “Il cantiere
E. Rawson, eds. Cambridge Ancient History 9 (Cambridge): del Colosseo.” RM 109: 341–75.
644–88. Reggiani, A. M. 1998. Il santuario di Ercole Vincitore. Milan:
Quilici Gigli, S. 1987. “Su alcuni segni dell’antico paesaggio Electa.
agrario presso Roma.” Archeologia Laziale 8: 152–66. Remesal Rodrı́guez, J. 1983. “Transformaciónes en la ex-
Quilici, L. 1974. Collatia (Forma Italiae I.10). Rome: Danesi. portación del aceite bético a mediados del siglos III d.C.”
Quilici, L. 1986. “Il Tevere e l’Aniene come vie d’acqua a In J. M. Blázquez Martı́nez and J. Remesal Rodrı́guez, eds.
monte di Roma in età imperiale.” Archeologia Laziale 7.2: Producción y comercio del aceite en la antigüedad (Segundo congreso
198–214. internacional, Sevilla, 24–28 febrero 1982) (Madrid): 115–31.
Rakob, F. 1961. “Litus beatae Veneris aureum. Unter- Remesal Rodrı́guez, J. 1986. La annona militaris y la exporta-
suchungen am Venus tempel in Baiae.” RM 68: 114– cion de aceite betico a Germania. Madrid: Universidad Com-
49. plutense.
Rakob, F. 1976. “Hellenismus in Mittelitalien. Bautypen und Remesal Rodrı́guez, J. 1991. “El aceite bético durante el Bajo
Bautechnik.” In P. Zanker, ed. Hellenismus in Mittelitalien Impero.” In A. González Blanco, F. X. Fernández Nieto,
(Göttingen): 366–86. and J. Remesal Rodrı́guez, eds. Arte, sociedad, economia y
Rakob, F. 1988. “Römische Kuppelbauten in Baiae.” RM 95: religion durante el Bajo Impero y la antigüedad tardia (Antigüedad
257–301. y cristianismo 8, Murcia): 355–61.
Rakob, F., and W. D. Heilmeyer 1973. Der Rundtempel am Remesal Rodrı́guez, J. 1998. “Baetican Olive Oil and the
Tiber in Rom. Mainz: Philipp von Zabern. Roman Economy.” In S. Keay, ed. The Archaeology of Early
Ramieri, A. M. 1992. La cisterna romana in via Cristoforo Roman Baetica ( JRA Supp. 29, Portsmouth, R. I.): 183–
Columbo. Rome: Arte Grafiche Jasillo. 200.
Ramieri, A. M. 1993. “Cisterne in via Cristoforo Colombo.” Ricci, C. 1929. “Il Mercato di Traiano.” Capitolium 5: 541–55.
Archeologia Laziale 11.2: 73–8. Rival, M. 1991. La charpenterie navale romaine. Paris: Editions
Ranieri, S. 1998. “A Contribution to the Study of Stratigrafic du Centre national de la recherche scientifique.
Sequence under the AD 79 Layers at Pompeii.” Opuscula Rivoira, G. T. 1909–1910. “The Roman Thermae: The Baths
Pompeiana 8: 135–41. of Diocletian.” Journal of the British and American Archaeo-
Ranieri, S., and T. Yokoyama 1997. “The Somma-Vesuvio logical Society of Rome 4.4: 353–60.
Eruptions Occurred Between 3760 y. B.P. and AD 79.” Rivoira, G. T. 1925. Roman Architecture and Its Principles of
Opuscula Pompeiana 7: 33–50. Construction Under the Empire. Translated by G. Rushforth.
Rankine, W. J. 1858. Manual of Applied Mechanics. London: Oxford: Clarendon Press.
Charles Griffin. Robertson, D. S. 1988. Greek and Roman Architecture. 2nd ed.
Rasch, J. J. 1984. Das Maxentius-Mausoleum an der Via Appia Cambridge: Cambridge University Press.
in Rom. Mainz: Philipp von Zabern. Robinson, O. F. 1992. Ancient Rome: City Planning and Ad-
Rasch, J. J. 1985. “Die Kuppel in der römischen Architektur: ministration. London: Routledge.
Entwicklung, Formgebung, Konstruktion.” Architectura 15: Rodriguez-Almeida, E. 1980. “Alcuni aspetti della topografia
117–39. e dell’ archeologia attorno al monte Testaccio.” In J. M.
Rasch, J. J. 1991. “Zur Konstruktion spätantiker Kuppeln vom Blázquez Martı́nez, ed. Produccion y comercio del aceite en la
3. bis 6. Jahrhundert.” JdI 106: 311–83. antigüedad (Madrid): 103–31.
Rasch, J. J. 1992. “Schalungstragwerke im römischen Rodriguez-Almeida, E. 1994. “Marziale in marmo.” MÉFRA
caementicium-Kuppelbau.” In Bericht über die 36. Tagung 106.1: 197–217.
für Ausgrabungswissenschaft und Bauforshung, 23–27 May 1990 Rodrı́guez-Almeida, E. 1984. Il Monte Testaccio, ambiente, sto-
(Bonn): 16–26. ria, materiale. Rome: Quasar.
261
P1: KIC
WORKS CITED
Rodrı́guez-Almeida, E. 1989. Lo tituli picti de las ánforas olearias Scrinari, V. S. M. 1979. Le navi del porto di Claudio. Rome:
de la Bética. Madrid: Universidad Complutense. Tipografia Centenari.
Rodrı́guez-Almeida, E. 1999. “Le anfore ispaniche per Sear, F. 1982. Roman Architecture. Ithaca, N.Y.: Cornell Uni-
l’alleggerimento delle volte.” In G. Ioppolo and G. Pisani versity Press.
Sartorio, eds. La villa di Massenzio sulla via Appia: il circo Seston, W. 1946. Dioclétien et la Tétrarchy. Paris: E. de Boccard.
(Rome): 235–45. Smith, S. 1981. Builder’s Detail Sheets. London: Northwood
Roldán Gómez, L. 1988. “El opus testaceum en Itálica. Edi- Books.
ficios Privados.” Archivo Español de Arqueologia 61: 119–40. Society, C. 1986. Formwork: A Guide to Good Practice. London:
Roldán Gómez, L. 1993. Téchnicas constructivas romanas en Concrete Society Institution of Structural Engineers.
Itálica (Santiponce, Sevilla). Madrid: Universidad Autónoma Spinazzola, V. 1953. Pompei alla luce degli nuovi scavi
de Madrid, Departamento de Prehistoria y Arqueologı́a. dell’Abbondanza (Anni 1910–1923). Rome: Libreria dello
Romanelli, P. 1973. “Il criptoportico del Palatino.” In Les Stato.
cryptoportiques dans l’architecture romaine (Paris): 208–18. Squarciapino, M. F. 1958. Scavi di Ostia 3 – Le necropoli 1.
Romualdi, A. 1991. “Populonia in età ellenistica.” In J. Le tombe di età repubblicana e augustea. Rome: Liberia dello
Remesal and O. Musso, eds. La presencia de material etr- Stato.
usco en el ámbito de la colonisación arcaica en la peninsula ibèrica Steinby, E. M. 1974–1975. “La cronologia delle figlinae do-
(Barcelona): 77–82. liari urbane dalla fine dell’età repubblicana fino all’inizio
Rose, C. B. 2005. “The Parthians in Augustan Rome.” AJA del III sec.” BullCom 84: 7–132.
109.1: 21–75. Steinby, E. M. 1978. “Ziegelstempel von Rom und Umge-
Rosi, G. 1931. “Pantheon.” BullCom 59: 227–9. bung.” RE Supp. 15: cols. 1489–531.
Rowland, I., and T. Howe, eds. 1999. Vitruvius: The Ten Books Steinby, E. M. 1986. “L’industria laterizia di Roma nel tardo
on Architecture. Cambridge: Cambridge University Press. impero.” In A. Giardina, ed. Società romana e impero tardo
Rykwert, J., N. Leach, and R. Tavernor 1988. Alberti. On antico II: Roma politica, economia, paessagio urbano (Bari): 99–
the Art of Building in Ten Books. Cambridge, Mass.: M.I.T. 164.
Press. Steinby, E. M. 1993. “L’organizzazione produttiva dei lateri:
Saalman, H. 1959. “Early Renaissance Architectural Theory un modello interpretivo per l’instrumentum in genere?”
and Practice in Antonio Filarete’s ‘Trattato di Architet- In W. V. Harris, ed. The Inscribed Economy ( JRA Supp. 6,
tura’.” ArtB 41: 89–106. Portsmouth, R. I.): 139–43.
Salza Prina Ricotti, E. 1988. “Villa Adriana. Un singolare Steinby, E. M. 2001. “La cronologia delle figlinae tardoan-
solaio piano in opus caementicium.” Palladio (n.s.) 1: 15–26. tiche.” In M. Cecchelli, ed. Materiali e techniche dell’edilizia
Salza Prina Ricotti, E. 1988–1989. “Le tende conviviali e la paleocristiana a Roma (Rome): 127–50.
tenda di Tolomeo Filadelfo.” In R. I. Curtis, ed. Studia Storz, S. 1994a. Tonröhren im antiken Gewölbebau. Mainz:
pompeiana & classica in honor of Wilhelmina F. Jashemski 2 Philipp von Zabern.
(New York): 199–240. Storz, S. 1994b. “La tecnica della costruzione delle volte con
Salza Prina Ricotti, E. 2001. Villa Adriana: il sogno di un im- tubi fittili a S. Sefano Rotondo.” Corsi di Cultura sull’arte
peratore. Rome: “L’Erma” di Bretschneider. ravennate e bizantina 41: 669–93.
Scavizzi, C. P. 1983. Edilizie dei secoli XVII e XVIII a Roma. Strong, D. 1968. “The Administration of Public Building in
