Sie sind auf Seite 1von 16
Pern een AJ Lb ISLAMIC STAR PATTERNS. ‘The geometrical ormament of Iain hae fascinated Western observers for well over 8 century, and con tinues to provide material for namerovs wien t= tempting to explain its nature in esthetic, mathernae seal, mystical, or even comological terms. The "geo" ietical” content of this ormament i obvious, but in spite af the fet that many works have been published fon the construction and analysis of Hlamie pater none bas delved very deeply ito the subject, and no comprehensive work has yet appeared dealing i9 a fomparatve and aystematic manner with the whole range of paterns, both geographically and historically.” “There is dearly 4 need for a more scientifically based sccount ofthese patterns, though it not 0 mich he Theoretical Knowledge of the professional mathemati can that is required for such a study, asthe precise mathematiea! language and rigor which e might bring to bear on the subjec. There is sll no-gevevaly accepted terminology for the many diferent kinds of motifs used in Islamic geometrical ornament, noe for the metheds of forming repeating pater fox them Tn the absence ofa definitive work along these nes, 1 ‘will ateaspt ere t show ways in wich some of the sitmper patterns might be develeped from x numberof ‘ementary principe, Opinions hive diferd_as co the part played by mathematics in the genesis, development, and con struction of the more complicated Islamic pater.” Some have seen a gradual, convergent evoltion from ‘many different typer of prelsamic _ormament, Clminatingin the often bewildering completes ofthe Jater, fly lifereniated Intamie patterns. According 10 this view ic was presumably the widening practical experience of skilled rafismen which alone chove the Dts taken at each level of increasing elaboration, (thers. an the contrary have seen clear evidence for the intervention of the profesional mathematician in the design and invention ofthese patterns, or atleast Ihave ateibed tothe arian themselves considerable Inowielge of thecrtial geometry and an ability to apply this knowledge tothe development of ew Kinds of geomeeeal ornaient We shall perhaps never havea final answer to thie controversial question? yet it seems that Western erties have serioury underestimated the ability of hatve craftsmen to rei large amounts of empirical Inowledge on pattern design and construction in the Absence of any undershnding ofthe theoretical back ‘gonnd which a. professional matheratcia might bring to beat on these froblems It may be an advan tage for 3 modern atthor to develop a systematic analyst of Islamic pacene in purely mathematical teem, but a Kaowledge of pire mathematics of seomeuy is unnecessrs for those who wish merely raw Islamic patterns o iavent new ones theoretical Dackground will often dlow the artist to see number cof combinaterial possibies more quickly than the use fof tiakand-eror methods, but it forms no ubstute {or tre critvity. Its prhape sigan that a geo fine application of mathematial insight to a eysematic analysis of Islamic geometrical patterns reveal far freater range of pombe than were cverdacovered By the Mins themadves 'A great deal ofthis ormament is immediately and unmistakably recogniasle a "Tslamie," and yet in ts fentrety it doct not fort such an easily dsinguishable body of geometrical omament Ia fact dhe whole range oflslamie patterns represents sn amalgam of many di ferent yes, some singly adapted and absorbed from lassi sources and fem various culkures with which Islam eame into conti during is early expansion Since itt not possible» coverall types of Tne pat terns here, [wl init myself tan exansinaion of eer tain ofthe more inereting and ‘ypicl of them, “There is one clans of geometial patterns which Islam has made its ows. ‘Tie group comprives what ‘one might term the sur patterns," since they indude starlike moti, linked or oriented according t certain Precise rules to produce endlessly repeating two ‘mensional patterns. The sat patterns are noques: Yionably the most beautiful and inricate of all Islamic patterns, and they ove their beauty in no small rearure to a high degree of symmetry at all levee, Thue, the star motif toemelves invariably posses {old rotational symmetry, # representing a range of whole numbers from 3 to almost 100. Although any pattern which repeats in two. dizections must of ecesity pertn to one or another of 17 fundamentally Gitinct arrangements? # claafcation by means of fuck symmetry groupe i of Kile use sn a detailed fnalyst of dame geometrical patterns. A given Islamic pattern wil equetdy employ a small number of precaly determined shapes, some of which become ‘repeated within the pattera in many ways nt allowed {or under the four elementary operations, or isometries, of classical lane symmetry; each ofthe latter must act fon the whole two-dimensional plane, noc merely locally fon small parts of it. Simiaey, although many mois ‘themselves possess higher than 6-old rotational sym- retry, they cannot form repeated centers of similar rotational symmetry in the plane a 4 whol, since the yermilted centers can have wo more than 2 5,4, or Gold symmetry (such rotocenters” are termed respectively