There are numerous lmages Of Varäha still in S1tu on temples in western and central lndla dating 伝 om the end of the tenth and first half of the eleventh century W1th virtually the same lconographlc charactenstlcs as this image. ln terms 0f regonal style, however, certaln features Of the anatomy indicate an ongln ln northern Rajasthan, ancient 、 Märude'a. For example, female torsos dlsplay a tubular nb area divided by a ndge at the waist to begln a gently bulglng but narrow belly and hips. The long male torso with defined diaphragm is also indicative of contemporaneous monuments in the regon. lt has been the common wisdom that reliefs W1th pierced backs are typical 0f central and not western lndia; however, they dO appear ln northern Rajasthan. AIthough half a century earlier than our image, the figures of the Nilakantha Mahädeva temple at Kekmd 0 asnagar) in the Nagaur distnct of Rajasthan, nudway between Jodhpur and Jaipur, bear some relauonshlp to th1S piece, even though the dehcacy of the limbs of figures at Kekmd link them to the Cähamäna idlom of the neighbonng Ja1pur and Sikar districts. The late tenth—century subshrines at the site Of Osian, north ofJodhpur, may be more comparable. Dunng the "medieval" penod, Varäha became one Of the most C0n11 Ⅱ on lcomc forms ofVisnu, found frequently on extenor 記 ra mches ofVm»ava temples. However, this image ofVaräha is carved 0f schist rather than the sandstone typical of temple bodies in northern RaJ asthan. This indicates that it was most likely carved for a sanctum, probably the sanctum Of a subsidlary shnne ln a Va1snava temple complex. Dark schist or serpenune/metasiltstone 、 Often utilized in western lndia for sanctum images. The finer gram allowed for a level of detading not possible with coarser sandstone (quartz or quartzlte aremte). R れ化 : Lefebre d'Argence and Tse Bartholomew 1969 : 70 , no. 30. 265

molding and the flattened, leaflike pattern—are common throughout Gujarat, Rajasthan, and the western half of Madhya Pradesh 伝 om the second half of the tenth century onward, reglonally and temporally dlstinguishable only by specifics 0f carving ・ The curved foliate scroll, however, appears most commonly in ancient at this penod at sites such as Gyaraspur and may indicate a regonal afflliation. The carving of the figures on th1S 24 川 is quite crude and choppy. Rather than indicating the quahty of carving of the remainder Of the temple 伝 01 れ which it came, however, the fairly large size and complexlty of the fragment make it likely that this crudeness indicates a hierarchy Of carvmg wherein the sculptor lavished care on the sanctum image itself but sklmped on the mmor figures of the frame. lndeed, the frame, like many smaller or more distant segments Of the temple's carved surface, was 254 god within. visually as well as symbolically dwarfed by the R ′化 : The Asia Society 1967 ー 68 : 102. 1. D ouglas Barrett in a letter to Richard Fullar, June 17 , 1966 ; Seattle Art Museum files. 2. Meister, "Phärhsanä in Western lndla,' pp. 167 ー 88.