Rome: Ministero per i beni culturali e ambientali, Ufficio Rome during the Late Republic and Early Empire.” BICS
studi. 15: 97–109.
Scherillo, A. 1941. “Studi su alcuni tufi gialli della regione Taborelli, L. 1980. “Elementi per l’individuazione di
Sabazia orientale.” Periodico di Mineralogia 12.3: 381–417. un’officina vetraria e della sua produzione a Sentium.”
Schlebeker, J. T. 1977. “Farmers and Bureaucrats: Reflections ArchCl 32: 138–73.
on Technological Innovation in Agriculture.” Agricultural Tannery, P. 1883. “Études Héroniennes.” Mémoires de la société
History 51: 641–55. des sciences physiques et naturelles de Bordeaux (ser. 2) 5: 347–
Scrinari, V. S. M. 1963. “Il porto di Claudio ed osservazioni 69.
sulla tecnica del conglomerato cementizio presso i Ro- Taylor, R. 2003. Roman Builders: A Study in Architectural Process.
mani.” L’industria italiana del cemento 33.7: 527–38. Cambridge: Cambridge University Press.
262
P1: KIC
WORKS CITED
Tedeschi Grisanti, G. 1977. I trofei di Mario – Il ninfeo dell’aqua van Wonterghem, F. 1976. “Archaeologische Zeugnisse
Giulia sull’esquilino. Rome: Istituto di study romani. spätrepublikanischer Zeit aus dem Gebeit der Paligner.”
Tedone, G. 1990. “Struttura della cupola del c.d. Ninfeo.” In P. Zanker, ed. Hellenismus in Mittelitalien 1 (Göttigen):
Bollettino di archeologia 5–6: 169–70. 143–9.
Tedone, G. 1993 (pub. 1998). “Roma. Velabro, la fabbrica Vendittelli, L. 2002. “La conservazione e la valorizzazione del
superiore dell’Arco di Giano.” Bollettino di archeologia 23–4: Mausoleo e Sant’Elena. Nuovi dati dai lavori di scavo e
195–202. di restauro.” In F. Guidobaldi and A. G. Guidobaldi, eds.
Terenzio, A. 1932. “La restauration du Pantheon de Rome.” Ecclesiae Urbis: Atti del congresso internazionale di studi sulle
Museion 20: 52–7. chiese di Roma (IV–V secolo) 1 (Vatican City): 771–92.
Thode, D. 1975. Untersuchungen zur Lastabtragung in Spatan- Ventriglia, U. 1971. La geologia nella città di Roma. Rome:
tiken Kuppelbauten. Darmstadt: Studien zur Bauforschung n.p.
9. Verità, M. 1999. “Le sabbie e il vetro.” In A. Ciarallo and E. De
Timoshenko, S. P. 1983. A History of the Strength of Materials. Carolis, eds. Homo faber: natura, scienza e technica nell’antica
New York: Dover. Pompei (Milan): 108–10.
Töbelmann, F. 1915. Der Bogen von Malborghetto (Abhandlungen Viollet-le-Duc, M. 1868. Dictionnaire raisonné de l’architecture
der Heidelberger Akademie der Wissenschaft, II). Heidelberg: française du XIe au XVI siècle 9. Paris: Morel.
Carl Winters Universitätsbuchhandlung. Volpe, R. 2002. “Un antico giornale di cantiere delle terme
Tosi, A. 1997. “Un esempio di analisi strutturale: l’ Eliocamino di Traiano.” RM 109: 377–94.
di Villa Adriana.” Materiali e strutture 7.2–3: 122–30. Waddell, J. J., and J. A. Dobrowolski 1993. Concrete Construction
Treggiari, S. 1969. Roman Freedmen during the Late Republic. Handbook. 3rd ed. New York: McGraw-Hill.
Oxford: Clarendon Press. Waltzing, J.-P. 1896. Étude historique sur les corporations profes-
Triglia, R., E. Agosta, C. Currado, A. A. De Benedetti, sionelles chez les romains depuis les origines jusqu’à la chute de
C. Freda, M. Gaeta, D. M. Palladino, and C. Rosa 1995. l’Empire d’Occident. 2. Louvain: Charles Peeters.
“Petrology.” In R. Triglia, ed. The Volcano of the Alban Hills Ward-Perkins, B. 1984. From Classical Antiquity to Early Middle
(Rome): 95–165. Ages: Urban Public Building in Northern and Central Italy AD
Tuck, S. 2000. “A New Identification for the ‘Porticus Ae- 300–850. Oxford: Oxford University Press.
milia’.” JRA 13.1: 175–83. Ward-Perkins, J. B. 1966. “Imperial Mausolea and Their Pos-
Ucelli, G. 1950. Le navi di Nemi. Rome: Libreria dello Stato. sible Influence on Early Christian Central-Plan Buildings.”
Uggeri, A. 1800. Journées pittoresques des édifices de Rome anci- JSAH 25: 297–9.
enne 1. Rome: Pagliarini. Ward-Perkins, J. B. 1981. Roman Imperial Architecture. 2nd ed.
Ungaro, L. 1995. “Mercati Traianei: l’articolazione degli edi- Harmondsworth: Penguin.
fici.” In E. La Rocca, L. Ungaro, and R. Meneghini, eds. Ward-Perkins, J. B. 1994. “Memoria, Martyr’s Tomb and
I luoghi del consenso imperiale: il Foro di Augusto, il Foro di Martyr’s Church.” In Studies in Roman and Early Christian
Traiano (Rome): 126–35. Architecture (London): 495–516.
Valeriani, S. 2003. “Historic Carpentry in Rome.” In S. Warren, J. 1976. Greek Mathematics and the Architects to Justinian.
Huerta, ed. Proceedings of the 1st International Congress on Art and Archaeology Research Papers. London: Coach
Construction History 3 (Madrid): 2023–34. Publishing.
Van Beek, G. W. 1987. “Arches and Vaults in the Ancient Wataghin Cantino, G. 1966. La Domus Augustana: personalità
Near East.” Scientific American (July): 78–85. e problemi dell’architettura flavia. Turin: Giappichelli.
Van Deman, E. B. 1912. “Methods of Determining the Date White, K. D. 1984. Greek and Roman Technology. London:
of Roman Concrete Monuments.” AJA 16: 230–51, 387– Thames and Hudson.
432. Williams, M. S., and J. D. Todd 2000. Structures: Theory and
Van Deman, E. B. 1924. “The House of Caligula.” AJA 28: Analysis. London: Macmillan Press.
368–98. Williams, S. 1985. Diocletian and the Roman Recovery. New
Van Nice, R. L. 1965. Saint Sophia in Istanbul: An Architectural York: Methuen.
Survey. Washington, D.C.: Dumbarton Oaks Center for Wilson Jones, M. 2000. Principles of Roman Architecture. New
Byzantine Studies. Haven: Yale University Press.
263
P1: KIC
WORKS CITED
Wilson Jones, M., P. Davies, and D. Hemsoll 1987. Yokoyama, T., and A. Marturano 1997. “Volcanic Products of
“The Pantheon: Triumph of Rome or Triumph of the Vesuvius Eruption in AD 79 at Pompeii, Italy, with Spe-
Compromise?” Art History 10: 133–53. cial Reference to the Effects on Human Life and Ancient
Wilson, R. J. A. 1990. “Terracotta Vaulting Tubes (tubi City.” Opuscula Pompeiana 7: 1–32.
fittili): On Their Origin and Distribution.” JRA 5: 97– Zalewski, W., and E. Allen 1998. Shaping Structures. New York:
129. John Wiley & Sons.
Yegül, F. K. 1992. Baths and Bathing in Classical Antiquity. Cam- Zander, G. 1958. “La Domus Aurea: nuovi problemi architet-
bridge, Mass.: M.I.T. Press. tonici.” BCSSA 12: 47–64.
Yegül, F. K. 2003. “Building a Roman Bath for the Cameras.” Zorzi, G. 1959. I disegni delle antichità di Andrea Palladio. Venice:
JRA 16.1: 153–77. N. Pozza.
264
P1: JPJ
0521842026ind.xml CB885B/Lancaster 0 521 84202 6 April 17, 2005 18:13
INDEX
The following index includes a reference to the entry for each monument listed in the Catalogue of Major Monuments in Appendix 1;
however, the details within each entry are referenced only when they refer to information not discussed in the main text. Likewise, each
catalogue of the building techniques included in Appendix 2 is also referenced in the appropriate place, but the details within it are not.
Adam, Jean-Pierre, 2, 7 Dressel 23, 69, 75, 76, 79, 81 structural analysis of effect on
aerarium, 170, 171 at Cologne, 84–85 thrustline, 160, 164
aggregate. See caementa identifying, 69 as structural innovation, 78–80,
Agrippa, 170 latest examples in vaults, 84 83–84, 168
Alatri, 87 location at crown of vault, 83, 84, 164 in walls, 70, 73, 74, 75, 81
Albano, Emissarium, 28 supercedes Dressel 20, 82 for wine, 69
Alberti, Leon Battista amphoras annona, 81, 84–85
on amphoras in vaults, 68, 85 in foundations, 74 Aosta (Augusta Praetoria), 59, 74, 132
on lime slaking, 54 for land reclamation, 71–74, 81, 83, Apollodorus of Damascus, 44, 46, 98
on washing pozzolana, 56 168 bridge over Danube, 23, 24
on water for curing concrete, 75, recycled, 69 Aqua Claudia, 187, 196
