PIA BRANCACCIO

the Buddhist Caves at

Aurangabad: The Impact of the Laity

T HE PRESENT STUDY of the caves at chitecturaland artisticaffinities with Caves 1, 2, and Aurangabadowes much to the enlighten- 26 at Ajanta.6The so-called Mahayanaphase of pa- ing work of ProfessorWalter Spink, who tronage at Aurangabadwas inauguratedwith the has unfolded the history of Ajantaand relatedcen- richly decorated Cave 3, which occupies the most ters.' It is fromwhat he calls "the period of disrup- privilegedposition next to the oldercaitya. This new tion" at the Ajantacaves that I intend to begin my cave is a small, perfectlydesigned unit (figs. 3 and surveyof the Buddhistcomplex at ,ad- 4), completewith sculptureand paintings-the lat- dressingthe developmentof this poorly understood ter survivingonly as a few traceson the ceiling.7The site' in termsof patronage,audience, and function. existence of a wealthypatron with greatvision can Through an analysisof the distributionof space and be certainlydetected behind the careftillyorganized imageryfrom the end of the fifth centuryonward, I spaceof this structure.In fact,from an accurateplani- hope to shed light on the site's specificrole as a sanc- metricanalysis of Cave3 it has been possible to indi- tuaryrooted in lay devotionalpractices rather than viduate the architecturalmodule on which the spa- in the exclusivemonastic tradition. tial distributionof the cave was probablybased-a Located on a slope of the Sihyachalrange be- unit equalto the radiusof the columns(40 cm).8The tween the Kaum River and the Devgiri basin, the imageryand the profusionof ornamentationwithin Buddhistcaves at Aurangabadare divided into two AurangabadCave 3 seem to confirmthe existenceof main groups (figs.1 and 2), with a third unfinished a planneddesign thatwas fullyrealized thanks to the cluster of later structuresto the north.3The oldest unbrokensupport of the sponsors. While the outer structureat the site, partof the westerngroup, is the faSadeis ruined, the inner squarearea is intact, de- severelydamaged caitya Cave 4, which dates to the fined by twelve lavishly decorated columns and beginningof the CommonEra. Surprisingly, no early flankedon eitherside by two cellaeand a rectangular attachedto this structurecan be identified, chapel. Opposite the entrancedoor, on a sculpted unless it was built in perishablematerials at the foot friezeabove the columns,is a depictionof theprincely of the hill or, as seems likely,carved to the east of the Sutasomajataka. A porch leads into the shrine, caiya alongpart of the escarpmentthat has now col- where a pralambapildasanaBuddha is flankedby lapsed. Another significantfeature of the religious two bodhisattvasin the Ajanta"style," preceded by complex at Aurangabadis the absenceof dedicatory two unique rows of life-sizesculpted kneelingdevo- or commemorativeinscriptions, in contrastto other tees (figs. 5 and 6). The strong affinitiesin design, cave complexesin the region.4 imagery,and sculpturaldetails between Aurangabad Afterthis firstphase of excavation,a revivalof Cave3 and some of the latestcaves at Ajanta(26 and patronage occurred in conjunction with the later 2) indicatethat a few of the same hands might have Vakataka'sactivity at Ajanta, as WalterSpink has bril- workedat both sites.Nevertheless, Aurangabad Cave liantlyshown.' In fact,units 3 and 4a and the unfin- 3 displaysa more "baroque"visual language (fig. 7), ished excavation1 at Aurangabaddisplay strong ar- possibly a conscious manipulation of the Gupta-

Ars Orientalis, supplement I (2000) PIA BRANCACCIO

FIG. 1. Plan of the westerngroup of caves at Aurangabad. By Dr. GiuseppeMonzo.

FIG. 4. Crosssection ofAurangabad Cave 3. By Dr. GiuseppeMonzo.

zi04n r\ll O 4m

FIG. 3. Plan ofAurangabad Cave 3. By Dr. GiuseppeMonzo.