the space flanklng the central window projections m1m1CS the intermediary offset of the sanctum (see Fig. 82 ). Poses and attributes of ロ 2 ′示 and レア show great vanety. However, the vanety IS probably a function of the multiplication of such images and seems largely aesthetically motlvated, although at times types bear literary significance. The placement of type withm any designated location does not appear to be canonically dictated, although specific cases exlst in which figures exhibit a clear relationship to the niches they flank by either pose or attribute. Thus the utilization and placement, including relative placement, of m or deities, 市た 2 , ロ′示 , and レア on はⅡ S seems tO have resulted 伝 0n1 a search for a hierarchy of significance to reflect the symbolic hierarchy 0 「 the floor plan. The differentiatlon made salient by framlng and scale allows the wall's ornamental rhythm to visually present this hierarchy. Axiality in relation to the sanctum and door, compass direction, sequential arrangement, clrcumambulatory order, hierarchy within the wall, and visual coherence are all at play on the "medieval' temple exterlor. The foregoing reduces the iconographic program of the wall frieze to its essentlals. After the ninth century the wall was seldom so simple, however, and vaned greatly 伝 om reglon to reglon. ln many parts of central lndia, for example, the frieze came to be divided by honzontal moldings into two or three repeatlng levels of images (see Fig. 83 ). Contemporaneous western lndia retained a single level of images but filled the larger walls with small figure groups or side offsets articulated as P111ars 、 vith elaborate raised niches supenmposed on them (see Fig. 4 ). Figure sculpture on the temple exterlor is by no means limited tO the wall frieze (see Fig. 38 ). lmages were placed in a vanety of locations on the superstructure Of bOth sanctums and halls, in niches on the binding moldings ( レルの , and in the moldmgs of the basement の (Fig. 84 ) and platform ( / ロの and covered the hall extenors as well (Fig. 82 ). The types of images placed ln each Of these locatlons were never arbitranly chosen, however. Their selection resulted 伝 0n1 reasoning that, if not always evident, can often be discerned by exammatlon 0f the histoncal lineage of each type. ln addition, there were Other forms Of contemporaneous stone archltecture that could include a retlnue Of images. These include the freestanding arch 00m 卩の and pillar, the monastery, and the sometimes elaborate step wells popular in western lndia (see Fig. 85 ). ScuIptures 伝 om these types of structures, divested of their onglnal context, are frequently iconographically and compositionally indis- tinguishable 伝 0n1 images that were once part 0 「 standing temples. The often multiple and multipurpose halls that came to front the temple were less stnctly related tO the cosmogonic function of the temple than were the sanctum walls. Thus architects were freer to be innovatlve with plan and decoratlon as lt suited use and patron. ln general, the interlors of these halls reutilize forms for bOth images and ornament found on the temple extenor. However, thiS space, sometlmes with an overarching 10tus ceiling (see Fig. 57 ) , multiple pillars ornamented with bursting and winding vegetatlon, and repeated depictions Of dancers and muslclans has a function beyond the representatlon 0 「 the germlnative nature Of divinity. lt is also the forecourt Of the palace Of heaven (the inner sanctum) in which the divine regent is celebrated in stone and li 応 . The door, while containing features and configuratlons charactenstic of specific reglons at particular points in tlme, also bears certaln lmmutable propertles. These include the rlver goddesses on the lower 」 ambs, flankmg the entry, along With Other aquat1C and protectlve motifs predominatlng in the 10 ℃ r portlon. Celebratory and apotropalc lmages were located in the side 」 ambs, and celestial lmages were featured in the lintel and overdoor. Vegetation entwined the whOle. Like the wall images, the elements of the configuration of doorways became Fig. 85 well established early in the tradition—in fact substantially earlier than was the Queen's step well (RäVi ki Väv) , case with wall imagery. Doors of the late Gupta penod make this evident (see Fig. Patan, Gu. 」 arat, about A. D. 1060. 135