dads, wads, eras, and lexads), On the surface ofthe spre the restrictions are somewhat di Ferent, whereas. in the hyperbolic plane. viewally Anything is posible? In their simplest form all Islamic geometrical pat tems are examples of period tilings (or esllasons) ofthe twovdimensiona plane, consaing of polygonal ‘reas or cals of various shaped abating on neighboring ‘ells lines termed the edges ofthe tiling, and with Ure oF more cals meeting at points termed the vee tices, or nodes, ofthe ting In genera, the cells ae not Fequied to be convex polygons, nor the edges to be straight lines, and there is no resection onthe number of edges of cells meeting at each vertex." From the eaves timer Islamic ornament adopted the widespread inteaing band form of linear decoea ion, whereby the original Hiner of the patern are represented by straps or hand, executed ia such a way fr to give the impression of weaving alternately over tnd under one anoher. This style of ornament had its ‘origins in antiquity. and must ultimately derive by imitacion from various types of weaving, plaiting, or bbrketwork. Ae an arti device it serves to give coe: ion to.s whole design. We may note that an inter Tacing-band style can only be achieved. when the ‘riginal ting consists entirely of way nodes, ti four edges (and therefore cll) meet a every node Weally, opposite angles at each node shouldbe equal, whic meus thatthe four edges esting atcha node Thecome a ait of fines intersecting ata comover point Paterns consisting entirely of fay nodes may be 18s referred to as tru iaring atoms, whether or not they are deawn au intestacing bands Trtcrlacing patterns have another important prop- ety: in ther linear, af epponed to interlacing-band, form che cell of any pater may be colored aernately Jn two modes—say, bck and white—so that no eo calls of the same mode meet at a shated edge. A chessboard i familia sxample. Strictly speaking, the possbilty of « ¢wo-mole coloring i inherent in any ‘ling in which an een numberof eels or elges mecte at every node Nowinaring atin conti least ome nodes which are rot way. It some of these are fodd-oummbered, then a re-mode coloring is no longer posse. The sar patens in general include examples from oun interlacing and non-interlacing eategorie lamic star motifs owe this beauty and regularity to 4 featare which they share with the regular convex polygons: all derive fom sete of points equally ‘stituted around the dreamference of circle, When Pris of adjacent points im any such set ate joa by Straight Hines anil a single circuit incompleted, the ‘result isa regular conves polygon. However itis post ble to continue joining up every other point, or every ‘hind point, and 0 0n,w produce many best stat like figures In loose sense chee may al be termed “star polygons,” altkagh a star polygon, properly speaking, celts only ven all the lines «0 produce form 2 single circuit strounding the center of the Figure more than once. The set of pots om the initia ireumference comprises the eter ofthe star polygon, bat the sides ofa star palygon intersect one anther at various addional pois, which are not counted as ver= ‘ies, although they dive each side into a number of segments “The earliest Islamic sar motif were based on » star polygonal construction, but complete star polygons were rarely weed ss ornamental motif Casall in iol ‘don, st medallions). Intilly nich constrution pro- duces a space at the center ofthe figure, i the shape ofa regular polygon (fg 1). In authentic Islan orn tment this cenval space is usualy transformed into star-shaped area by the sisson of one of more of the riddle segments on allies ofthe star polygon. Us ally ll but the lst two segments each end of side fre omitted, thus producing the typical Iai star mot fig. 2. This figue therefore consists ofan inner call, or central star, anda number of outer ells in the ‘shape of kites. Occasionally all but the outermost seg ren are omited (fg. 3), ad we thereby artive at he Simplest form ofa reguar geometrical sar es Star moti of these types canbe ditinguished by 2 concize notation, giving data on three quantities: the ‘puber of initial verties; the meted of pining up the vetices to produce the original star polygon (ie. #60 by two, three hy three, andl soon}; sd te number of end segments remaining at each end of ee sides ofthe ‘sar polygon, Those three quantities ca be represented by md, and s,respctvey, and dhe complete symbol forthe basic Islamie sar as afd. Thus, the star shown in fig. 2 ca be designate an (892 (this may be ead san “eight over thee, two-segment tar”). Obviously this notation, whichis derived fem the mathematical sotation for sar polygons, is appliable to any Larmic "ar construct by drawing straight lines between pals of points on a circle. Indeed, it may even be adapted fo certain other types af construction, Hf we allow non Integral values for ‘Motif bated directly on star polygons (sens. lat.) are ‘eadly constructed uring a single sce and a nambce of prints equally ditributed on the citeumnference ofthat Sle. Noe inal construction is necesary. Many later Isamie star mosis, however, are not derive in this way, although they thay sl consi of central tar land