235n12 in vaults. See also names of individual Aqua Virgo, 170
Alexander Severus, 148 buildings arch, 6, 7, 21
Alexander the Great, 44 in Byzantine architecture, 238n68 arc of embrasure, 132–133
Alexandria, 50 examples, early, 69–74, 83 relation to concrete vault, 7
Amici, Carla Maria, 2, 60, 106, 118–125, examples, late antiquity, 75–80, 172 structural behavior of, 6–7, 151–152
156 explanation, to aid curing of failure patterns, 151
Amphitheater of Statilius Taurus, 170, 186 concrete, 68, 74–75 limit state, 155
amphitheaters/theaters, importance for explanation, to improve sound statically determinate, 152
vaulting, 168–169 resonance, 68, 70 three-hinge, 151
amphora types, 70 explanation, to reduce materials and Arch of Constantine, 83
Africana 1, 69, 75, 76 labor, 68, 76, 80, 81, 167–168 Arch of Janus, 80, 202
Almagro 51c, 69, 75 explanation, to reduce weight, 68, Archimedes, 10–11, 46
Dressel 20, 69, 75 70, 76, 77, 85 architect, 10
early examples in vaults, 70–71 list of examples, 215 legal obligations, 20–21, 233n89
identifying, 69 medieval imitation of, 79, 80, 84, 85 architrave bars. See Hadrian’s Villa
at Monte Testaccio, 81–82 at Ostia, 69–70, 72, 81 Arco di Malborghetto, 63–64, 105,
tituli picti, 81–82 reuse, 82–83 203–204
265
P1: JPJ
INDEX
Arles, 115, 239n13 ribs, Plate XIII, 98, 100, 102–104, 107, brick linings (on intrados), 29–32
Athenaeus, 50 111 explanation for, 29–30, 49
Augustus scoria, Vesuvian, Plate IX, Plate XIII, list of examples, 207–210
building projects, funding of, 170 62–63, 66, 177 pattern of use, 31–32, 49
commissions, establishment of, 18 analysis of, 64, 222–224 relation to brick industry, 31–32, 49
Aurelian Walls, 28, 50, 108 tie bars, 118, 120, 125, 126 upright bricks in, 30–31, 96
Baetica, 69, 70–71, 81, 82, 84–85 vaulted terraces, 128, 169 brick stamps
basilica, Christian, 179 Baths of Constantine, 178 anepigraphic, 186, 195
Basilica Aemilia, 185 Baths of Diocletian, 177, 197 information from, 17, 19
cross vaults, 35 buttressing, 137 bricks, 17
metal bands in architrave, 116 “Planetarium,” 109, 110, 234n69 bessalis, 17
tie bars, 116, 117, 118, 128 pozzolana, 57 in brick linings, 29, 30–31
Basilica Julia, 196 pumice, Plate X, Plate XIII, 63, 64 for suspensurae, 59, 60
scoria, Vesuvian, 63, 66, 67, 177 ribs, Plate XIII, 102–103, 110 bipedalis, 17
analysis of, 64, 223–224 tie bars in brick linings, 29, 30
Basilica Neptuni, 192 anchor blocks, 82, 118, 120 horizontal courses in domes, 48, 110,
Basilica of Maxentius, 106, 197–198 with pumice, 126 144
amphora in vault, 69, 75 spacing of, 125 protection for paving, 58
buttressing, 137–138, 139 Baths of Maxentius, 198 for suspensurae, 59, 60, 65
coffers, 198 ribs, 103, 104–105, 107, 110 unit in modular design system, 131,
cracks, 8 Baths of Nero, 148, 241n14 132–133
deformation of vault, 34, 36, 37, 49 Baths of Trajan, Plate VI, 189–190 in vaulting ribs, 88. See also ribs:
pumice, 63–64 buttressing, 137 bipedalis
ribs, 106 cracks in semidome, 8, 139 early use in Rome, 17
roof tiles, 58 graded caementa, Plate VI, 60–61 reused, 82
tufo giallo, 63–64 ribs, 96, 97 sesquipedalis, 17
Basilica Ulpia scoria, Vesuvian, 66 for brick linings, 29, 30, 190, 191,
architrave/cornice block, 126 analysis of, 64, 223–224 194
dovetail clamps in entablature blocks, with tufo giallo, Plate VII, 61 for relieving arches, 96, 99
121, 124, 156 suspensurae at, 59 Brown, Frank, 46
fallen vaults from, 121–125, 240n25 tie bars, 116–118, 119, 126 Brunelleschi, Filippo, 22
Mons Claudianus columns, 128 tufo rosso a scorie nere, 61 Brunt, Peter, 18
scoria, Vesuvian, 61–62, 66, 167 vaulted terraces, 128 building industry, 18–21, 65, 175
analysis of, 64, 222–224 Bauer, Heinrich, 115–116 clientela, 19, 20
with tiebars, 63, 170 bitumen, as waterproofing, 173 collegia, state control of, 20, 21
structural analysis of, 156, 157 Blackfriar’s Bridge (London), 27, 28 freedmen, 19, 20, 21
tie bars, 118–119, 121–125, 126, 156, Blake, Marion E., 2, 7, 12, 113 in late antiquity, 21
170 on pozzolana, 56 as means of social advancement, 19, 21,
truss, 23, 44 Bloch, Herbert, 176, 177 112, 175–176, 178
vaulted terraces, 128 Bolsena, 175, 224, 236n61 Opera Caesaris, 232n71
bath buildings, importance for vaulting, brick industry, 175–176 slave labor, 18–19
168–169, 172 gens Domitii, 175–176, 242n36 building materials. See also names of
Baths of Agrippa, 195–196 in late antiquity, 82, 178 individual materials
date, 108 Marcus Aurelius, inheritance of, 175 supply of, 21, 170, 177–178
ribs, 108, 109, 110 relation to brick linings, 31–32, 49 transport of, 17–18
Baths of Caracalla, 12, 195 relation to ribs, 111, 112 buttressing, 94, 124–125, 137, 145–146,
buttressing, 137 senatorial class involved in, 175, 176 147, 190
centering, 37, 39, 49 women in, 176 of arch, 7
266
P1: JPJ
INDEX
balancing thrusts, 133–134, 135, 146–147 for cross vault, 34–40, 49–50 S. Pietro, 179
side walls of semidome, 140 deformation of, 34, 36, 40–42, 49, truss, 23, 24, 37
“Temple of Minerva Medica,” 146, 147, 167 S. Sabina, 129
168 for dome, 40–48, 49–50 S. Sebastiano, 55–56
travertine ties as connectors, 145, 146 earth, 188, 190 S. Stefano Rotondo, 50
window glass influence on, 147 hanging, 44, 48 Circus of Maxentius (Via Appia), 80,
buttressing arch holes for, 32–34, 35 198–199
Basilica of Maxentius, 137–138, 139 laying out the forms, 22, 36–37, 42, amphoras in vaults, 75–77
Baths of Caracalla, 137 49, 187 clamps
Trajan’s Markets, 135–137 for pavilion vault, 35, 38 to counter lateral thrusts, 10
buttressing walls permanent, for lintel arches, 127 dovetail (iron), 121, 124, 128, 156
Domus Aurea, 142–143, 144 removal of, 26, 29, 49 dovetail (wood), 113, 127, 239n4
“Temple of Mercury” (Baiae), 142–143, Colosseum, 36, 38 early use of, 113, 115
144 Hagia Sofia, 29 for formwork, 42, 43
caementa, 3, 59–64 Pantheon, 45–46 at Palestrina, 9
change under Severans, 62–63, 66 for segmental vault, 46–47 pi-shaped, 113, 119, 124
graded according to weight, 59–64 support of, 27, 32–34 Claudius, harbor at Portus, 55
to counter lateral thrust, 10 Baths of Caracalla, 37 clay, 12, 17, 19
development and understanding of, Colosseum, 36, 38 clerestory windows, 49, 111, 137, 143, 144,
167 Domus Aurea, 43 147, 168
efficacy of, 164 Pantheon, 44–45 cocciopesto, 58–59, 65
structural analysis of effects, 160–161 Trajan’s Markets, 37 amphora shards, 69, 81
laid horizontally, Plate IX, 7 wedges for removal, 27–28 list of examples, 211–212
laid radially, 5, 9, 59 weight of frames, 37, 49 pot shards, 236n49
at Palestrina, 9, 10 Champlin, E., 148 properties, 58, 65
provenance of, 64 Chieti, cistern at, 239n18 codicariae, 67. See also transport: river
refuse from older buildings, 59 Choisy, August, 153 coffers, 8
types in vaulting, 13–16 on brick linings, 29–30 Basilica of Maxentius, 36, 198
Caesarea Maritima, imported pozzolana, on lattice ribs, 98, 107 Pantheon, 44, 45–46
55 on amphoras in vaults, 78 Sanctuary of Fortuna Primigenia, 9, 10,
Çakmak, A. S., 8 Chronographer of A.D. 354, 82 11
Canina, Luigi, 63 churches Villa of Domitian, 87, 97
Capitoline, Roman insula. See “Casa di via S. Anastasio (Tre Fontane), 84 collegia. See building industry
Giulio Romano” S. Bernardo, 197 collegium fabrum tignariorum, 20
Caracalla, increased tax base, 171 S. Clemente, structure below, 34, 35, colonnade supporting vault, 121
Caraffa, Paolo, 78 188–189 architrave bars, 125, 126–127, 128
Casa de la Exedra (Italica, Spain), 70–71, S. Costanza, 127, 129, 179 lightweight caementa, 61, 63, 126, 170
73 S. Croce, 179 methods of stabilization, 114–115, 128,
Casa del Sacello Iliaco (Pompeii I.6.4), 54 S. Giorgio (Riofreddo), 84 168