someume in the first half of the eleventh century. This is confirmed by 100k1ng at an lmage 0f Gane'a dancing 伝 om Paranagar, thlrty-eight miles southwest ofAlwar, dated on its base inscnpuon tO A. D. 1044. Although clearly not by the same hand, correspondences ln anatomy, detail, and ornament are close enough tO speculate that the seated Gane'a onglnates in a nearby area and dates only slightly earher. An even closer stylistic accord is with some Of the Other lmages 伝 0n1 the numerous ruined temples at Paranagar, such as some on the Nilakantha temple, which most likely dates to the eleventh century. R ( 肥 The AS1a Society 1991 : 106 , fig. 7 ; Museum of Fine Arts, Boston 1991 : 206 ) 7 , no. 198. 1. These two figures cannot be positlvely identlfied as Riddhi and Siddhi. Other consorts of Gane'a include B uddhl ("knowledge , " percepuon") and 、ⅱ (Lak}mi), Goddess ofWealth. The attributes of these consorts are complementary charactenst1CS Of the God. 2. For sculptures 伝 om Paranagar, Alwar D istnct, Rajasthan, see PupulJayakar, "Medieval Sculpture: 11. Paranagar (Alwar)," M , V01. 12 , no. 2 (March 1959 ) , pp. 61 ー 73. This inscribed image of Gane'a is in the Government Archeolog1cal and Art Museum, Alwar, no. 2 / 3 and is dated Vikrama-Sarhvat 1101. The AIIS archive label states that the image is sald to be 伝 om RaJ0rgarh in the Alwar Distnct. Jayakar, however, states that it comes 伝 om Paranagar, which seems likely. (AIIS Neg. NO. 157.61 ) 3. Accordmg to Jayakar, op. cit. , p. 62 , "An inscnptlon beanng the date 0fSa1iwat 1010 い . D. 956 い s sald t0 have been found ascnbing the [ N ak 叩山司 temple to one AJayapal Raja of the Bargu. 」 ars. Another source ascnbes it tO Lach, alSO one Of the BargtlJar kmgs. ' On stylistic grounds, however, the temple can date no earlier than ca. A. D. 1100. 170

adequately published, and the clear chronologlcal similanty 0f this piece t0 images 伝 om R0da makes this early attnbuuon both feasible and insightful. A second piece (Dayton ArtInstitute, No. 66.25 ) , the nver goddess Gafigä ( 丘 om a doorway 0 「 aboutthe same date) was likewise attnbuted tO ROda. ln publishing the two prces, Bruce H. Evans was hard put tO account for their clear stylistic differences. ["Two EarIy MedievaIIndian ScuIptures," D 畔な川 月れんⅵ tu Bu 〃 etin, V01. 25 , no. 1 (September 1966 ) , pp. 1 ー 6 」 3. Allahabad Museum No. 283 , AIIS Neg. No. 85.89. ln general, however, the Lachhag1r lmages show much lower reliefthan this Umä—Mahe'vara or figures of this penod 伝 om Kannat1J. 25 180

Fig. 64 Detail of the exterlor wall, temple, Modhera, Gu. 」 arat, A. D. 川 27 ー 75. 120 Sürya about $ にツ・ゞ をを : 1 ス ~ groupings likely, but not necessanly, based on common craft groups. 伝ん , will designate more site-specific (smaller and more closely related) elements out 0 「 the んⅡ vocabulary available, regardless 0 「 patronage. 〃川 , like commonalities Of formal charactenstics bOth in the treatment and choice 0 「 This work will 応Ⅱ ow Dhaky's looser usage, with 立 ) , ん designating broad they may interact. style as independent rather than dependent vanables, however much they remam essentially cumulative and self-defining. I see idiom and and though idioms may absorb the general charactenstics 0f a style, WhiCh affects and influences the craftsmen responsible 「 br idioms; 0 「 10Ca1 idioms, lt can aISO denote a generatlve force ln art, one