surrounding kites, In there later star mots the lope ofthe ines forming the sar esc tha they can ‘ot be constacted siemply by joining pairs of points on the cieumscbing cede of the moti. Th these eases, ‘one oF mare additional concentric ctcles are needed 10 ‘determine the inner point of the star motif and hence te complete the lines (ig. 4), Islamic snr motte are of ‘any diferent sypes, and the main varieties will be indieated belo, but ll include a simple nae a8 8 cen teal cel, The precise metrical properties of any star ‘motif may sometimes be arbitrarily chosen, but fe ‘quently depend on geometrical considerations concera~ ‘ing the relation of the mif to otber elements in pattern "The tymaetscal properties of any rgular pointed star motit oe wsided regular polygon (» being sy ‘whole number greater than two) can be represented as 1 aystemn of 2n rad diverging from the geometecal enter ofthat moi, This figure convenienly tered ‘iar; iconsists of wprincipal rai, dough the tain, ater point ofthe mot, ad w secondary rai iterating with them. The continuing invention of fundamentally new star patterns ental a search for ll suitable areangement of sar-centers in the two-dimen ‘Sonal plane, but dhe tunuber of possibiides i imited bythe precise way in which cach sarscenter must be coviented in relation eo ite nearest neighbors. I the fof bes Cex Fr OUW®€X method of linking starcenters that is preferred above {il others jn Islamic pattern, one radius fom cach tarcenter of neighbering pair x coinident with the ‘trsight line joining thr centers. Inthe case of a pair fof sar motif, their cenvers and shared point of contact, Tie om a single straight line (ig. 3). This i therefore termed a cling link ‘Another sceeptable method of linking nearby star cemters is by having their eaves adit parallel, rather than calliear. In thi cae the straight ine joining ee ‘enters no longer coincides with» radius of either star; ch a relation may be termed a pall! fink (Hg. 8) Parallel inks between tar mous are principally found in certain derivative puter (described below) which ate usually fx more lieu to costeuct with ruler Sn compast than thee wing collinear links. (I is probable that mow of tase derivatives were originally Composed cough cearangements of certain elemen tary mova shaper-for example, cit tiles or pieces of ‘wood inlay) Rules sich a these for linking adjacent star motifs swore never expicily stted by Muslin artists, but were {plied probably unccnsciously, in large numbers of ‘aried pateras througout Islam. Ideed, the employ ‘nent felines Hinks join star= or Newer ike motifs stedates their use in Islamic patterns by many cea: Curie and has probably always been fl inatinetvely to represent » mote elegant method of pattern con position “Many cacy forms of repeating pattems were based fon simple Tractional divisions of underlying grids of ‘ilateral anges or squares, Both of which require {he ruling of sete of parallel lines right eros the pat termed area. It wast sch grid thatthe Rit designers of Telatie patterns eure fo produce the eats sat patterns, The simples procedure isto canter a star Ino ou every verter of the gid, with a radius equal to half she ecg lenge ofthe grid paygons. The mots te then oriented ao that certain of thee principal radi fwincide wih grid lines, ad each pale of adjacent Imotls meets at the midpoint of an edge of the grid ‘Thus, collinear Hinks ate automatically established, Since the triangular and square grids themselves have respectively & and Away verte, the simplest movie whic canbe used in his way wil have 6 of & points Ta fae, i should be obvious that motifs with any ‘numbers of point which are itegral multiple of these ‘values con be similarly placed on the grid vertices. * 185, Win oteasy to estabish the exact date at which sims ple star patterns ofthis kind were frat used in Islami chitestural decoration, Some attempts have been made to explain the histrieal and developmental ‘origins ofthe first rectlinear star patterns,” but since ‘curviinear versions of many of these cceurred even ‘adie, perhapa one sbould really try 0 explain the forgins of the latter. In any escussion of historical origins, we mus beat nid that although ti til theoretically possible © locate the eaves surviving version of every ditinet patter, i cannot necessarily be ataumed that any one of chee realy represeats the first hintorical occurrence of that patcular pater, It i more thin likey that most eaely examples ofthe art have long ago crumbled to dust” Nevertheless, sa ‘Seat material la survived fom dhe eighth fo the tenth fenturies (0 give @ atalzing glimpte of what must have been an extremely rich fund of very easly geo- retical oenament. Rewgnizable precursor of the first Star paterns started to appear in the Middle East as ‘arly asthe beginning othe eighth ceneury, in the form ‘of open work window grilles inthe Great Mosque at Damascus (713)," the palace of Quer al-Hayr al- Gharbi in Syria (222, and the palace of Khirbatal- Maar in Jordan (748). Although some of these design coniain starlike clement, ss ot always dear wether they representa distines mots i the atis's ‘original conception, ot whether they arose merely at residual