HAL Id: halshs-02556480https://halshs.archives-ouvertes.fr/halshs-02556480

Submitted on 28 Apr 2020

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Of Gods and Mortals: Lila CopaCharlotte Schmid

To cite this version:Charlotte Schmid. Of Gods and Mortals: Lila Copa. Catherine Jarrige; Vincent Lefèvre. South Asianarchaeology 2001 : proceedings of the Sixteenth international conference of the European associationof South Asian archaeologists, Vol. II, Éditions Recherche sur les Civilisations, pp.623-636, 2005,9782865383016. �halshs-02556480�

6 2 3

Of Gods and Mortals: Lila CopaCHARLOTTE SCHMID

appeared, in North India, in the 4th century CE, when a new type of literature was emerging: the Puranas. Thetraditional five components of a Purana (creation, re-creation, genealogies, ages of Manu and the history ofdynasties) are all linked in one manner or another to thenarrative genre with which puranic literature seems to bein close correspondence. The epics otherwise comprise anarrative totality abundantly drawn upon by the P u ra na s.The illustration, on sacred buildings, of the stories told inthese texts give access to a much wider mythologicalrange than is provided by the textual corpus alone.

Presentation of the Corpus

The first representations of the brahmanical narrativecycles appeared in North India during the Guptadynasty. They are connected with the two most repre-sented of the avataras of ViÒnu, Rama and KrÒna, to befound again on the Copa temples. These first cycles aremostly fragmentary and, with few exceptions (the tem-ples of Apsa∂h (Bihar), of Deogarh and Nachna-kuthara(Madhya Pradesh), it is no longer possible to study themin the architectural context to which they belong. InSouth India, it would seem that the first extensive brah-manical cycles were Copa, even if they were preceded byrather more sporadic attempts in Calukya art, in Badamiand Pa††a∂akal, in the North of Karnataka. The mostextensive cycles decorated the rich Hoysala temples ofKarnataka from the beginning of the 11th century; theyare influenced by lesser known Copa experiments whichwill be the subject of this paper. The corpus consideredhere is constituted by Copa temples dating from the

anugrahaya bhutanam manuÒam deham asthitaÌ /bhajate tadrikiÌ kri∂a yaÌ krutva tatparo bhavet //“It was for the sake of bestowing grace upon beings that he took on a human body and that he took part in games such as on hear-ing about which one becomes devoted to him.”

Bhagavata-purana X.33, 37

In South India, the bases, or socles, of some Co pa tem-ples are decorated with small sculpted panels placed oneafter the other to form a frieze running all around the tem-ple. These panels portray the lila of the gods: of ViÒnu andhis a v a tar as, of Kiva and his son Skanda and portray, too,the human drama through the dancers and musicians whoparticipate in the divine adventures. These largely unpub-lished friezes, sometimes constituting actual “narrativecycles” interspersed with separate representations, areoften inspired by epic and puranic literature and bear wit-ness to the popularity of that literature. As we shall see,they provide firm evidence of the great popularity of thenarrative genre and allow for the consideration of thatgenre from a fresh angle the better to comprehend theiconographic ensemble and the specificities of templesdecorated with this type of representation. The Centre ofthe École française d’Extrême-Orient in Pondicherry hasinitiated the study of these series of representations.

“Narrative cycle” here refers to the continuousrecounting of a story, whether in a text or on a temple. Inthe latter case, the cycle corresponds to the representationin images of a narrative text, made up of a succession ofactions appearing in those sections of the friezes whichbring onto the scene the characters in the given situationsaccording to narrative chronology. The earliest icono-graphic examples of brahmanical narrative cycles