FIG. 5. 42 Aurangabad Cave 3, shrine, devotees,eastern wall. After C. Berkson, The Caves at Aurangabad (New York, 1986). THE BUDDHIST CAVES AT AURANGABAD

FIG. 2. Plan of the easterngroup of caves at Aurangabad. By Dr. GiuseppeMonzo.

Vakatakaartistic idiom, which can be interpretedas the most logical choice for this site-overlooking a a statementof powerby the new Aurangabadpatrons vast plain crossed by traderoutes, easily accessible, wantingto outdo the imperialproductions at Ajanta. and alreadyestablished in the local religious tradi- Thus, Cave 3 at Aurangabadappears to be deeply tion with its earlierBuddhist caitya. connectedwith the localpolitical landscape, perhaps sponsoredby the samefeudatories that took over this territoryafter the collapse of the Vakatakaempire.' It seems that the new local kings excavatedAurang- abadfor political reasons, in orderto createcontinu- ity with practicesof patronageinitiated by their il- lustrious predecessor. To do so they chose a new centerwith no imperialconnections where they could glorifythemselves and the Buddha.Aurangabad was

rt#>, '''"V. .;

.h~~~~~~~~~~~V'

lX:' : " p .'..

FIG. 6. FIGe7. AurangabadCave 3, shrine,devotees, western wall. AurangabadCave 3, interior,detail of column. 43 AfterBerkson, Caves at Aurangabad. CourtesyAmerican Institute of Indian Studies. PIA BRANCACCIO

1-N ~ ~ ~ ~ V - *~~~~~~~~~~~

1'~~~~~~~~~~~~~~~~~~~~~~~~~~1

FIG. 8. FIG. 9. AurangabadCave 4a. AjantaCave 26, interior. AfterBerkson, Caves at Aurangabad. CourtesyAmerican Institute of Indian Studies.

Unfortunately,no donativeinscriptions survive ing recognition would opt for such visually powerful fromCave 3, and one wonderswhether they everex- solutions rather than poorly visible inscriptions to isted. In fact, the patrons of Cave 3 at Aurangabad obtain both legitimation of their power and merit for did leave a differentkind of long-lastingsignature to themselves. their dana (gift) in the form of the life-sizekneeling The fact that these individualized figures of lay devotees sculpted in the sanctum. Such unique fig- devotees were represented inside the sanctum, the ures, almosta tutto tondo,are located along the two holiest of places, raises questions about the function sides of the sanctum convergingtoward the monu- of the and about the possibility of mentalBuddha image. They arecertainly not generic ritual access to the main image by sravakas, who were devotees, since they appearto be individuallychar- not members of the samgha. " In this case the depic- acterized members of a royal group, and it seems tion of worshippers of royal rank could also be taken likely that the male and femalefigures at the head of as a reference to the direct connection between the each row are the actualpatrons of the cave.1?Their spiritual cakravartin, the Buddha, and the temporal impressiveand portraitlikeappearance mightjustify cakravartin, the king. Nevertheless, I believe that the the absence of inscriptions,as the princelypatrons occurrence in such a privileged position of images would have left behind a powerfultrace of their do- representing not gods or monks but lay devotees can nation in these figures eternallyin devotion, as pe- be interpreted as an indication of the prominent secu- rennialreceptacle of merit. In a circumstanceof po- lar nature of the complex at Aurangabad. In contrast 44 liticalinstability such as the collapseof the Vakataka's to the monastic emphasis at Ajanta, Aurangabad empire, it is conceivablethat local feudatoriesseek- seems to have been more open to laity, emerging as a THE BUDDHIST CAVESAT AURANGABAD