spaces between groups of overlapping ccs, CGreular ares, or oer shapes. In some cases, however (or example, that shows ing. 7, of which almost iden- ‘eal versions surive frm Qasr alHayr abGharbi and Kira al-Mafjar), we are probably correct in inter ig, 7 A pte rm he ip cer lh carlin Spit 186 Ay tee pretng the pattern as consiting ese of an array fof star motifs (i this case, of curiinetr pointed fan) in conte. Two ofthe eae repeating piteroe using simple rectilinear stars cocsed of 6 and pointed star, ‘eapetivey (Giga. 8 and 8), bth of which had thet ‘nig in classical ansiquity" Ta an Islamic context the Starand-ross patern (Fig. 9) frst oceuts in ninth fentry Samarra" version of the pater with pointed sae occurs inthe move of fon Talon at Cro (876-79) while both cam bese in te fom the palace of Madina al-Zahea (09676) in Spa,” in alain onquein Gao (1003)" However, the aand-ron pattern probaly speared even cA” ig, 69. Tw of eit rein pate ie, snce arctica version oscuro one ofthe win does from seal Hayr a Gharbi in wich a Tine seaments are extended ae straight ines ronning in interiaced form right though the pattern. Another ‘ample ofthe patter with pointed war ecco inn Saay-leventrerntary house excavated in Sia "Thaw by the endo the tenth century mt the Its patter containing and pointed reine start rere Uy widespread, and i probably required ile inetectual effort onthe part othe origin artists to conceive the idea of incorporating simple stars with treater numbers of points in cepeating pater Kit not un the second hal of he eleventh cenry that tramps survive, however, A pattern with 12-pinted ‘nn ruperimpored nm wlangalar gid occur on the Chir ofthe two Kharragan tomb towers (1067-9 [X21 NL in norwestern Iran (i. 10) A related pacer, cing teint deka thee (12°92 tan ba ote sare i i. 1 7 a tight be expected obave been discovered at about the ‘ame period, but no ery examples apps o haves vived, These pattems were later among hemos wide- toresd of sll_xar pater, they ate found fom Morocco to Central Asi, ndcaingin both cas very ‘arly dicovery an diseminaion. Snialy a patera Of(@/32 sar fig 12) had probly been ccovered at thin time, since the bree sar vl woe aed ‘tine, bt gain no erly examples appen fo hare survived, though what could be rgnrded te 2 Cue vlna version fom th igh entry inated fg. ‘The patterns developed s far imcditly trate an importane feature in the execution of ney all Isami tar patterns. When the number of plein the stasis greater than he number of ines ating from Vv ‘ch node ofthe grid om which they are superimposed, there remains the prem of what to-do with the 1 pend on gi me “eee” vertices of each star, Le, those notin diect contact wi neighboring stats, The simplest and most ‘eyant solution is fo continu the edges of the stars ‘hough each fre vertex until they meet similarly pro ‘duced lines from neighboring stare. Ta the cae of the square array of Bepointed tars (ig. 12) this resus in ‘smal intertial pointed star, whereasin the case of the triangular array of 12-poined tars (ig. 10) a stall ‘qullateral tangle produced, In these two pater ‘each star coneributes only a single fee vertex in each sd polygon, andthe teal number of fee vertices in ‘cach polygon is obvioutly equal othe numberof sider in that polygon. The majority of sar patterns obeain a greater measure of continuity between thei constituent Sar mots by bridging the space between lee vertices inthis way, although the arto ix always at Hiberey not 10 do 00 i he wishes. The additional lines thus pro- ‘duced by joining up such fee vertices may be eelerred to as constituting the ional pate, since they occur in the residual space between groupe of three or more neighboring stats. The greater the number of free ver tices available in thie space, the more complex i the incerta patern, "The nature and symmetry of the ‘ements inthe interstitial patter are elesely related to the symmetry of the pattern a2 whole, ar well art the Iyper of constituent ar motif and thir prette com Paterns using 10-peinted stars occur in the now dome chamber of the Masjid Jami® at Isfahan (1038) ‘These “decagonal” patterns represent & ‘departare foun traditional methods of geometrical pt term design, although they result inevitably fom a logical generalization of the principles developed so far =e x a CRISS om: L se | DA Lhd eB 187 Ie is not immediately obvious bow to form repeating terns with 10-poited sary, since neither ofthe grids Considered sari able, Ifthe tars are required 9 remain in contact, using eolinear inks, however, i Found that there is ony one simple arrangement posi ble, and this could easly have been obtained after & litle rial and error (6.13). This pattern and the new gid which underlies require «more detailed analysis (Geich wil be given below) thar leads tothe formal Aliacovery of rany Kinds of patterns, These inlade Simultaneously wo kinds of star mot, nchuding some ‘ofthe most widesprel a all amie star pattern. I it by no means suggested, however, that the original dlscovery of such patterns fllowed ram the methots of analysis outined here: Rather wore elaborate patterns ‘th 10-fld motifs occur on the later, so-called Victory ‘Tower of Mand IIT at Ghazni