6 2 4

Of Gods and Mortals: Lila C opa ❘ s c h m i d ❘

beginning of the 10th to the beginning of the 11th c e n t u r y .It comprises, in the region of the Kauveri delta, theKa∂aiyar temple of Tiruchennampun∂i (Fig. 1), theNagekvara of Kumbakonam, the Brahmapurikvara ofP ul la m aœgai, the Naltuñai I kvara of Puñjai, theTirumexxatikvara of Paccil, the Samavedikvara of Tiru-maœgalam, and farther North, near Pondicherry, theVaradarajaperumal of Tirubhuvanai (Fig. 2). To thisbody corresponding to a totality of about six hundredrepresentations, I add the small VaiÒnava temple ofKonar, near the village of Tirumalpur, not far from Kañ-cipuram and dating from the second half of the 10th cen-tury.1 The narrative series in this temple appear on thepilasters2 and not on the plinths but is of the same type ofrepresentation and, moreover, the VaiÒnava character ofthe temple makes it possible to add to observations madeon the comparison between the Kaiva corpus of the Kau-veri delta and the VaiÒnava Varadarajaperumal.

The temples of the Kauveri are in fact all Kaiva. Theyare of modest dimensions, sometimes probably becausethey have been reestablished; they frequently bear a greatnumber of inscriptions. The bas-reliefs which are ofinterest here would seem to have been sculpted in the

space of fifty years, most of which belong to the secondhalf of the 10th century. The dating of these temples is anuncertain area since each one poses specific problemseven though they are linked elements of one small corpus.If, however, one wishes to understand what could havebeen the relationships, iconographic and other, betweenone temple to the other, a brief reminder of the elementsof the dating is necessary. With four Pallava inscriptions,the temple of Tiruchennampun∂i has existed in one wayor another since the Pallava epoch. These inscriptions arefound only on non-fonctional parts of the temple, on pil-lars and on the jambs of the door of the ardhaman∂apa:they may have been used for a second time during thereconstruction of a temple, Copa in architecture andiconography, probably in the reign of Parantaka I (907-955). The earliest incription we can be certain of (SouthIndian Inscriptions, vol. VII, no. 518) is dated from thefourteenth year of the reign of Parantaka I, that is 921.3

The dating of the temple of Kumbakonam, betweenthe second half of the 9th and the beginning of the10th century, has been the cause of a great deal of con-troversy. The epigraphy is not much help. A Pan∂yainscription (Rangaracharya 1919, vol. II, no. 22, p. 1235)

Fig. 1 – Tiruchennampun∂i, Ka∂aiyar temple, 10th century. (Photo: EFEO-Evenisse N 26 30).

6 2 5

engraved on a reused block in the shrine to the Goddessis a little later and may indicate that the foundation of thetemple also predates the Copa period. We have to noticealso that, anyway, the site is attached to the corpus of theTevaram, devotional Kaiva hymns attributed to the 7thcentury: a temple existed here as soon as the 7th century.So here too it is a case of a reestablishment. The earliestCo pa inscription of confirmed date is 934 (Rangaracharya1919, vol. II: no. 34, p. 1237). T.V. Mahalingam (1967:83-84), in attributing the title “Parakekarivarman”, foundin three inscriptions, to Parantaka I, puts the even earlierdate of 910 on the first epigraphic reference. This titlebeing, however, common amongst Copa kings the attri-bution remains unconfirmed. David T. Sanford (1974:182-183) dates the temple from the second half of the9th century. It should be mentioned that Sanford con-siders the narrative series of the Nagekvara to be very lit-tle earlier than those in the Pullamaœgai temple. The dat-ing of the later temple is also, however, a subject ofcontroversy even if dated from the beginning of the10th century, that means even if we do not take intoaccount that, as a temple attached to the corpus of theTevaram, the temple has existed in one form or another

since that epoch. The eight inscriptions on the walls ofthe main temple give rise to discussion.4 One of thesedates from the eleventh year of the reign of Parantaka I,that is 918: J.-C. Harle (1958: 5) dates the temple to thevery beginning of the 10th century while S. R. Balasub-rahmanyam (1971: 6, 223-224) dates it from the openingof the reign of Parantaka I, without being more precise.Sanford (1974: 175; 1987: 280-281) considers anotherinscription which records a donation from KembiyanMahavali-Vanarayar made in the sixth year of the reignof a Copa king Sanford identifies as Parantaka I: men-tioned in this inscription is a title Parantaka I had givenat another time to his Ganga vassal, Prthivipati II. Theinscription would thus date from 913 and the templefrom 910-911, at the moment when the vassal had helpedAditya I to gain a decisive victory over the RaÒ†raku†aKrÒna II. The fact that this temple is connected with thecorpus of the Tevaram never seems to have been consid-ered: once again it is a matter of reconstruction.