“Casa di via Giulio Romano,” 69, 75, 76 S. Giovanni in Laterano, 179 tie bars, 63, 118–125, 126, 170
Case a giardino (Ostia III.9.22–3), 30 St. Jerome (Cologne, Germany), 84 Tor de’Schiavi porch, Plate XII, 80
Castra Praetoria, 74 S. Lorenzo Fuori le Mura, 129 Colosseum, 186
center of gravity, 10, 225 SS Marcellino e Pietro, 179 caementa, laid horizontally, 59
centering, 22, 23 S. Maria Antiqua, 188 caementa, tufo giallo, 60
assembly of, 22–26 S. Maura, 83, 84, 164 cross vaults at, 36, 38
Blackfriar’s Bridge, 28 S. Paolo Fuori le Mura formwork imprints, 36, 38
central tower for domes, 42, 44–45 pot in vault of cloister, 84 funded from spoils of war, 171
corbels for, 32–34, 37, 101, 104 truss, 23, 24 labor, Jewish captives, 232n70
267
P1: JPJ
INDEX
Colosseum (cont.) Culmann, Karl, 152 buttressing of large barrel vault, 134,
pumice in mortar, Plate V, 56, 60, 167, Cumae, pozzolana from, 52 146
224 cura operum publicorum, 18, 232n66 cross vaults at, 35
ribs, 88–91, 92 Curia, pozzolana at, 57, 235n30 octagonal room, 42
bipedalis, 90, 93, 105 De Angelis d’Ossat, Gioacchino, 64, buttressing walls, 142–143, 144, 147
ladder, 90–91, 93 222 dome centering, 42–43, 45
lattice, 93, 101–103, 105, 111–112 De Angelis d’Ossat, Guglielmo, 43, 64 Domus Flavia, 187–188
travertine, 88–90, 91 de’Ficoroni, Francesco, 80 aula regia, 23, 187–188
concrete Degrassi, A., 5 hypocaust floor of triclinium, 178
to control light and space, 135–137, 142, DeLaine, Janet, 12, 18, 20, 37 Domus Tiberiana, 186–187
148 Baths of Caracalla, 2, 37, 98, 100, 175 hanging gardens, 58, 60, 173–174
curing of, 26–27, 53 design, 130 lime paste, 234n42
difference between ancient and mistakes, 42, 146 ribs, 187, 239n28
modern, 3 modular system, 131, 132–133, 146 Dressel, Heinrich, 74
early development of, 3–6 rule of thumb, 130, 131, 146 Elba, iron ore, 113
as fireproofing, 1, 6 barrel vaults, 132–133 Esposito, Fabrizio, 131
shrinkage, 52–53 bridge arches, 132 Eusebius, 180
strength of, 26–27, 242n21. See also cross vaults, 135 extrados, protection of, 58–59, 65
mortar: pozzolanic: strength dome, 140 Faventius, 17
Constantine, 84, 178, 180 wall thickness, 133 Ferentino, market, 6, 134
ecclesiastical building program, 178–179 Diades (engineer), 44 figlinae, 232n50, 240n31, 243n51
Constantinople, 180–181, 243 Digest of Justinian, 20 Fine Licht, Kjeld de, 2, 34
contractors, 18–21, 170 Dio Cassius, 46, 52, 75, 167, 174 fires in Rome, 169, 170
legal obligations, 20–21 Diocletian A.D. 64, 1, 11, 169
contracts building under, 177 A.D. 217, 101
locatio conductio, 19, 20 reorganization of provinces, 85 A.D. 283, 63, 169, 177
relation to collegia, 20 reorganization of brick industry, 82 fiscus, 171
Corsica, fir from, 17 tax reform, 21, 67, 172 flat vaults, 231n18
Cosa, pozzolana at harbor, 55 Diodorus Siculus, 173 Florence, Duomo, 22
Cozzo, Giuseppe, 30, 99 Dionysius of Halicarnassus, 18 Fondi, land reclamation, 74
cracks in concrete, 7–10, 62, 69, 139, 145, Diribitorium, 23 Fontana, Domenico, 24
161 dome, 2, 139 Forma Urbis. See Severan Marble Plan
crane, 25 in bath buildings, 49 formwork
capacity, 37 celestial symbolism of, 46, 174–175, adhesion to concrete, 28, 29
for centering, 45, 49 180 as insulation, 27
lifting tower, 45 centering for, 40–48, 49–50 board length, 25, 34, 40, 48, 57, 111
creep, 8, 9, 48, 52–53 hollow spaces in haunch, 193 brick linings. See brick linings (on
Hagia Sophia, 53 parabolic form, 164, 165 intrados)
“Temple of Mercury” (Baiae), dome of heaven. See dome: celestial imprints in mortar, 35, 36, 38, 42–43,
157–158 symbolism of 47, 107, 124, 186
Trajan’s Markets, 137 domed mausolea, 179–180 list of examples, 205–206
cross vaults Domitianic Vestibule, 188 wood grain in, 186
in bath buildings, 49 ribs, 91–94 radial, for dome, 40, 46, 48
centering for, 34–40, 49–50 Domitius Alexander, 82 reed mats, 28, 233n31
early examples of, 34 Domus Augustana, 187–188, 240n35 relation to ribs, 110, 111
for lighting, 49 ribs, 94–95 relation to step-rings, 144
cruma. See pumex pompeianus; scoria: from Domus Aurea, 185 reuse of, 28, 30, 48, 49
Vesuvius brick lining, 233n33 surviving examples, 28, 186
268
P1: JPJ
INDEX
Forum Boarium, round temple, 113 Hall of the Doric Pilasters, 124, 125, importance for centering, 48, 167
Forum of Augustus, 87, 113, 127, 170 126–127 metal bands, 24
Forum of Caesar, 191 Heliocaminus Baths, 30, 31 scarf joint, 23–24, 25
caementa, 61 Large Baths, 30, 40, 46, 50 in shipbuilding, 23–24, 48, 107
cocciopesto, 59 Maritime Theater, 126 Jones, A. M. H., 171
red pozzolana, 235n30 Piazza d’Oro, 46 Juvenal, 169
scoria, Vesuvian, 60, 61 Serapeum, 46, 47, 50, 126 Krause, Clemens, 174
analysis of, 64, 223–224 Stadium Garden, 126 Krautheimer, Richard, 179
Trajanic latrine, 98 Hadrianeum, 194 Lamprecht, Otto, 3
ribs and relieving arches, 96, 98, 99, scoria and tufo giallo, 62, 66, 167 lapis Albanus. See peperino
105 Hagia Sophia (Istanbul), 181 lapis Gabinus. See peperino
impost blocks, 96, 98, 99 cracks, 8, 140 lapis Tiburtinus. See travertine; quarries
travertine keystones and impost blocks, creep, 53, 140 lead, as waterproofing, 173
87 removal of centering, 29 Leger, Alfred, 33, 34, 153
Forum of Trajan, 171, 190–191. See also Hanging Gardens of Babylon. See Lehmann, Karl, 50
Basilica Ulpia; Trajan’s Markets Seven Wonders of the Ancient leucititic lava, Plate I, 13–16, 198
tie bars, 114, 118–125 World Liber Pontificalis, 179, 180
Frank, Tenny, 12 harena fossicia, 55 lime, 16–17, 53–54
Frontinus, 18, 52 Helena, mother of Constantine factors affecting quality, 167
funding of imperial building projects, power in Rome, 243n62 firing of, 53, 54
170–172 property in Rome, 178, 179 as glue, 30
bona damnatorum, 170 Hemsoll, David, 50 kiln at Lucus Feroniae, 18
debasement of coinage, 171, 172 Heres, Thea L., 57 quicklime, 53, 54
in late antiquity, 177–178 Heron of Alexandria, 10–12, 25–37, 45 slaking, 53, 54
from spoils of war, 171 On Vaulting (Camarika), 10 from spolia, 54, 65
funicular polygon analysis. See structural Stereometrica, 10 supply of, 65, 167, 172
analysis: funicular polygon Heyman, Jacques, 2, 154, 165 transport of, 17, 18
method Hooke, Robert, 152, 165 limestone
Gallienus, 82 Horace, 25 as caementa, 9
gardens, hanging, 58–59, 60, 65, 173, horizontal thrust. See structural behavior: for lime, 12, 16
174 lateral thrust sources of, 15, 16
Gatti, Guglielmo, 5 Horrea Agrippiana, 94, 184 transport of, 17
Gaul, transhipment ports, 81 tie bar, 115–116, 128 line of thrust, 152
geology, Rome, 12, 14 Horrea Galbana, 81 Lint, H. F. van, 80
geometry horti pensiles, 58–59, 174. See also Livy, 5
centering for octagonal dome, 43 gardens, hanging; Domus locatio conductio. See contracts
golden section, 131 Tiberiana; Seven Wonders of the Lucus Feroniae, 18
symbolism at Pantheon, 44, 46 Ancient World Lugli, Giuseppe, 2, 5, 8, 12
Giovannoni, Gustavo, 153 Horti Sallustiani, 46, 193 MacDonald, William L., 1, 3, 8
Giuliani, Cairoli Fulvio, 2, 58, 86, 99 Hutchinson, Paul, 160–161 Magazzini “Traianei” (Ostia.20.1), 69–70,
granite, Mons Claudianus, 118, 128, 170 impost blocks, 98, 99, 125, 126–127 72
Guidobaldi, Federico, 34 iron, 48, 113 Mainstone, Roland, 3, 8
Gullini, Giorgio, 5, 10 mines at Noricum, 25, 170 marble trade, 113, 128
Hadrian, 46–47 processing for steel, 25 giallo antico, 129
Hadrian’s Villa Isodorus of Miletus, 10 Greek marble, importation of, 113,
architrave bars and impost blocks, Jerusalem, Vespasian’s victory at, 171 127
126–127, 128 joinery Roman desire for colored columns,
Edificio a Tre Esedre, 231n18 bolts, 25, 46 128, 169, 170
269
P1: JPJ
INDEX
Mark, Robert, 2 properties, 3, 51, 53, 65 black pozzolana, 56