in Afghanistan (prob bly dating from aroun) 1100) This srocrre also presents what apposss to be the fist oecartence in Islam of geometsica patterns incorporating 7-10 ‘motifs (in fet dese same patterns alin nce 20+ and ‘epointed stars, expectivey) ‘These “hepiagonal” patter, in which the propor: tionsof the bisc repeat are determined by angle whic fare mulkples of 180°7, ave extremely diffi ta work wth. This iicuy is rected in the paucity of amples throughost the whole range of Islamic geo metrical omament. About 6 percent of Bourgoin’ el lection™ const of tue heplagonal patern, and in this respect iis probably representative of Ime pat ferns a6 a whole. Indeed, pattern sing many other kinds of odd-numbered starmatfs are often very di Sule co incorporate in repeating patterns, and were 188 wp therefore seldom or never used. This is particularly so in the ease of those motile in which the numbers of point are prime nambers—say, 11, 15,17, 19, and 0 Belore analyzing sme of these new kinds of pat terns, we must Git conser the discovery and develop ‘ment of what is pethaps dhe most typically “Llanie™ fal sar mots the email rte (Bg. 18). The pro totype for the general weaved rosette alrnotceeainiy onsite of» pointed sar surrounded by sx regular hexagons (fig. 15). This configuration becomes sito ratiealy incorporated in the pattern of pointed are Geseribel above (Bg 8), but the prototypical Sayed Tete sctnst have been use fr the fits! time ae fdstinet moc on the Arab-Ata mausoleuan (078) at Tim, in Unbekistan® (fig. 16), and the same pavers reappears, with slighty diferent proportions, on the eather ofthe two Kharragan tomb towers.” A method fof constructing the isolated 6-rayed rosete ean be Gerived frm the consiriction used to produce the Ge & TOG COG OR oe oor 16, The cao ocurance of «Ge ete a dere ‘see th ey oh orginal pattern of pointed stars (fig. 17). The centers tf the (xx) surrounclng erp! tar orn the vertices ‘A ofa lining lyon of oe rset, and cieces centered ‘on dhese vertices, with radius AB, equal to hal the ‘dg length ofthe limiting plygon, eerie the pos tions of points such as €, and hence the radius CD of the interior sa of the sett. The hexagons may be termed the outer ile of the eowette; the terminal segments ¢ of these oute cells are obtained by drawing {eeaight lines beeen such points as Band Ey hil the Sides of the outer eels are obtained by drawing ines through points such as C, parallel o a principal radius [BD of the rosete. Ln such a rave the outer cel of the rosette may be said to posses illo terminal Segments, and paral sles" Alternatively the rosette ie ean be deseribed ae @ paralesided rose with collinear terminal segments, The construction just given has purposely wade no reference of-fold symmetry ort abaolute angle sss, to itr eapable of immediste generalization to include ‘geomewealrosetes with any numberof rays (fg. 18), provided only that the Fimiting polygon it a regular polygon. A'gcometsical rvette of thi kind will sito ‘matically be produced, with collinear terminal seg ‘ments ad paral! sides Tn aditon, although iis not ‘immediately abvious, serene a bare anys ual (this ean be proved quite eal. The outer cell of the general mrajed rosette are always symmetrical hew Sone, but only when n = 6 are they regular ‘When is greater than 6, the geometrical rosette i early always constricted «0 as to inchade a 2xegment star, ae defined above, in the ctle with radius CD (Bi 1). This may be termed the utr sero the rosette, and his in ts turn contains the ne oF seta rar With regard to constituent polygonal areas, the waved rosette now consists of whexagonal outer ell, kit shaped midde cls (or mie), and nn pointed cen tral star Since the angle BAC is greater when » is greaterthan 6 (ef. igs. 17, 18) iin ebvious that exaealy eqular 6;pointed peripreral stay are only. powible when n ="6. For higher values there i not enough oom fora perfetly regular Bpointed peripheral ta, bata regular 5-poined peripheral star only becomes possible when m= 10, Furthermore, since the ange between & pair af adjaceat side of any regula polygon in never 180°, it it obvious that regular pointed peripheral stars are impossible, with the properties een above. Thus, S:pointed peripheral” stary Inevitably come to be included in maay pers cone taining geometrical roses, and although dey can be ee Eadie weomeloaaien ag regulary formed in relatively few cass, ideally they are made as regular a is posible without reducing the fymmetry of the main rvetes. The highest degree of regularity ix achieved a8 follows: the ower pointe of a peripheral sar amt Hie ona cle, the center of which {ea vertex ofthe limiting polygon ofthe parent rete. the angles its outer points should be equal; the sides fr segments bordering all outer points should be equal; tnd the peripheral sar sbould form elliner links with the outer tar of ix parent rosette), and with neigh boring peripheral ars. Furthermore, ip this most reg ‘lar form, the bitecors ofits outer angles meet at 4 ingle pt, which may be termed the center of the peripheral star, since it coincides with the center ofthe Greumsesbing cele. I's pair of peripheral sas i shared Between (wo equal sete which ae