religioussanctuary serving primarily the nonordained Cave2'4 (see fig. 1) shows a distinctiveplan con- membersof the community. sistingof a simple shrinechamber enclosed within a The lack of residentialstructures for monks, es- corridorfor circumambulation, which in turnwas ap- peciallyin the westerngroup of caves, confirmsthis proachedthrough a small,now collapsed,mandapa. hypothesis(see fig.1). Even Cave3, which was visu- This structure,squeezed into the last availablerock ally and conceptuallymodeled afterAjanta Cave 2, withinthe westerngroup of caves,has a centralsanc- seems to have clearlyabandoned the vihara pattern tum containingthe usual seated Buddhain dharma- so recurrentat the imperialVakataka complex. Only cakramudraflanked by two bodhisattvas.The en- four residential-likecells showing scarce traces of trance to the cella is guarded by Maitreya and use12 were excavatedalong the side walls of Aurang- Avalokite?vara,both attended by nagarajas.The sty- abadCave 3, as the plannerschose to omit additional listic idiom of these figures is far from the late cells and to introducetwo axialrectangular chapels. Vakatakaone at Ajantaor Aurangabadand seems to Immediatelyto the rightof the earliercaitya hall be in linewith the sixth-and seventh-century Kalacuri is an open shrine,Cave 4a-also datingto the latefifth Brahmanicalcave sculpture found at sites such as century-that atteststo the lay activityat Aurangabad Elephanta,Jogesvari, Mandapesvar, Mahur, and because of its prominenceand accessibility(fig. 8). Ellora. Currentlyvery damaged, this deep nichein the basal- What is particularlyinteresting in Cave 2 at tic rock,once framedby foursmall columns, contains Aurangabadis that the corridor surrounding the a majesticimage of a Buddhain dharmacakramudra shrineis literallyfilled with a multitudeof heteroge- seatedon a highlydecorated throne that is flankedby neous panels sculpted on the walls (fig. 10). Most of two bodhisattvas.It is surprisingto findsuch an elabo- them display an established iconographic format, rateimage in an independentoutdoor shrine, barely with the Buddhaseated on a lotus throneand flanked protectedby a ledge of rock, as all othercomparable by two bodhisattvas.In many of these images, the imagesof this type are alwayscarved in the most sa- "triad"pattern intersects with the so-called depic- credcores of the caves.In fact,Cave 4a's image is very tions of the Miracleof ?ravasti,when the Tathagata similarin styleand decoration to the one carvedin the multiplieshimself on lotus flowers. To explore the body of the stupaof AjantaCave 26 (fig.9) or the one source of this imageryis beyond the scope of the in the darksanctum of Cave 3 at Aurangabad.Here present article,but it is sufficientto point out that the focal iconographyof the large caves crosses the such an iconographicpattern occurs invariablyin thresholdof the innershrine to be easilyapproached votivepanels donated by individualsin the Buddhist andviewed in the4a chapelat theentrance of thecom- caves of the Deccan during the so-called plex. This smalland independentstructure contain- phase. At Ajanta,Walter Spink has suggested that ing only a mainimage is an unprecedentedarchitec- these "intrusive"panels were added duringthe dis- turalsolution'3 that illustrates the innovativespirit of ruptionof the site, when the Vakatakaslost control lay patronageand audienceat Aurangabad.It is pos- over the caves and patronage suddenly collapsed sible thata wealthyindividual sponsored this public prior to the abandonmentof the complex.'5 unit in conjunctionwith the princely patronageof Conversely,at AurangabadCave 2 the design of Caves 1 and 3. the cave seems to have been conceived in order to The secularorganization of the Buddhist cave accommodatethese attestationsof individualdevo- complex at Aurangabad,which was intimatelycon- tion, as thepradaksinapatha was left undecoratedto nected with lay patronageand devotion, continued makespace for such imagery.'6A singularfeature that beyond the end of the fifthcentury. The next phase betraysthe popularvotive origin of this body of im- of excavationat the site is characterizedby the diffu- ageryis the frequentdepiction of worshippersat the sion of new cave plans focusingon an increaseduse bottom registerof these panels: they are mostly lay of public space. The innovativedesigns appearto people, often women, and rarely members of the reflectthe ritualneeds of the lay communityand in- samgha. Further,in Cave 2, images of a squatting 45 dicate a differentlymediated approach to the deity. female,commonly identified as Lajj Gauriholding PIA BRANCACCIO

FIG. 10. Augangabad Cave 2, interior, detail of the westernpradaksinapatha wall. CourtesyAmerican Institute of Indian Studies.