59. Guardian Lower IeftJamb frieze ofa doorway Provenance unknown, probably K0ta reglon, Rajasthan (Uparamäla) Abou い econd halfofthe 9th century Sandstone; 92.7 x 54.6 cm Los Angeles County Museum ofArt, From the Nasli and Alice Heeramaneck Collection, Museum Associates Purchase; M. 74.5.5 Th1S fragment of a lower Jamb frieze ( の伝 om the le 丘 of a door bears a large Saiva door guardian ( 叩の . Minucking Siva in h1S horrific 応 rm as Bhairava, this ternfY1ng four—armed guardlan cames a tndent and snake ( れ畩の in h1S rear hands and a skull cup ( た叩の in h1S extant front hand. His bearded face dlsplays Siva's thlrd eye, a cavernous mouth, and bared fangs and is crowned by the ascetlc's matted locks 四川材た町 4. Next to the 叩 4 stands a female attendant facing lnward. H er breasts are hlgh and round W1th gentle rolls of flesh below. At her feet, a dwarfish female, apparently holdmg a cosmetlc bag, gazes upward toward the (now nussing) nver goddess who would once have completed this po 血 on of the compositlon (Nos. 52 ー 54 ). The river goddess's 応 ot remains, along W1th the foliate tail Of her sea creature below her attendants' feet. The scales of both this fragment and the previously dlscussed segment Of overdoor ( No. 58 ) imply that the d00 伝 om which they came were quite monumental—probably between thlrteen and fourteen feet hlgh 伝 om the floor tO the top Of the overdoor. AIthough the scale and carving on the two P1eces are identlcal, however, the lconography raises questlons as tO whether they come 伝 0n1 a single doorway or 伝 0n1 doorways on different temples created by the same artlsans. At 6 t glance, the repetltlon ofthe Smva theme 、 seem tO reinforce the connec— t10ns between the t 、 MO P1eces, but, as discussed ( No. 58 ) , an examnatlon ofnumerous standmg monuments points toward the conclusion that the overdoor segment came 伝 om a temple to the G0d Visnu. ln terms of reglonal and chronologlcal ongm, however, the t 、 VO fragrnents undoubtedly coincide, onglnatlng in the Kota repon of Rajasthan (ancient 238

KhaJuräho had attalned a tenuous balance bet 、 veen the reqtllrements Of the autonomous lmage and those Of the entlre C01 れ posltlon. Even on the temple 伝 om which this image came, it is hkely that figures oflesser importance already tipped that balance t0 the needs of the whole ( No. 46 ) , as was true of all sculptures on temples bullt by the subsequent generatlon. 27. Vi ロ Probably extenor wall, central offset Provenance unknown, probably central-eastern Madhya Pradesh (Dähalade'a) About early 11 th century Sandstone; H. 119.3 cm Museum ofFine Arts, Boston, David PuIsifer KimbalI Fund; 25.438 1. This is not certaln, however, since the large temples of Khajuräho frequently dlsplay such d1Vine images, even couples, in Other OffsetS as well. For example, the unusual Pär'vanätha temple repeats images Of standlng Lakyni-Näräyapa without mches flankmg the wlndows on the extenor Of the ambulatory path. 2. Deva, 川 2 研 K ん可 0 , V01. 1 , pp. 23 , 240.

38. Warrior Rondel CentraI sectlon of flat celling, hall, or porch Provenance unknown, probably southern Rajasthan (Medapäta or Gurjarade'a-Arbuda) About late 9th century Sandstone; 65 x 78 cm Ashmolean Museum, Oxford; 1985.5 Eight flying male figures, each brandishmg a sword in h1S nght hand, lnterlock in a circular formatlon renumscent of dance. Each nght 応 ot emerges behind the subsequent figure while the le feet 応 rm an lnner nng. Each man grasps the sword arm Of the one in front, and they lean forward in a rhythnuc mterchange that makes the rondel appear t0 spin 0f its own accord. Filhng the space between the figures' heads are multl- 204 layered flames that seem tO burst 伝 0n1 the sword points. Such rondels, depicting wamors, female dancers, or multiple deitles ln 川 4 〃 d ロ信 format are not uncommon ln western lndla. Judgng 伝 om extant monuments, they tend tO appear in the square center sectlon Of a flat celhng. They are held in place visually by surroundmg 0 hogonal pseudobeams. Structurally, however, they tend to have