joined bya clinar link (is 19), then tis obvions dat not only do the roses share tn outer point B, but their limiting polygons share an edge dy in other words the edge length of the wo Timing polygons i the ane. In fat, pe of Tink 109 rosettes could be constricted on the basis of pair of ‘regular polygons sharing an edge, and indeed the con "Hruetion is easly gonertzed eo ilo different Kinds (of regular polygons, ar therefore rosetes of different ‘Soe, but in this ease only one ofthe rosestes can be Constructed suey according to the principles we have ‘established so far. The slopes of the various line seq- ‘ments of the other rset are then largely determined by those in the ist ste, but the sides and terminal segments ofthe ovter cls ofthe second rosette can ill Ibe constructed as equal lengths, i required, by noting that the shoulder, or subterminal point (Gy fig. 18) should always ion thebnector af angle BAC In order to produce repeating patterns with geometrical ont. ‘one might then sare fr teselations or open arrange rents of regular polygons in which pare of adjacent polygons share an cdge. These are the essentials of what might be rermed te “polygons in contact” (PEC) tmethod of constructing certain Islamic star paterns that was fist enunciated as a general principe by EH Hankin although it was cceasonaly used eater by Bourgoin. However, in most cases iis not necessary to draw a complete arangervent of limiting polygons since the constrvetion a just one shared edge itr ‘ent to determine the wlatve circumradi of pai of fijacent rosettes, and subsequently the rat of heir ‘ter stars, In fact thee ae other enact constuctions posible which achieve the same resule without using ‘he limiting polygons sal, "The east paterns with geometrical rosettes were mostly consiructed with the properties, if pot the methods, outlined above, ie, with collinear terminal segments and pale der. Apart Irom dhe ease when , rosetes are completly absent ‘ror the ‘xtremely rch ornamestation of the two Kharragan Tomb towers (1067-68 and 1093, respectively), but feumetrical rosettes wah 10 faye occur in the north Home chamber ofthe Masjid Jani, Isfahan (1088) Rowetes with 8, 1 12 and 16 rayt occur in various perms round the tibrab ofthe mosque at Basia, Isfahan (1154)"—and, nidentally, include the east example oa remarkable pattern with 45, 67>, and ‘pointed stars." Patteme with and Hsrayed roves fceur on wooden minkare from the moague of Als Din at Konya (1138), and from the Ages Mosque at Jerusalem (1158) "1 ems posable, however, thatthe ‘old eave was generalzed to include @, 10+, and per= ‘haps 12-rayed rosete well before the end of the dl eth century, and that the majority of ealy examples Ihave simply not survived, Amare thorough search of 190 foe literature may bring to light many more fol star-centers les from this early period. the relevant dated may then be Both varieties with respectively the outer sets aze reduced t te form (10/41, Unfortanately itis ot posible to enter iato con sieuctonal and other deals here, 0 farther pattern ‘ppeson the same basis can only be briefly described sd ilirated. Types TV (Gg. 28) and V (ig. 24) are ‘imply laboratons of type I pater and are confined to Central Asia. Type VI (ig. 25) i alo derived from ‘type I this ime by expanding each kiteshaped eal ‘in ic overlap ie neighbors i small rhombs. Initially ‘he si of the expanded cells can be ariteaelly chosen, bu in eerain derivative f this type the relative sizes the pater ells become righly determined. A fe tquent dition to this patten is the incorporation of {ype Il rosetes a the center ofeach type VI motif (ace fig. 25, right se). Types VIE lig. 26) and VII ( 27) are ao Central Asian snd are ultimately derivable from elements of types Tand I and ther later elabor tions. In type VII all star-centers are surrounded regular deeagons, but oly altemate centers of type VIL fare 90 murrounded. ‘The interior of the decagons are ‘artably tested ip authentie sourees. The fae that pe VIT patterns simultaneously contain two kinds of centers means that either kind can be given prominence ina small partern area, cteaing quite different effects ‘The same remarks apply to types 1X (fig. 28) and X (lig. 29), akernate centers of which employ regular ‘eeagons and 20-gons, respectively. These last Two, land many related patterns, are almost exclusively sed in wooden Ince work. Types 1 (Bg. 30) and XI i, 81) again form a related pair of patterns, with similar peripheral clement, Other varieties are possible, but these telve types ate the mest common Variants Th addon there are many derivative variations, some of which are described below. ‘There is some justifeaton in giving distinct designations to chese ‘variations, since analogous treatments were applied, oF can be applied, to many other arrangements using sar motif of diferent sien, Similar variations unaythere- fore be given the same type designation, irerpectiv of llferences inthe number of points inthe costtuent sar maui “The pattem of 10fold sta-centers which undesies all the foregoing pattem types is formally related to a up which shares the same structural basis a6 the 2/108" chorus, with respect the patra of prin cipal