The temple of Puñjai certainly seems to be a littlelater, but it offers a very wide margin for dating and theinscriptions are of very little help. The beautiful aestheticof the temple makes it tempting to ascribe an earlier date

Fig. 2 – Tirubhuvanai, Varadarajaperumal temple, 11th century. (Photo: EFEO-Ravindran E 75 7).

6 2 6

Of Gods and Mortals: Lila C opa ❘ s c h m i d ❘

to it than it could possibly be accurate. Once again it is amatter of reconstruction: this is another site connectedwith the Tevar a m which means that the temple must havebeen in existence since the 7th century. For Balasubrah-manyam (1971: 266-267), since the earliest inscription ofthe complex (which is found however on the base of asecondary temple) dates from 968, the Copa structure ofthe temple dates from the short reign of ƒAditya II (964-969). Sanford (1974: 178-181) sees both the architecturalcharacteristics of the temple and the representations ofdonors on its South wall as corresponding with the mid-dle of the 10th century even though D. Barrett (1974: 81-82) dates it around 940. G. Hoekveld-Meier (1981: 182-183, 204), on the basis of the convergence of threearchitectural criteria, dates it from the very end of the10th century or even the beginning of the 11th. Lastly,P. Kaimal (1999: 87-89) sees 964 as a terminus ante quemand the temple as more likely to date from the first half ofthe 10th century. There are very precise iconographicresemblances between this temple and the KumbakonamNagekvara one, however, such as the two representationsof Varaha threatened by a naga king, a version of themyth absent from the texts, and these similarities bringeither its dating back into question or that of the Kum-bakonam temple which, as is clear from all the publica-tions mentioned, is uniformly said to be earlier than thedate of its epigraphic corpus.

We see that the three temples of Puñjai, Pullamaœgaiand Kumbakonam are included in a relative chronologywhich, constructed according to the different architec-tural considerations of each author, does not match theone established with the aid of the inscriptions (which arenot yet all published). It is possible that a detailed studyof the narrative series may be of use in more accuratelyestablishing the chronology of the three temples.

The temple of Tirumaœgalam has three series ofpanels. The third series, found on the bases of thepilasters, offers an iconography which is unusual and inseveral places obviously linked with the deities in theniches. The positioning and the unusual iconographytend to support Sanford (1974: 173-174) who, on thebasis of considerations of style, particularly architectural,puts forward the hypothesis that the series on the basesare earlier by almost a century than the one on the basesof the pilasters: the temple could have been rebuilt on anearlier base. The first two series would be from the begin-

ning of the 10th century, while the third would date fromthe end of the 10th or the beginning of the 11th century.A single inscription from the Copa age has been found onthe temple (Archaeological Report on South Indian Epig-raphy for 1929-30, no. 251). It is dated from the 15th yearof the reign of Rajaraja I, 1000. Sanford’s theory isreasonable, but by no means confirmed and the dating ofthe series could be more accurate than it is. It is obviousthat all the different series were not sculpted in the sameworkshop, yet several teams were able to work at thesame time, rather than successively, on this importantsite.

Only three short reports on the four inscriptionsrecorded from Tirumexxatikvara of Paccil are publishedin the Annual Report for Indian Epigraphy for 1992-1993(1998: 4-5, 67-70). The earliest inscriptions date fromthe third or fourth year (both dates are given in thereport) of the reign of “Parakekarivarman”, a titlebelonging to we know not which king. The characteristicsof the inscriptions correspond to the epigraphy of the10th century. Another inscription dates from 990 and thetwo remaining from the beginning of the 11th century(the reign of Raj a raja I). The iconographic analysis is dif-ficult as many of the statues are very worn away. Thearchitectural style corresponds to the 10th century, with-out it being possible to be precise.