finite element model of Pantheon, proportion of mix, 55 centering, 43–46
160–161, 164 strength of, 6, 51, 52 coffers, 45–46
Hagia Sophia, 8 See also concrete: strength of cracks in dome, 8, 139
Marseilles, land reclamation, 74 simple lime, 3, 53, 55, 75 dome, celestial symbolism of, 46,
Martial, 148 properties, 51, 53 174–175
mathematics mosaic, 50, 58, 59, 180 geometry of, 44, 46, 130–131
ancient knowledge of, 10–12 Nemi, ships from, 24 graded caementa, 62, 77, 167
calculations, 10, 132 Nero, 148, 170–171. See also fires in relieving arches, 97–98, 101, 144
units of measure, 10 Rome: A.D. 64; Domus Aurea ribs, 96–98, 100, 144, 192
Mattingly, David, 81–82 imitating Seven Wonders, 173, 174 roof tiles, marble and bronze, 58
Mausoleum of Augustus, 180 Newstead (Scotland), bolts from, 25 scoria, Vesuvian, 62, 66, 167, 222
Mausoleum of Diocletian (Split), 179 Newton, Isaac, 150 stability of, 144, 146, 147
Mausoleum of Helena, 179–180, 202–203 Niederbieber (Germany), bolts from, 25 step-rings, 141, 143
amphoras in vaults, 57, 68, 69, 76, 80, Nimes, 115, 239n13 structural analysis of, 153, 158–161
84 Noricum, 25, 48, 170 Peacock, D. P. A., 128
bipedalis course in dome, 48, 57 Nymphaeum Alexandri, 34, 104, 196 Pearse, J. D. L., 20
cracks in dome, 69, 139 ribs, 101 Peña, J. T., 82
formwork imprints, 110 O’Connor, Colin, 132 peperino, Plate I
pozzolana, 57–58 Odeum of Agrippa (Athens), 23 caementa in vault, 80, 194, 203
radial formwork, 110 olive oil supply, 81, 82, 84–85 lapis Albanus, 13, 203
ribs, 69, 109, 110 opus caementicium. See concrete lapis Gabinus, 5, 13, 113, 232n40
as setting for liturgical ceremeonies, opus signinum. See cocciopesto photogrammetry, 40, 42, 48, 110, 157,
180 opus spicatum, 58 167
step-rings, 57, 110 organization of building process Piranesi, G. B.
two phases of, 57–58 advances in late antiquity, 167 Blackfriar’s Bridge, 27, 28
Mausoleum of Romulus (Via Appia), 179, centering, assembly and removal, 26, Pons Cestius, 33
198–199 28–29, 49 pitched brick vaulting, 50, 108, 181
Maxentius, 82, 177–178 construction sequence, 106, 189 plague in Rome, 65
Maximian, 177 dome construction, systemization of, Pliny the Elder
Meiggs, Russell, 17 48, 110 on cocciopesto, 58
Michetti, Antonio, 131 separate work crews, 76–77 on damming the Tiber, 17
Middleton, J. H., 7 stabilization during construction, 127 on fir, 17
Milani, G. B., 153, 158 Ostia on lime, 17, 54
Misenum, pozzolana from, 52 amphoras from, 69–70, 72, 81 on mortar mix, 55
Monte Testaccio, 81–82 building industry at, 232n74 on pozzolana, imported, 55
“piccolo Testaccio,” 238n55 impost block, 240n35 on saws, 26
mortar, 3, 51–59 lime kilns at, 54 Pliny the Younger, 175, 176
Greek, 3 pozzolana in private structures, 57 Pompeii
oil added to, 58 scoria, Vesuvian, 66, 67 amphoras in wall (I.20.5), 74, 75
Portland cement, 51–52 transhipment port, 12, 16, 66 concrete vaulting, early examples, 6
pozzolanic, Plate VII Otricoli, Hadrianic bath, 46 scoria from, 64, 222
curing of, 52, 167 Ovid, 148 tools from, 26
early use of, 21 Packer, James E., 118–125, 156 window glass from, 147
factors affecting quality, 27, 52, 56, Palladio, 148 Pons Cestius, 33, 183–184
65 Palladius, 56 Pons Fabricius, 32, 183
ancient understanding of, 52, 56, Pantheon, 179, 192 Pont du Gard, 33, 34
166–167 bipedalis course in dome, 48 Pont St. Martin (near Aosta), 59, 132
270
P1: JPJ
INDEX
Populonia, iron ore, 113 porphyry, Mons Porphyrites, 170 Guadalquiver (Spain), 71
Porta Latina, 84 pozzolana, Plate IV, 16, 55–56 Nar, 17–18
Porticus Aemilia, Plate II, 5, 133 travertine, Tivoli, 16, 88, 112, 127 Nile (Egypt), 128
Portus, imported pozzolana, 55, tufo giallo della via Tiberina, 13, 66 Pallia, 17
235n25 tufo lionato, 13, 66 Po, 74
Postumus, usurper, 82 tufo rosso a scorie nere, 13 Rhine (Germany), 84
pozzolana, Plate III, 3, 54–58. See also Rakob, Friedrich, 42, 44, 157, 167 Tania, 17–18
mortar: pozzolanic Rankine, W. J., 158, 159 Tiber
Bay of Naples, 16, 54–55 Rasch, J. J., 2, 44 clay, 12, 17, 127
black, high quality, 56, 57 “Temple of Mercury,” 42 cura alvei Tiberis, 18
early use in concrete, 3, 55 late Roman domes, 48, 108, 110, 167 lock system on, 17, 232n58
pozzolanella, 55, 57–58, 166 Tor de’Schiavi, 80 transport, 12, 16, 17–18
properties, 51, 52 redemptores. See contractors Rivoira, G. T., 106
red, high quality, 55, 56, 57 relieving arch, 86 robber holes, 114, 120, 125
sieving and washing of, 56, 57, 65, protecting architrave, 87, 97–98 Rodrı́guez -Almeida, Emilio, 77
167 protecting door openings, 86–87, 168 Roldán Gómez, Lourdes, 71
sources of, 14, 16. See also quarries: protecting vault, 89, 91, 96, 98 roof tiles
pozzolana sequence of construction, 189 for cocciopesto, 58
Pozzuoli (Puteoli) Remesal Rodrı́guez, Jose, 85 protection of extrados, 58
amphitheater, 239n28 ribs. See also names of individual buildings in ribs, 95
pozzolana from, 52, 54–55 bipedalis, 90, 91–98, 101 upright in vaults, 239n28
praefectus urbi, 21, 67, 243 bipedalis, at crown, 95, 239n18 in wall facing, 17
Ptolemy II, banquet, 50 conceptional change in application, Saepta Julia, 170
pulvis puteolanis, 55. See also pozzolana 102–104, 111–112 Sanctuary of Aesclepius (Cos), 6, 21
pumex pompeianus, 64. See also scoria: from development of, 91, 111–112 Sanctuary of Athena (Lindos, Rhodes), 6,
Vesuvius ladder, 90–91, 98–108 21
pumice lattice, 98–111 Sanctuary of Fortuna Primigenia
caementa of vaults list of examples, 216–221 (Palestrina), 9–10, 11, 49, 146,
Basilica of Maxentius, 63–64 network of, 90, 96, 105 173
Baths of Diocletian, Plate X, 63, purpose of caementa, laid radially, 59
126 to distribute loads within vault, coffers, 10
“Temple of Minerva Medica,” 110–112 dating of, 3–5
Plate XI, 63–64, 78, 162 to form divisions between work Sanctuary of Hercules Curinus
list of examples, 213–214 stages, 86, 91, 95, 99, 100 (Sulmona), 5
mortar of vaults to reduce construction time, Sanctuary of Hercules Victor (Tivoli), 5,
Colosseum, Plate V, 56, 60, 167 94–95 49, 146, 173
Domus Tiberiana, 187 to regulate creep, 48 caementa, laid radially, 59
properties, mineralogical, 64, 224 to transfer loads, 86, 90, 91, 110, ribs, travertine, 88, 89, 111, 112
properties, physical, 16, 162, 165 111–112 Sanctuary of Jupiter Anxur (Terracina), 5,
use in late antiquity, 172 related to centering, 34, 101, 107–108, 49, 146, 173
pumpkin dome, 46, 50. See also segmental 110, 167 caementa, laid radially, 59
vault relation to brick industry, 111, 112 scoria, 61, 62, 66, 167, 187. See also names
quarries, 14 structural analysis of, 162, 164–165 of individual buildings
granite, Mons Claudianus, 128, 170 travertine, 88–90, 91 effect on structural behavior, 164, 167
leucititic lava (selce), 16 rivers, 15 list of examples in vaults, 213–214
marble, Cararra (Luni), 170 Aniene, 12, 13, 17, 127 properties, 16
peperino (lapis Albanus), 13 Clanis, 17–18 from Vesuvius, Plate VII, Plate IX,
peperino (lapis Gabinus), 13, 80 Farfa, 127 61–63, 67
271
P1: JPJ
INDEX
scoria (cont.) “Temple of Minerva Medica,” 111, 144, suspensurae, 59, 65, 96
analysis of, 222–224 162 Symmachus, 172
as ballast, 66–67 Trajan’s Markets, 141–142, 144 Tabularium, 5, 28, 35, 113, 240n35
importation of, 59–60, 66–67 Stevin, Simon, 150 Tacitus, 170, 174
list of examples in vaults, 213–214 Strabo, 17, 25, 64, 173 taxation, 82, 170, 171
properties, 64, 222–224 stress, 7, 139 under Diocletian, 21, 67, 85, 172
segmental vault, 47 hoop tension, 139, 159, 160–161 Taylor, Rabun, 37, 45
centering for, 46–47, 50 structural analysis, 153–155 technology as display of power, 50, 128,
imitating hanging canopies, 50 absolute vs. relative applications, 155 173, 177, 179
selce. See leucititic lava; quarries assumptions, 155 technology, military
semidome basic concepts, 7, 149–151, 152 influences on construction, 166
as a buttressing element, 111, 144, elastic theory, 154, 155, 161 siege tower, 44, 45, 48
162–163 finite element modeling, 154, 157–158, Temple of Castor (Forum Romanum), 87,
buttressing for, 140 163–164 113
Seneca, 129, 148 funicular polygon method, 152–153, Temple of Concord (Forum Romanum),
on hanging gardens, 174 225–229 239n2
on luxury, 147–148, 169 graphic statics, 152–153 “Temple of Diana” (Baiae), 164, 165
on window glass, 147 limit analysis, 154–155, 165 Temple of Divine Hadrian. See
Sentium, window glass from, 147 middle third rule, 159, 241n16 Hadrianeum
Septimius Severus, 82, 171 plastic theory, 154–161, 164 “Temple of Mercury” (Baiae), 2
Sette Sale, 34, 190 static equilibrium model, 154 buttressing walls, 142, 143, 144, 147,
Seven Wonders of the Ancient World, thrust line analysis, 152–153, 155–164, 156–158
173–174 225–229 caementa, laid radially, 59
Colossus of Rhodes, 173 structural behavior, 6–10 centering, 40–42, 45
Hanging Gardens of Babylon, 173–174 ancient understanding of, 9–10, 155, deformation of dome, 40–42, 157–158,
Severan Baths (Palatine), 194–195 165, 167, 168 167
ribs, 100, 101, 103, 104 arch, 6–7, 132–134, 151–152 structural analysis of, 156–158
Severan Marble Plan, Plate VI, 5, 121, 133, hinge, 151 windows in dome, 158
137 barrel vault, 34, 132–134 “Temple of Minerva Medica,” 111, 163,
Severus and Celer, 174 cross vault, 34, 134–135 178, 201–202
shipwrecks dome, 34, 138–140 amphoras in vault, 78, 80, 83, 168
Albenga, 237n84 lateral thrust, 7–10, 136 bipedalis course in dome, 48
Cabrera 3, 75, 238n57 from cracking, 131 buttressing, 146, 147
La Madrague de Giens, 24 in cross vault, 134 cracks, 139, 145
Nemi, 24 methods of control, 116, 124, 138, pumice, Plate XI, 63–64, 78
Sidonius, 17 167 radial formwork, 110
siege tower. See military technology material properties ribs, Plate XI, 109, 110
Sila (Calabria), fir, 17, 18 elasticity, 151, 154 step-rings, 110, 111, 144–145, 162
spolia, 82–83, 118 stiffness, 151 structural analysis of, 153, 161–164,
lime from, 54, 65 strength, 154 168
Stabian Baths (Pompei VII.1), 237n13 monolithic quality of concrete, 7–8, 9, structural experimentation at, 146, 168
steel, 25, 48. See also Noricum; iron 139–140, 145 “Temple of Venus and Cupid” (Sessorian),
Steinby, Eva Margareta, 2, 178 semidome, 140 200
step-rings, 110, 141–144 stone arch vs. concrete vault, 7 amphora in vault, 80
Pantheon, 77, 141, 143, 159, 160–161 structural form, stability, 6, 10, 165 buttressing, 145–146
purpose, 141–142 Sulla, 5 “Temple of Venus” (Baiae), 46
structural analysis of effects, 160–161 suspended ceiling, metal bars for, 30, 31 Terme del Nuotatore (Ostia V.10.3), 81
272
P1: JPJ
INDEX
Terme di Sette Sapienti (Ostia III.10.2), radial formwork, 40 Van Deman, Esther, 59
240n35 ribs, 109, 110 on pozzolana, 55–56
Terracina. See also Sanctuary of Jupiter Tor Pignattara. See Mausoleum of Van Nice, Robert, 8
Anxur Helena Varro, 174
limestone from, 16 Trajanic latrine. See Forum of Caesar Velia, Porta Rosa, 87
tetrarchy, building under, 176–177 Trajan’s Markets, 191–192 Veneto, land reclamation, 74
Theater of Balbus, 170 Aula Vespasian, fund raising, 171
Theater of Marcellus, 170, 184–185 buttressing, 134, 136, 147 Villa alla Vignaccia, 72, 193
balancing of vault thrusts, 114 centering, 36–37, 39, 49 amphoras in walls and vaults, 69–70, 73,
caementa, laid horizonally, 59 cross vaults, 135–137 81, 167
relieving arches, 88, 89, 90, 91 caementa, tufo giallo, 61 Villa di Sette Bassi, 102, 193–194
travertine keystones and impost blocks, barrel vaults, arc of embrasure, 132 ribs, 99–100, 101, 102, 103,
87, 88 black pozzolana, 56 106–107
Theater of Pompey, 5, 235n30 roof tiles on extrados, 58 Villa of Domitian (Castelgandolfo)
Theodosian Code, 65, 67, 83, 108, 172 semidome brick lining, 234n33
Thode, Dierk, 158 stairs, 142 cryptoporticus, 203
tie bars, 115–126. See also names of step-ring, 141–142, 144 ribs, 95–96, 97
individual buildings wall thickness, 133 coffers, 97, 234n62
anchor blocks, 115–118, 119 transport, 6 “Villa of the Gordians”
Augustan, 115–116 cost, 18 Octagonal Hall, 199–200
combined with lightweight caementa, of building materials, 17–18. See also amphoras in vaults, 79, 80
63, 126 names of individual materials ribs, 79, 109, 110
concealed, 118, 123, 128–129 river, 12, 13, 16, 66, 67, 173 radial formwork, 40
cuttings, dovetail, 116, 117, 118 road, 18 segmental semidome, 47
cuttings, L-shaped, 115–116, 119, sea, 66–67, 70 Vinci, Leonardo da, 151
121 travertine, Plate I, 12, 16, 87, 113 Viollet-le-Duc, M., 45
exposed, 115, 116, 128–129 Trier, Constantinian basilica, 23 Vitruvius, 11–12
list of examples, 221 Tripolitania, olive oil, 82 on accustical pots, 70
spacing of, 125 truss, 22–24, 25 on arch behavior, 134,
surviving example, 114, 115, 118–119, centering, 37, 44, 45 155
128 composite tie beam, 24 on cocciopesto, 58
tie beams, Byzantine, 129 ancient depictions of, 23, 24 on design principles, 130,
timbrel vaulting, 30 tubi fittili, 50, 68, 181 133
tituli picti, 81–82 tufo giallo della via Tiberina, Plate I, Plate V, on fir, 17
Tivoli. See also Hadrian’s Villa; Sanctuary Plate VII, 13, 59 on hot rooms in baths, 49
of Hercules Victor; quarries: list of examples in vaults, 213–214 on lime, 30
travertine reappearance under Maxentius, 63–64, on mortar, 52, 55
market building, 6, 134 67 on pit sand (harena fossicia), 54–56,
Tomb of Caecilia Metella, 17, 239n11 systematic use in vaults, 60, 65–66, 166
tools, 25–26, 48, 167 167 on pumex pompeianus, 64
frame saw, 25–26, 48 with tie bars, 126 on siege towers, 44
pit-saw, 25–26 tufo lionato, Plate I, 13, 62–63, 66, on trusses, 23
saw blade setting keys, 26 70 volcanic districts, 13
Tor de’Schiavi, 199 tufo rosso a scorie nere, Plate I, 13, 61 Colli Albani, 12, 13, 16, 55, 64,
amphoras in vaults, Plate XII, 78–80, Tunisia, oil amphoras from, 69, 81 222–224
83 Uggeri, A., 202 Monti Sabatini, 12, 13, 55, 64,
decoration of dome, 199 umbrella dome. See segmental vault 222–224
273
P1: JPJ
INDEX
volcanic districts (cont.) Wilson Jones, Mark, 46, types