and secondary rai within the som. The 60 sex or dngonale ofa rhombus dive ent four equal Fightangsl triangle (lig. 21) and in the present ease ach has, in addition eo the vight angle, interior angles of 54° and 36°. Radi fom the sarcenters oo these two vertices divide their respective anges into dhvce m1 a 14.1.8 pte th 8d mio Tp UO eh = am Istanue stan paris 193 3 dyiont Pedy ( Nye? divisions ig 32 hon afte 2/108 sh no ea ange of 1 stag (2) som. 8,16; and 7,28 (in each case the center ie the fist of the pair of value)" These solutions represent in fet a series of rhombs in which m-fold and fold centers occur in opposite pair on alternate vertie of each rhombus. We cannot Immestinely build up repeating patterns with these ham, however, since mone of ther wil frm a sin ple ged like the paterr of H0-Eld tarcenters with Thich we began, But pttens incomporsting all except the first and lant solution have been used inauthentic formanment, although only four of them—18,6, 12.8, 10,30 and 9,12—ean be used by themselves to form repeating pater) tke others require aditions Shapes. The beauty of thie sere is that similar varie tions, based om the decagonal ypes IMU that we have brielly described bow, can be adapted to each ‘hombus in the series, Tacoetclly, one might there fore expet atleast 96 dtinet lone, but in caee where the faulting staremotf are {oo dissimilar in size the patterns are often not very satisfactory ‘Using the same symbols as before, we may refer to serice of 39) shombs, and the specific solutions of uch 3 series may be expressed in the form (pq) "Thus, the chombe in the grid shown in fig. 24 become (@x3}10,10 shombs, The general notation for par ticular pattern type in the (3x2) shomb series may be ‘wren at (3% Z)mn/T, where T sands for one of the twelve dant type gen above. Each rhombus inthe pattern of fig 20 thus becomes (32)10,10/ Strictly Speaking, of course, this notation ought to refer tothe ghtangied triangle wich constitutes one quarter of hombus, but iis nateral (extend io include not only the complete thombus, but ala if necessary the two kinds of roaeles tangles which an be predced {fom two uc identical right angles. The contest will mabe it clear to which we are elerting at any particular ime. It must be printed out that the notation developed so far des ct designate a repeating pattern taut merely a potential dementay unit for one oF more repeating puters, Adcional symbols are requced inate ways in which nich elementary nite ae incor porated in epeatng paterns, but this unfortunately {question which canna be purried here ‘As we have remarked above, the decagonal types T- TI for special cave ofan infinitely variable see, in “which the oo lover cells are congruent 0 the ‘outer cells of the star motif in each case, A similar property it charactersic of other (332) rhombusce trth-disimilar star number, but i these cases the 194 na inter! cells cat be made congruent to the outer ‘als of only one ofthe two moti, nately the mente (Glthougt thie congrucnce was! early achieved in Authentic patterns. Thins an inescapable consequence ofthe undevying geometry ofthis seris of thowab ad itis eapable of rigorous geometrical poof (2 similar geometry is exhibited by the (2 1) rhomb series. Fa ‘Samentalcstinctions ofthis nature between the m= and toenters of the (3 <2) and other thomb serie were teldoa understood by the original atin, and have t= ainly not been appreciated. by Western authors, ‘Among a number of lines of evidence leading to this tondusion, we may cite the many difereny com ucted versions of patterns using the (3*2)12,8/11 thombus in existence (the “rhombus” being in Fact & square), some of which indicate chat the attists com cemed had no idea ofthe core” method of con: simuction*® The PIC method wil eetainly allow the correct relative sizes of diferent star mois to be Achieved and this method i essential for type I pat terms with similar stare, but it not appropiate for all woelve variations dealt with above, and there are ven certain rote patterns in which ean he used "Among other shombs in the (3%2) series which become incorporated inthe more common patterns, we ‘may mention (2>2)12,/1, whieh occur in a variety fof verdons from India to the Maghrib a» docs (9 209,121 Type T patterns using thee two rhombs cer fiom Inia to Egypt, but seem tobe abet from the Maghrib. (3% 2)8,16/1 ix commonly incorporsied in Maghbi pater bue is absent oF rare elewhete (9 *2)6,167, onthe other hand, sems to appear occa sionally” in Iran, but is entcely absent elewhere (2)128 ypes VIF and VII are common in Central Asia, but do ot appear in other ar, It, however dilfealt to comple an inventory of pattern typee throughout Islam from published soutees, since Ht is ‘extremely rare to find a work which illustrates every pattern occurring on 4 single monunient, lt lone & Shots gropaphien aren or historical period. Of recent Publications “notable exceptions are. the work of Suronach and Young reerred to above, and that of he Eremanns on thieenticeentury Anatolian carnean ‘Variations such asthe 1ypes EX we have been ca sidering do not of coutse constitute ll posible sources foe variety in patterns which use (2) shombuses. An, ‘extremely