The inscriptions of the temple in Tirubhuvanai arebeing prepared for publication. As far as the temple itselfis concerned, the earliest epigraphic date we can be sureof is 1017 (inscription no. 92 from the corpus of inscrip-tions in the territory of Pondicherry, found on the Eastwall of the ardhaman∂apa). This temple, literally coveredwith inscriptions, is certainly not much earlier than thatfirst known inscription.5 The temple at Tirumalpurseems to be a little earlier. The inscriptions of this templeconverge towards the middle of the 10th century.6 It hasbeen greatly renovated and the positions of the inscrip-tions are not always the original ones (Balasubrah-manyam 1971: 92, 226).

V a iÒ nava Stories

From one temple to another we find the same depic-tions, even sculpted in exactly the same way. Theybelong to friezes made up of a series of sculpted panels

6 2 7

whose format varies between 11 µ 11 centimeters and2 7 µ 13 and they are worthy of the name “miniature pan-els” sometimes given to them. Each temple comprisesone to three series of panels, one above another, on thesocle of the temple and then, sometimes, on the bases ofthe external walls.7 According to convention, we havenumbered them from the temple ground level.

The bas-reliefs of these series are often of very highquality; myths that they represent can be the same asthose that we find in the devakoÒ†has, which are of muchlarger dimensions. The most common Kaiva myths, suchas the representation of the Gajasamhar a murti atT i r u c h e n n a m pu n ∂i and at Pul la m aœgai, or that ofBhikÒa†ana in the above mentioned sites (Fig. 3), as wellas in Tirumaœgalam and Paccil, are once again found onthe friezes at the base. But these representations do notreally belong to the narrative cycles. With few exceptions,Kaiva mythology is represented only in a single panel andwe cannot speak of the continuity of the narration in thiscase: most of the cycles are VaiÒnava.

The friezes to which the narrative cycles belong aremade up, in fact, of VaiÒnava narrative cycles, of inde-pendent Kaiva or VaiÒnava myths and of worldly repre-

sentations, such as the dancers and the musicians whooften appear. The importance of the place accorded toVaiÒnava representations in the Kaiva context is surpris-ing. Since certain independent panels also show VaiÒna v amyths, such as the avataras of Trivikrama, Varaha andN a r a s imha, the majority of these series of representationsare VaiÒnava. In the present corpus, only one temple isVaiÒnava, the Varadarajaperumal of Tirubhuvanai, nearPondicherry. This temple is an exception in a number ofways, which is perhaps explained by its being of a laterperiod than the rest and not of the same region, but alsobecause of its being VaiÒnava temple.8 These representa-tions certainly date back to the beginning of the 11th cen-tury. If we compare it to the Kaiva corpus of the Kauveridelta, it has a very curious iconography: it does not rep-resent any scene that we can associate with the Kaiva con-text. Hence, as we observe in other domains, it seems thatthe Kaiva movement was more tolerant than the VaiÒna v amovement. This temple shows that we can, anyway, nei-ther ignore the opposition between the two movements,nor consider that, in the case of narrative cycles at least,the Kaiva-VaiÒnava opposition was not operative, or wasnever operative.

Fig. 3 – Tiruchennampun∂i, Ka∂aiyar temple, 10th century, Kiva BhikÒa†ana. (Photo: EFEO-Evenisse N 24 9).

6 2 8

Of Gods and Mortals: Lila C opa ❘ s c h m i d ❘

Architectural Traditions and A v a tar a s

One of the first questions to arise in relation to thesenarrative series concerns their positioning. The panelsmay be placed very low down on the bases of the templebut, once again, the Varadarajaperumal of Tirubhuvanaiis an exception with its representations placed a littleabove eye-level. In all the other temples, the series of rep-resentations go upwards, one above another. The firstseries of the temple of Tirumaœgalam, is level with thefloor and some of its panels are today completely buried.At Pullamaœgai and Kumbakonam, the representations,belonging to the first series of bas-reliefs, are placed verylow down in a sort of ditch surrounding the shrine. It iscertain that the ground around the temples has risen inthe course of the centuries and that the panels were muchmore visible in the 10th century than they are today.However, this positioning does not seem likely ever tohave enhanced the visibility of the scenes represented.