Mount St. Helen’s (USA), 237n83 130–131 chestnut, 12, 28, 37
Vesuvius, 16, 63–64, 66, window glass, 147 elm, 12, 24
222–224 wood fir, 12, 15, 17, 37, 49
Vulsini, 224 deforestation, 17 oak, 12, 113, 184, 186
Ward-Perkins, Brian, 21 supply of, 17, 172 pine, 49
Ward-Perkins, John B., 7 transport of, 18 Yegül, Fikret, 172
274
P1: KDF
0521842026fig.xml CB885B/Lancaster 0 521 84202 6 April 18, 2005 13:57
plate i. Examples of major local building stones used in vaulted structures around Rome. All are shown
at actual size. (Euro = 2.3-cm dia)
1
P1: KDF
plate ii. Porticus Aemilia (second century b.c.). Detail of opus caementicium vault.
Note the heterogeneous materials used as caementa and the lack of the distinctive
red pozzolana found in later imperial mortar (compare Plate VIII).
plate iii. Four types of pozzolana showing the difference in color between them.
The Campi Flegreian pozzolana (pulvis puteolanus) at upper left is easily distinguish-
able from the other pozzolanas quarried near Rome. (Euro = 2.3-cm dia)
2
P1: KDF
plate iv. Section through an abandoned pozzolana quarry on Via Nesazio (near Largo Martiri delle
Fosse Ardeatine) showing the stratigraphy of the three different types of Colli Albani pozzolana.
plate v. Colosseum (a.d. 70–80). Detail of mortar with white pumice used in the upper level vaults.
“TG” = Tufo giallo della via Tiberina. “Pu” = Pumice.
3
P1: KDF
plate vi. Baths of Trajan (a.d. 104–109). Reconstructed plan. Detail insets show location of ribs, tie
bars and different types of caementa. Dotted lines show outlines of preserved fragments of the Severan
Marble Plan. Location of Plate VII is indicated with hollow arrow at section H.
plate vii. Baths of Trajan (a.d. 104–109). View of section H