Gwalior by Mihira Bh0Ja. ln this positlon, Alla must have been a respected member Of the pratihära nobility, for Gwalior fortress was an lmportant locatlon for the rulers—it guarded the temtory between Kälinjar and Chittaur and was integral tO their campalgns agamst the Rästrakütas tO the south. After assuming his post at Gwalior, AIla built the CaturbhuJ temple of Visnu for the increase 0 「 his and his wife's ment, 当 receptacle of his fame, cut by the chisel and marked with his name. '' The temple is described as a single piece of rock ( 戓 . 励の and a great ship for crosslng the ocean Of existences. " The temple was known as VäiIIabhattasväm1n in honor of AIIa's father. AIIa also built a second temple dedicated tO the goddess that has not survived. Endowments were lavished on both buildings by vanous sections 0 「 the community and the city council made a grant on behalf of the entlre town. The Caturbhu. 」 temple inscnptions also provide an unusually thorough eulo 0 「 the Pratihära dynasty. More typical is an mscnptlon made durmg Ya<ovarman s reign descnbing how his officer Bäläditya built a temple of the Buddha at NäIanda and how it was endowed by Malada, the son of Ya'ovarman s 33 mlnlster. Virtually no informatlon IS glven, however, about the ruling monarch. 34 The same sltuatlon is found in the copper plates 伝 om Una. These were issued by Bälavarman and his son Avanivarman Yoga 0 「 the Cälukya family; they record the 仕 of villages to a temple of the sun god called Taruvädityadeva. They mention 0 that the Pratihära monarch Mahendrapäla I (ca. A. D. 885 ー 910 ) conferred the title of the "five great sounds" 明朝川 ) on Bälavarman and that the glfts were sanctioned by a frontier-guardian ( 4 厩叩の named Dhiika. This Dhiika appears t0 have been a representative Of Mahendrapäla's court. Despite these lmpenal connectlons, a royal genealogy is not glven and all the details focus on matters Of lmmediate local importance. The same sltuation is found in the plates 伝 0n1 35 HaddaIa. Citing the example of the nobility, we can conclude that officers enjoyed considerable autonomy, g1V1ng only bnef acknowledgement 0f their sovereign When issumg such documents regarding their temple patronage. The wives of important Off1Cers were alSO active patrons. An image Of Siva and Pärvati 仕 0n1 Gwalior cames an lnscnption stating that it was commlssioned 36 by RIJükä, the wife 0 「 Sri Rudra, a Pratihära feudatory. Another inscnption, of unkno 、 provenance but no , in Udaipur, records the actlvitles Of Ya'omati, who built a temple of Visnu. She was the wife of Varähasiri1ha, a commander in 37 the service of the Guhila pnnce AparäJita. Though separated by two centunes, both the Caturbhu. 」 and Udaipur lnscnptions descnbe temples as a means of crossmg over this world—a reminder Of hOW little inscnptions and their contents changed with the passage of time. P 酊 ro ″ 6 0 〃石 The foregomg examples are fairly simple in that they represent donors constructing indlvidual temples or making individual grants. Relig10us centers Of importance, however, Often attracted an extended Of temples and endowments. This led tO complicated inscnptions recording numerous glfts by a range Of individuals over a span Of time. Such collective are 佇 0n1 38 Kämän, Ahar, partäbgarh and Siron. The Siron mscnption lists a number 0 「 grants by persons without title—an appropnate brldge tO the most common level 39 0 「 patronage. Temple 応 at this level often consisted 0f plots 0f land, the rents 佇 0n1 which were intended tO beneflt a particular god. However, land was not the only source Of temple revenue. This is shown by an inscnptlon at Delhi dating tO Mihira Bhoja's time, which records the 仕 ofrent money 伝 om a house for lamps, 40 sandal paste, flowers, and worship at a particular shrme. Similar, but somewhat unusual, is a monthly gl 往 of wine for the worship 0 「 Visvu (probably in the Täntnc fashion). A number 0f records show that some kind 0f taxing was used tO support temples. An inscnption 丘 0n1 Pehowa (ancient Prthüdaka) recounts hO 嶬 , a group Of horse traders imposed certam taxes upon themselves and upon their customers and the は y ln which the proceeds were tO be distnbuted tO 42 certam temples in fixed proportlons. Fig. 18 Memor1al stele, Gupte'var temple, Mohangarh, Madhya Pradesh, early l()th century. 55