We straightaway think of the kind of symbolic valuethese representations may have had, quasi-invisible asthey are and yet often so finely wrought. It is, however,necessary first to think in terms of the architecture and ofthe traditional positioning in the temple. Narrative repre-sentations require a succession of scenes one afteranother and have their place just where the horizontalcharacter of the structure is most marked. There are twoplaces in the temples which are especially qualified fromthis point of view: the base and the meeting pointbetween wall and roof. It is here that we find, all overIndia, bas-reliefs presented in the form of friezes. InSouth India, the finely carved narrative friezes of theHoysala corpus are found on the bases which seem tohave been extended to provide space for them. Yet ifthese placings are, in a text devoted to architecture, suchas the Maricisamhita, meant especially for kÒatriyas,9 it isperhaps because it is here that we meet with the storieswhich put the gods on the stage, very often doing battle,that is, behaving as kÒatriyas. Beyond the architecturallines of the temples, the placing of the Co pa narrative rep-resentations appear particularly well adapted to a sym-bolic function which stresses the horizontal or terrestrialelement of the mythologies represented. The storiesinvolved always have for their theme the incarnation of adeity, it may be one of the descents of ViÒnu or a formtaken by Kiva, or by the Goddess, to fight one or another

demon; in either case the deity descends to earth, and itis that deity’s adventures on earth which are depicted.

The contrast with the traditional images featured at the centre of the exterior walls of the temples offersclarification in this respect. In most of the temples underconsideration, a DakÒin amurti is found on the South wall,a Liœgodbhava or a Hari-Hara on the West and, on theNorth, a representation of Brahma, and also often of theGoddess standing on the head of a buffalo. These are allimages belonging to worship. Hierarchical and divine,they stand isolated in the vertical space of a niche, unlikethe representations of the series that are engraved hori-zontally. Near the ground, and defining a path of prayerwhich most often follows the pradakÒina, the figures fea-tured in the series correspond to the terrestrial aspects ofthe deities on the exterior walls.

The representations in Tirumaœgalam which corre-spond to the representations in the niches, in whichregard this temple is exceptional, confirm that aspect ofthe iconography of the series. This temple offers threesuccessive series of representations. The sculpted panelswhich take up the iconography used in the niches belongto the last series, the one closest to the niches. Some ofthese representations, in fact, flank the images in theniches. This last frieze in its series seems to be an inter-mediary between the horizontal series of narrative repre-sentations and the vertical images in the temple. This isthe only series whose representations are similar to those

Fig. 4 – Kañcipuram, Kailasanatha, BhikÒa†ana, 7th century.(Photo: EFEO-L’Hernault 78 32a).

6 2 9

of the niches: its functions as an intermediary betweentwo parts of the temple, the base and the exterior walls,may explain this. Lastly, as in the series at Tirumaœga-lam, though placed further down in the temple, most ofthe series are situated on the extensions of the pilasterspunctuating the walls of the buildings; their wide rectan-gle would certainly appear to be providing a base for theimages for the worship, both those on the niches andthose found inside the interior of the building.

The base of the temples is also seen to be an interme-diary space, close to the ground and connected with it,where divine incarnations gamble or play the roles giventhem in the texts, putting on the costume of the boar, ofthe naked mendicant or of the Goddess mounted on alion. This is the realm of the lila of the gods, the cosmicplay which intervenes between gods and humans. Theprescription in the M a ric i s amh i tamay also be understoodthus. According to some authors, Rama may be repre-sented on Copa Kaiva temples as the model of the king10

but I am not sure that in this corpus Rama has alwaysbeen considered as a royal incarnation. These templeswere not founded by kings; it is often a matter of recon-struction, and the link with royal power hardly appears inthe inscriptions except during a specific period at Tiru-bhuvanai which, once more, is seen to be an exception.11