showing the use of the lighter tufo giallo della via Tiberina in the
semidome and of the heavier tufo lionato and brick below impost
level.
4
P1: KDF
plate viii. Baths of Trajan (a.d. 104–109). Detail of vault from section E showing
tufo giallo della via Tiberina and Vesuvian scoria laid in alternating courses. The
pozzolana in mortar consists largely of pozzolana rossa. Note the variation in both
the vesicle size and the color of the scoria.
plate ix. Baths of Caracalla (a.d. 212–216). View of different types of caementa
used in the semidome of the exedra of the west palaestra. From bottom to top:
brick, tufo lionato, Vesuvian scoria.
5
P1: KDF
plate x. Baths of Diocletian (a.d. 298–306). Detail of the east palaestra vault. Gray pumice
is visible at the crown and tufo lionato below at the haunch.
plate xi. “Temple of Minerva Medica” (first half of the fourth century a.d.).
Detail of dome. Dashed line marks change from tufo giallo della via Tiberina below
to the darker pumice above.
6
P1: KDF
plate xii. Tor de’Schiavi (early fourth century a.d.). Top: Detail of painting by Henrik
van Lint (1684–1763) showing the remains of the pediment with amphoras embedded
(Alinari/Art Resource). Bottom: Reconstruction of building showing section through
pediment (following reconstruction by J. J. Rasch).
7
P1: KDF
plate xiii. Plans of the Baths of Caracalla (top) and the Baths of Diocletian (bottom) showing the
location of lightweight caementa, the location of tie bars, and the differences in the types and distribution
of vaulting ribs. Red indicates bipedalis ribbing. Green indicates lattice ribbing.