a lotus,'7occur in four instancesin associationwith representationsof the Tathagata'sepiphany (fig. 11). The emergenceinto the Buddhistimagery of such a figuretied to the world of local and ancestralbeliefs seems to confirmthe lay devotionalmatrix of thisunit and its sculpted panels. The whole conception of Cave 2 at Aurangabad seems to respond to the devotionalnecessities of the KiiPln)1 sravakas,which is not surprisingat a site thatshowed a stronglay orientationsince the inceptionof the so- called Mahayanaphase of patronage.The plan with a centralsanctum certainly allowed for easieraccess to the deity and the performanceof devotionalprac- tices. Of particularsignificance is the factthat the core of the shrine is projectedforward to meet the needs of a largercommunity. This architecturalformat ap- pears sporadically in many parts of the Buddhist world in conjunctionwith structureslocated outside the exclusivemonastic areas and generallyassociated with lay devotionalpractices. Besides CentralAsia,'8 relevantexamples of such shrines can be found in Sri Lankain associationwith the patimaghara, or "shrinereceptacle of the image of the Buddha."'9It FIG. 11. 46 is interestingto observethat such a structure,identi- Aurangabad Cave 2, interior, detail of a votive panel. fied in Pali also as p&sadaor palace, the residenceof AfterBerkson, Caves at Aurangabad. THE BUDDHIST CAVES AT AURANGABAD

FIG. 12. Plan of Ellora Cave 21.

LIJi

the Buddha, was central only to the suburban and new ways of approaching the deity appear to have pabbata -religious complexes with a more been in practice. distinct devotional purpose. The unfinished Cave 9 represents the next and At Aurangabad, the plan with a central sanctum final stage at Aurangabad-in which the expansion and pradaksinapatha is also found in Cave 5. Fur- of the public space makes the structure even more ther, it occurs with some additions in Caves 6 and 7 physically and emotionally accessible to devotion. of the eastern group,20providing us with a useful pa- The unusual plan of this cave (see fig. 2), with three rameter in support of a chronological framework for sancta opening onto a large, rectangular porch, has the second phase of activity at the site. A comparable been generally attributed to its hasty completion, as layout appears in a Brahmanical context at Ellora patrons tend to finish the main Buddha images to gain Caves 14, 20, and in particular Cave 21, the so-called merit before abandoning a site. Contrary to what we Ramesvaracave (fig. 12). The architecture and sculp- would expect in a similar rushed situation, the Bud- tural evidence seems to indicate that Caves 2, 5, and dha in the central shrine, likely the focal icon on the entire eastern complex at Aurangabad belong to which the most effort would converge, was only the same phase of patronage as the above-mentioned roughed out, while the main images in the two side excavations at Ellora,2' which Walter Spink has at- sancta were fully carved. It seems possible that the tributed to the Kalacuri kings,22who probably con- various shrines had different patrons, who inter- trolled these parts of .23 rupted their work in the cave at different times. Nev- With the excavation of Caves 2 and 5, all the rock ertheless, I believe that the uncommon layout was available around the Hinayana caitya at Aurangabad part of the original plan, and the abrupt interruption was exhausted. Thus, a new cluster of units was ini- of work at the site did not change the basic organiza- tiated to the east with Caves 6 and 7 (see fig. 2), still tion of space. The large rectangular mandapa was in keeping with the devotional orientation of the site. probably part of the original design, as it functioned During the last phase of activity at Aurangabad the to unify and allow more direct access to the images 47 space accessible to public devotion was maximized, in the sancta. In Caves 5 and 6 at Aurangabad we PIA BRANCACCIO

FIG. 13. Plan of Mandapesvar Cave.