It is, however, certain that the illustrations of theRamay ana, some of the forms of the Goddess and of Ki v adoing battle or, again, the stories of KrÒna, correspond to

the representation of the protective function of a godmanifesting in the world of mortal beings. In the Northof the Tamil land, one of the inscriptions on the site ofTiruvannamalai does link Kiva and Rama in a formulaintended to solemnise the gift made: “Kiva permeateseverything, Rama conquers all”.12 Dated to the 17th cen-tury, the inscription is later than the temples in this studybut appears as another manifestation of a religious stateof mind according to which all divine forms may be seenas forms of Kiva even though Rama is the conqueringform of the god, the one seemingly most fit to solemnisethe gift made to the gods on earth. Some of the icono-graphic caracteristics of the represenations allow for thesetting up of a more accurate hypothesis about the inte-gration of these VaiÒnava series in this Kaiva context.

The Theatre of the World

A large number of representations of Kiva appearingin these narrative series seem to be derived almostdirectly from Pallava art. The BhikÒa †ana of theKailasanatha of Kañcipuram (Fig. 4) is found again inthe reduced dimensions of, for example, Tiruchennam-pun∂i (Fig. 3), Pullamaœgai and Puñjai: the iconographyis quasi-stereotypical. The representations of MahiÒasur-a m a r d i ni are the most remarkable from this point of view.The panels of Tirumaœgalam (Fig. 5-6), of Pullamaœgai

Fig. 5 – Tirumaœgalam, Samavedikvara temple, 10th century,MahiÒasuramardini. (Photo: EFEO-Evenisse N 40 20).

Fig. 6 – Tirumaœgalam, Samavedikvara temple, 10th century,MahiÒa. (Photo: EFEO-Evenisse N 40 19).

6 3 0

Of Gods and Mortals: Lila C opa ❘ s c h m i d ❘

(Fig. 7), of Kumbakonam and of Paccil (Fig. 8) are con-structed on the same plan as that of the famous bas-reliefof MahiÒas u r a m a r d i ni in Mahabalipuram. It is mainly theKaiva scenes, including those that bring the Goddess intothe picture, which are constructed according to aniconography that hardly alters from temple to temple.13

Even if some of the representations of avataras are alsooften very like one another,14 the sculptors have shown

more inventiveness when treating a VaiÒnava subject towhich they were drawn to give more variety. This is not,in any case, because they lacked earlier models in thefield.

This different treatment of VaiÒnava and Kaiva scenesbrings us back to the place reserved for VaiÒnava imagesin a Kaiva context. Whatever pertains to the Kaiva domainseems to be more strictly codified. The VaiÒnava imagesmay be proofs of more inventiveness, perhaps justbecause they stand outside the field of the worship con-ducted in the temples. They are not, for all that, consid-ered as VaiÒnava, or at least not uniquely. It would seemthat they are stories first and foremost. The placereserved for the series indicates their links with the ter-restrial aspect of the things with which the temple is pro-vided. The exterior walls of these temples are coveredwith inscriptions in with mundane matters, such as taxes,redistribution of lands and cleanings of tanks, dominate.They appear to correspond to the skin or surface of thetemple, whose inner ritual is much more secret, fragileand therefore protected. The exterior walls and, evenmore often, the bases, are places of exchange with otherworlds, divine or perfectly human. The narrative panelsare imposed on the surface of the temple in contrast tothe niches which carve a space, a priori protective, out ofthe exterior walls. The predominance of the VaiÒnava tra-ditions is comprehensible in this case; ViÒnu is the incar-nating deity par excellence and it is unquestionablydescents of the deity into the human world which are rep-resented. Beyond these incarnations of the god, or simu-taneously with them, it is observable that there alwayswas a connection between the narrative mode and theVaiÒnava tradition. The epics are both VaiÒnava eventhough Kiva has some prominence in the Mahabharata.In the field of narrative representations the VaiÒnavamovements have always had primacy. Kiva is presented,first of all, in the transcendent form of the liœga butKrÒna, from the first images onwards, is a divine figure.The Puranas demonstrate that there is no lack of narra-tive strands with Kiva as hero. Puranic literature never-theless offers the same view of a definitive VaiÒnava pre-dominance as that found on the Copa bases; this issometimes explained in terms of efforts to brahmanize a“ n o n - V e d i c ”1 5 deity such as Ki v a .1 6 Right from the start ofbrahmanical narrative art, which may be dated from thesecond century CE with the appearance of representa-

Fig. 7 – Pullamaœgai, Brahmapurikvara temple, 10th century,MahiÒasuramardini. (Photo: EFEO-Evenisse N 36 32).