common derivative typeI pattern is shown in fig 3. Here the stars are separate! by # “"iwinned pentagon” shape, forming linear sequence which also cccurs along the shor axis of the original ype I hombus (5g. 13). One could continie this proces, {aking linear sequences along the axce ofthe showbus ofa sie pattern type ad using the sequence obtained fs the edge ofa new, larger rhombus ofthe sae shape (or ofa diferent shape: there are two ind of rhombs ‘ossible with 0-ollstar-eenters, which may be distin: {ished as (32) andl (41) smb, using the present Dotaton). However, although thi ix very Suit method for generating derivative pattems, and may indeed have Been used om accasion by Muslin arti, there are many similar patterns whieh cannoe be Adesived inthis way. ‘A diferent source of varity makes use of parallel links (fg. 34) Only one example i illustrated, but it it posible +0 canily such patterns in many different ators; there it no end to this kindof variation, Of ven greaer intrest, i posible to imitate many’ of these derivative deagonal patterns in ater rhombs of the (3 % 2) series, especially in the ease of (352)12.8 and (329,12, sce in thee there is very litle distor the decagonal patterns. A number of such derivatives fevst as authentic paterns throughout Isla the limited space avilable i as been posible to sive no more than a glimpse of some ofthe snethods by Imean of which these star patterns can be investigated ‘There ae of course many numerical procedures which can be used 0 investigate diflerent combinatorial fspects of starpatern construction beyond. those ‘riely mentioned above. We have deat with only ‘hombie configurations of sarcentersand even hee tur rey par ake saying ed me ee rama ao one ple mats 18" wth only one series, the (3% 2) seien—and yt we have ‘ot descibed ways in which thombs of one or more Kinds ca be arranged to build up repeating patterns Tie would entail a dassifeation of rhombic tesla tions “along lines somewhat illerent from thowe employed in clasifying tenellations of regular polygon “There are also many possibiliies for eepeating pat- terns using almost regular str-centers, relatively Tew of Which exis a authentic Ilamie omamest, and these ‘so lend themesves to numerical methods of investiga tion, Although by no means all authentic pateros are ton thombie teselistions this does form ery ul approach sine accounts for most ofthe cont ‘moner patterns, incading hexagons and sqeareased larangements ae special eaten, Tis not suggested that the firs designers of Ilamie patterns used sear Lines fof reasoning to thowe oullined abuve, but theoretical studies ofthis kindl are of particular value in that one ‘can attempt a systematic and exhaustive enumeration ‘of al posible ways of combining authentic star motif 1nd their vaviation, Since noone can posiby study the rofusion of existing Ini pater a fst hand jer to be able to clasiy all authentic varieties, theoretical approsch is clearly an advantage. It then becomes of interes to compare the results of such staies with the reds setally achieved by the original ‘Moslim artes, since the comparison may provide ues to the eary craftsman’ understanding of the deeper geometry of his patterns. 195, KALC Crome Bigpond Cf is ny, at heen Ur 1 ‘nm oie (Lend, 893 Ko, ame it htt Arn tn 0) Dfmes “The Hees of Deen Sula, ated Li Gi chs fam oe Win sr la! (A Pao Wve nd ME Eee Mornay La Dawn pds mune (Me "Wa, Areata wpb man lecst Lata i Gai 473, ina ce en we Can SSM 07 #7 Se the ron Oy idly dn alas #502050 Peto ns Hisoy 3) (Ono 879, 9p. 281-28 ‘tes Grabr (The dn Loon, 1978), rein om 195.98) on wap eco hos tee" Ova, Ho Gay Se ENA" Lstwod and RH. Mac, Gane Senay (Gangs 10 ‘shacrcton ehh cmos (2) an () in toe ‘egave tunate oer 2 erene ses) ‘ieee by Bear mths i te ay inte ‘itp ome he wien ae mae tee he ts ae ng Sey tater et woo Bur (New Yan 197) Nosocntn ye obama icp “Toe da af ng or sibel ere ve ee ee Seton ings However ser es et wh ‘Sea ows vers iia ae fone ei pr ‘Syaegancdie a an adente Srme ams pacers eet re ee Re eee Epes ete el tl te ee ener ceo tenes Se oa ere Die een oem Fe eer ree Lee neat ee Ses eer SES Tyr ioe nomen en eer rece eecupoes mo | cess Cet emmamne Reeceee areca oe chee ooiaam it Fete eee, Pe ee Be en Be cnn” secs tel iasmhaaynonsacserciomes sivbetemma Seat lov as oes eer aains igre mie ‘Donme ofthe Maj fami of Iban,” in Si i Tl At Suen ae eat SOLA aes ge Ror atermae tt et ic ware eae nsarae ie ces tat Sree Petes, oo eee a a ie tgencomete nigeccny Ere crate cderetee Se rear, anes ee a oo ere at icine enn Beare, Hla ot Pt yp 16, ad spar ane Ems ca ote Teeetaaenerincriaayncaas bootmaereniiadir Patti Emil eee Speen cree ew feieiacinoen ava sobbtteemancnies etouc isan Cee EI oe ne 2 st ter a ts apt craton Nan Tt ne au Snir meahch os eter, SEESr cleat S'S at rio as adam 8h $3 Sit Seah ste staan sea warreens 7 as mop op, Back rin ha pa vale oe hich ‘ese of many namic pero abieary te Se atin were ew op sae Sy mk be ut dt {Sine we wring Sh igh el sm ‘pny saan each perm, ae poprions ‘Whe yt tint pews ve aed werd ig tp of ep pe my emt akon he Heo dreamer pombe ‘Sth trai al ie, Se thee rey gees uvealy aye eomsacion tan vn 7 Be Oe

Das könnte Ihnen auch gefallen