_q~~~~~~~~~~~~~~~~~~~~~~16

already notice an increasing emphasis on the at Aurangabad9 had alreadyemerged in the Bud- mandapa as a bridge between the outer world and dhist world. Within the same region it occurs in the the innersacred space of the cave. Cave9 represents fifth-centurycave of Ghatotkacha26and in the unfin- the ultimatedevelopment of this concept, in which ished Cave 21 at Ajanta. the barriersbetween the two spheres are almost re- In conclusion,I hope thatthis briefsurvey of the moved and the mandapabecomes the cave. It marks development of cave planning at Aurangabadhas the climaxof the prominentlay orientationof the site shed light on the functionof this complex as a center that grows in keeping with the devotionalneeds of oriented toward popular devotion and secularpa- the sravakas. The presence of a largeparinirv?nzna tronage. Its so-called Mahayanaphase appears to scene carvedon the westernwall of Cave9, usuallya have been intrinsicallyconnected with the collapse populardevotional icon, supportsthe hypothesisthat of imperialVakataka and their patronageat Ajanta. the structurewas conceived as a place for collective Aurangabadrises in response to this exclusive mo- worship. In fact, such images are generallylocated nastic center, controlledby imperialpatrons, to tes- in more accessibleshrines like the caitya Cave 26 at tify to the triumphof the regionalpowers and local Ajanta. popular Buddhist forces at the end of the fifth cen- The cave layout, with three shrines opening on tury. The accessibilityof the site, the smallnumber a mandapa, is a pattern that occurs elsewhere in of cells for the samgha, and the presence of life-size Kalacuriarchitecture. The Saiva temple at Manda- depictions of lay devotees, probablynoble donors, pesvar in Konkan (fig. 13), attributedto Kalacuri in the sanctumof Cave3 certainlyindicate the grow- patronageand also thoughtto be the resultof a hasty ing importanceof the "secular"at Aurangabad. excavationas it was nevercompleted,24 is surprisingly Its life and prosperitycontinued to be strongly similarto Cave 9. Thus, it is reasonableto suggest rooted in the world of lay devotion and patronage that Mandapesvarmight have had some impact on through time, as illustratedby Cave 2 and related the planningof Cave 9 at Aurangabad.25 structuresdatable to the Kalacuriperiod. The strong While innovative in format, with its ample linkage of the site with popular religiosity is par- 48 mandapaleading directlyinto the most sacredunits ticularlyevident in Cave 2, with its centralsanctum of the structure,the basic three-sanctatype adopted and pradaksinapatha for circumambulation left THE BUDDHIST CAVESAT AURANGABAD