Fig. 8 – Paccil, Tirumexxatikvara temple, MahiÒasuramardini.(Photo: EFEO-Evenisse N 27 7).

6 3 1

tions of KrÒna fighting against Kekin at Mathura, thesmall numbers of Kaiva narrative representations isnoticeable.17 The fact that the epics are predominantlyVaiÒnava may be significant here yet it may equally besaid that the epics are VaiÒnava because the Kaiva move-ments were much less interested in the narrative genre.18

In the first Copa series the narrative mode appears, inany case, to be essentially VaiÒnava and, as with the plac-ing of the friezes, this VaiÒnava thematic must have beenconnected with their mundane character. In fact repre-sentations connected with the world of dance and music(Fig. 9) are quite numerous in these temples, on the orderof 15 and 20 per cent per temple. They are sometimesinserted for the sake of solution of continuity with thecycles, all the more so as the dancing forms of gods areequally present: those of Siva (Fig. 10) but also those ofKrÒna dancing with pots and playing the flute (Fig. 11).In the Hoysala temples this type of representation oftenmarked the boundary between one narrative cycle andthe next: it remains very common. We are familiar withreligious festivals whose atmosphere seems very close tothe presiding spirit of the representations on the bases. Atthe time of festivals dedicated to Kiva, plays were per-formed featuring the avataras of ViÒnu, such as theMahaviracarita and the Uttararamacarita of Bhavabhuti,which have Rama for their hero and were performed inhonour of Lord Kalapriya, one of the forms of Kiva. Theinscriptions evoke, too, the drummers who made theirinstruments resonate during those festivals just as they do

today.19 The drummers are by far the best representedmusicians on our narrative series and they often accom-pany Kiva dancing.

The presiding spirit of these representations inV a iÒnava narrative cycles appearing in series of otherwiseKaiva representations, and scenes of dance and musiccorresponds very well with the spirit of these rejoicings.The world of the theatre, of the divine action representedand played out on the earth, comes close to the CopaKaiva series on the bases. The way they are divided intosmall panels irresistibly evokes the sequence of scenes of

Fig. 9 – Puñjai, Naltuñai Ikvara temple, 10th century, musicians. (Photo: EFEO-Evenisse N 6 30).

Fig. 11 – Tiruchennampun∂i, Ka∂aiyar temple, 10th century, KrÒna playing flute. (Photo: EFEO-Evenisse N 24 28).

Fig. 10 – Pullamaœgai, Brahmapurikvara temple, 10th century, dancing Kiva. (Photo: EFEO-Evenisse N 37 22).

6 3 2

Of Gods and Mortals: Lila C opa ❘ s c h m i d ❘

a play in a theatre. The poses taken by certain charactersevoke the world of dance in the same way. The fightbetween two monkeys, Sugriva and Vali, in theRamayana, for example, is obviously in the form of adance (Fig. 12) and sometimes has no significancebeyond its decorative value. We find it represented com-pletely out of context in Kaiva temples such as that atDarasuram, where it is the only representation having todo with Rama but where it is, however, represented atleast four times. It may, too, have inspired the pose ofN a r a s imha fighting Hirany a k akipu, which also appears indistinctly dancing mode in, for example, the temple inTirumaœgalam (Fig. 13). The outstandingly aestheticnature of these representations where the wrestlers” limbscorrespond to one another emphasizes the theatricalcharacter of certain narrative representations. It is truethat other representations of deities, such as those of bat-tles or of leaping gods and demons, indirectly evoke theworld of dance and theatre. The base of a temple is a the-atre where the living gods appear and where “appearsadorned with peacock feathers, red ochre and with buds,Mukun∂a-KrÒna who sometimes dresses up to play therole of a wrestler, friend !”.20