GDA Arquitectos

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

GDA Arquitectos

Hochgeladen von

Copyright:

Verfügbare Formate

P1: JPJ/KNP

0521842026agg.xml CB885B/Lancaster 0 521 84202 6 April 18, 2005 20:36

This page intentionally left blank

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

Lynne C. Lancaster is assistant professor of classics at Ohio University. An architect and

CONCRETE VAULTED CONSTRUCTION

Cambridge University Press

© Lynne C. Lancaster 2005

This publication is in copyright. Subject to statutory exception and to the provision of

First published in print format 2005

isbn-13 978-0-511-16068-4 eBook (EBL)

isbn-13 978-0-521-84202-0 hardback

List of Illustrations and Tables page xiii

2 CENTERING AND FORMWORK 22

3 INGREDIENTS: MORTAR AND CAEMENTA 51

6 METAL CLAMPS AND TIE BARS 113

7 VAULT BEHAVIOR AND BUTTRESSING 130

8 STRUCTURAL ANALYSIS: HISTORY AND CASE STUDIES 149

“Temple of Minerva Medica” 161

9 INNOVATIONS IN CONTEXT 166

appendix 1. catalogue of major monuments 183

appendix 2. catalogues of building techniques 205

appendix 3. scoria analysis 222

appendix 4. thrust line analysis 225

LIST OF ILLUSTRATIONS AND TABLES

Frontispiece. “Temple of Minerva Medica” in the Horti Liciniani.

LIST OF ILLUSTRATIONS AND TABLES

29. Detail of cross vault from the Colosseum. Rome. 38

LIST OF ILLUSTRATIONS AND TABLES

74. Sketch plan of cryptoporticus at the Villa of Domitian. Castelgandolfo. 95

LIST OF ILLUSTRATIONS AND TABLES

110. Reconstruction of lintel construction at the Hall of the Doric Pilasters at

LIST OF ILLUSTRATIONS AND TABLES

C oncrete vaulted structures represent one

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

Palestrina (ancient Praeneste). The sanctuary has been

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

So, what prompted the early development of con-

ning a distance so that the stresses within the material

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

With the increased interest in preservation

During the second century b.c., when con-

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

way in which they relate to the local trans-

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

R oman concrete vaults are praised for the

CENTERING AND FORMWORK

In any complex structure involving joinery, the

8. Reconstruction of wooden centering for concrete barrel vault.

Trusses were in use at least by the first century a.d.,

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CENTERING AND FORMWORK

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CENTERING AND FORMWORK

CONCRETE VAULTED CONSTRUCTION IN IMPERIAL ROME

CENTERING AND FORMWORK

sesquipedales with a grid of bessales covering their joints

the use of brick linings on the intrados