undecorated to display a number of individually Qureshi,Art and VisionofAurangabad Caves (Delhi: Bharatiya commissionedvotive panels. Finally,in the eastern KalaPrakashan, 1998). groupof caves,the unusualand unfinishedstructure 3. We do not know the ancientname of the complex, though it 9, with three shrines on a wide man.apa and the has been suggestedthat the site was known as Rajatalakabased parinirvana, seems to markthe culminationof the on an earlyinscription found at Kanheri.S. Gokhale,"Ajanta: popular tendency at the site, reducingeven further The Centerof MonasticEducation," in TheArt ofAjanta,New the distancebetween the common devotees and the Perspectives,ed. R. Parimoo (New Delhi: Books & Books, 1991), 52. holiest of images.D 4. Donative epigraphic records appear at most of the Bud- dhist cave sites of the Deccan, such as Bhaja, Karli, Pitalkhora, and Ajanta. See J. Burgess and I. Bhagwanlal, Inscriptionsfrom the Cave Templesof WesternIndia (Delhi, 1880) or V.V. Mirashi, Inscriptions of the Vakatakas,Cor- pus Inscriptionum Indicarum 5 (Ootacamund: Government Notes Epigraphist for , 1963). 1. It would be impossible to cite here all the significantcontri- 5. WalterSpink has tracedthe connectionsbetween Aurangabad butions by Walterjito the understandingof Ajantaand other and Ajantasince his 1967 studyAjanta to Ellora. rock-hewnreligious centers in the westernDeccan. I would like to rememberhis Ajanta to Ellora, which appearedas a special 6. The latercaitya Cave 26 at Ajantashows columns and deco- issue of Margin 1967, and some ofhis most recentarticles: "The rativeelements comparable to the ones occurringin Aurangabad Great Cave at Elephanta:A Study of Sources," in Essays on 3. The viharas 1 and 2 at Ajantaalso share a similarspatial or- Gupta Culture, ed. B. Smith and E. Zelliot (Delhi: Motilal ganizationwith unit 3 at Aurangabad. Banarsidass,1983), 235-82; "The Archaeologyof Ajanta,"Ars Orientalis21 (1991): 67-94; "Beforethe Fall:Pride and Piety 7. Remnantsofpainted medallions that survive in the antecham- at Ajanta,"in ThePowers ofArt: Patronagein Indian Culture, ber to the sanctumare almost identical to the one foundat Ajanta ed. B. StolerMiller (New Delhi:Oxford University Press, 1992), Cave 17. 65-77; finallyhis booklet Ajanta, A Brief History and Guide (AnnArbor: Asian Art Archives, University of Michigan,1994). 8. This study has been conducted in collaborationwith the ar- chitectDr. GiuseppeMonzo. See P. Brancaccio,"11 Complesso 2. Surprisingly the caves at Aurangabadhave scarcely been Rupestre di Aurangabad" (Ph.D. dissertation, Istituto investigated by scholars. A first description of the complex UniversitarioOrientale, Naples, 1994), appendix 1. appearedin 1858, in Dr. Bradley'sAccount of Statistics of the Sarkarof , done on behalfof the Nizam'sgovernment. 9. According to Walter Spink, the local kings, the A?makas, More scientific was the approach ofJames Burgess in his Re- controlledthe regionafter the fallof the Vakatakas.See W. Spink, Port of theAntiquities in the Bidar and AurangabadDistricts, "The Vakataka'sFlowering and Fall,"in Art ofAjanta, 71-99. 18 75-76 (London: W. H. Allen, 1878). The more recent and scarce bibliographyon Aurangabad,besides guidebooks, in- 10. On portraitsin ancient India, see V. Dehejia, "The Very cludes: G. Yazdani, "The Rock-Hewn Temples of Aurang- Idea of a Portrait,"Ars Orientalis 28 (1998): 41-47. For the abad:Their Sculptureand Architecture,"Indian Art and Let- visual idioms and meanings behind patron's portraits,see P. ters 1 (1936-37): 1-9; R. S. Gupte, "An InterestingPanel from Kaimal,"Passionate Bodies: Constructionsof the Self in South the AurangabadCaves," Marathwada Universityjournal 3.2 IndianPortraits," Archives ofAsian Art 48 (1995): 6-16. (1963):59-63; D. Brown-Levine,"Aurangabad: A StylisticAnaly- sis," ArtibusAsiae 38 (1966): 175-88; A. Ray, Aurangabad 11. Figures in devotional attitude already appear at Ajanta, Sculptures(Calcutta: Firma K. L. Mukhopadhyay,1996); K. sculpted on the bases of the main Buddhas'thrones. They are N. Dikshit, "A Newly Discovered Buddha Image from small,never acquire a prominentposition in the sancta,and of- AurangabadCaves," in Madhu: RecentResearches in Indian ten seem to makereference to the audiencewitnessing the First Archaeologyand Art History, ed. NagarajaRao (Delhi: Agam Sermonat .The presence of devotees by the feet of the KalaPrakashan, 1981), 235-36; J. Huntington, "Cave Six at Buddhain the shrineseems to become more establishedduring Aurangabad: A Tantrayana Monument," in Kaladars'ana: what Spink recognizesas the later excavationphase at the site AmericanStudies in the Art of India, ed. J.Williams (Leiden: (A.D. 475 onward), culminatingwith nine small figurescarved E.J. Brill, 1981), 47-55; C. Berkson, TheCaves atAurangabad on the pedestal of the Buddha's throne in Cave 1 sanctum.In (New York:Mapin, 1986); R. Brown, "A LaiiaGauri in a Bud- this lightthe impressivelife-size lay devoteesat Aurangabadrep- dhist Context at Aurangabad,"The Journal of the Interna- resent the culminationof a tendency that had alreadyemerged 49 tional Associationof Buddhist Studies 13.2 (1990): 1-16; D. in nuce at Ajanta. PIA BRANCACCIO

12. Only two inner cells show, in the upper part of the door 21. The chronologicalissues related to the later phase of pa- frames,traces of a door hinge, usuallytaken as indicatinguse of tronageat Aurangabadare too complex to be included in the the chambers. present paper. It is probably enough to rememberthat the af- finitiesbetween the Kalacuricaves at Elloraand the laterunits 13. We do not come acrosscomparable isolated chapels at Ajanta at Aurangabad,in particularCave 7, were alreadyput forward with such elaborateshrine-type icons. They arealways attached by W. Spink, "Ellora'sEarliest Phase," Bulletin of the Ameri- to largercaves, such as the side wings of Cave 26. can Academyof Benares 1 (1967): 10. The Brahmanicalrock-cut caves at Mahurand Dhoke, with 14. Cave 2 is located between the earlierunits 1 and 3, at the a centralshrine, also representinteresting comparisons for the same level as Cave 3. units excavatedin the laterphase of activityat Aurangabad.For Mahursee SoundaraRajan, Cave Templesof the Deccan (New 15. See Spink, "Checklistof SculpturedIntrusions at Ajanta," Delhi: ArchaeologicalSurvey of India, 1981), 164-67; for the in Ajanta, A Brief Historyand Guide,36-3 7. cave at Dhoke see G. Tarr, "The Siva Cave Temples at Dhokesvara,"Oriental Art 15.4 (1969): 260-80. I do not en- 16. Donative panels were primarilylocated in visible areas of tirely agree, however, with the author's interpretationof the the structure.It seems likely thatsome of them might have also Kalacuristructure at Dhoke. been painted, as at Ajanta, and simply have not survived at Aurangabad. 22. For the chronologyof thesecaves at Ellora,see Spink,Ajanta to Ellora. 17. RobertBrown, "A LaiiaGauri," identifies this image as re- lated to the Buddhistgoddess Vasudhara. 23. Unfortunately,we do not have a clearpicture of the histori- cal developmentsin the regionafter the collapseofthe Vakatakas. 18. A surveyof the squaretype of plan appearsin H. G. Franz, It seems reasonableto suggest that aftera conflict between the "AmbulatoryTemples in and Hinduism," South Traikutakasand the Visnukund.ins, the Traikutakasbriefly con- AsianArchaeology 1979, ed. H. Hartel(Berlin: Dietrich Reimer trolled the area. Finally at the end of the sixth century the Verlag,1981), 449-58. The layoutwith a centralcella that could Kalacuriseem to have established their supremacy.For a re- be circumambulatedbecame very popular at the cave temples view of the various historical issues, see Brancaccio, "Il of Kyzil, Kumtura, Bazaklik, and Qutcha. A. Griinwedel, Complesso Rupestredi Aurangabad,"21-29. AltbuddhistischeKultstatten in ChinesischTurkestan (Berlin: G. Reimer, 1912); A. von Le Coq, Buried Treasuresof Chinese 24. Spink, "The GreatCave at Elephanta,"243. Turkestan.An Accountof the Activities and Adventuresof the Secondand Third German Turfan Expeditions(London: G. 25. Although it is hard to establish secure chronologicalse- Allen & Unwin, 1928). quences with regardto the structuresin consideration,it seems reasonableto suggest that the caves in Konkanrepresent the 19. R. Prematillekeand R. Silva,"A BuddhistMonastery Type first stage of Kalacuripatronage in Maharashtra. of Ancient Ceylon Showing MahayanistInfluence," Artibus Asiae30.1 (1968): 61-84; S. Bandaranayake,Sinhalese Monas- 26. With threeshrines on the backwall, the centralone with the tic Architecture(Leiden: E.J. Brill, 1974). mainBuddha icon and the side ones probablysubsidiary chapels with paintedimages. See W. Spink,"Ajanta and Ghatotkacha:A 20. In these caves thereare, in additionto the centralshrine and PreliminaryAnalysis," Ars Orientalis5 (1966): 135-56. pradaksinapatha,small cells lining the side walls.

50