The Lord of the Stage

The spirit which animates the theatrical representa-tions in the temples is close to the little world populatingthe narrative series and it evokes one of the fundamentalconcepts of the VaiÒnava tradition. André Couture(2001) in the course of a study of the use of vocabularyreferring to the idea of the a v a tar a, brings to light the spe-cial link apparent in the M a hab har a t a, the H a r i v am kaa n dthe Puranas between the world of theatre and the mani-festations of ViÒnu. This connection seems to me to beilllustrated by the narrative series of the Co pa temples too.Avatarana designates, in the theatre, the entrance of theactors onto the scene, raœga, just as the same term desig-nates the appearance of the god here below. The fight ofKrÒna against his uncle’s wrestlers staged by Kamsa, ona r ang a, in the town of Mathura, is also represented as thescene in which are revealed the powers of the god whomanifests himself in the world. It is, however, the wholecollection of episodes from the childhood of KrÒna, in theHarivamka, which must be considered as an entranceonto the scene, an avatarana.21 It may be said that whatappears on the scene is the visible form of the god,human or visible to the human eye, which KrÒna evokesin the Bhagavadgita (XI, 8), that which plays on earthamongst men and women after having put on the cos-

Fig. 12 – Tirumaœgalam, Samavedikvara temple, 10th century, the fight of Sugriva and Vali. (Photo: EFEO-Evenisse N 42 5).

Fig. 13 – Tirumaœgalam, Samavedikvara temple, 10th century, the fight of Narasimha and Hiranyakakipu. (Photo: EFEO-Evenisse N 39 20).

6 3 3

tumes (veÒa, as quoted by Couture, Id.: 323) which giveshim a human or half-animal form.22

A detail of the organisation of the narrative cycles inthe space of the Copa temples corresponds very closely tothat textual aspect of the manifestation of the a v a tar a s. Inour Copa corpus, the representations of ViÒnu KeÒakayi,ViÒnu sleeping on a serpent, precedes the entrance ontothe scene of the avatara at Pullamaœgai (Fig. 14) andTirubhuvanai (Fig. 15). In the Hoysala corpus, that fig-ure marks in the same way, but in a quasi-systematicmanner, the beginning of a narrative cycle. It may corre-spond to the beginning of the Ramayana, where the godscome to ask ViÒnu to incarnate, or, in a more general way,with the beginnings of the stories of avataras when thegods come in procession to ask Narayana to descend andrelieve the Earth of her burden. But, in fact, this form ofthe god appears to be related in a very general way, sincethe first images of it in the North of India, to manifesta-tions of the god, very often as avatara. Uniting humanand animal forms in one double-bodied god, ViÒnu rest-ing on a serpent deity, reveals a flesh and blood body ris-ing up from another world invisible to mortal eyes as anintermediary between the world of the gods and that ofhumans. His image representing a transition from thenon-manifest to the manifest, is very well adapted to therepresentation of the beginnings of the narrative cycleswhich are the cycle of the a v a tar a s. In Tamil Nadu more-over, that form of the god seems to maintain a specificrelationship with the idea of theatrical representations;ViÒnu on the serpent is known by the name of Raœga-natha, the Lord of the Stage. This cosmogonic figure ofViÒnu sleeping on his elder brother, the serpent, marks

the introduction of a new universe. In the Copa cycles,amongst others, the act of creation by the Lord of theStage may corresponds to the introduction of the newscene, the raœgavatarana, the importance of which isemphasized by Couture, for it must be set out so that thea c t o r -a v a tar a, usually KrÒna or Rama, is able to manifest.The Apvars in their poetry like to place the figure of ViÒnuon the serpent before the evocation of his incarnations.23

In Tamil Nadu the special link between the figure ofV iÒnu sleeping on the serpent with the idea of the a v a tar aof god as the visible form of the deity appears from thefirst known images onwards. First of all, these figures aremost often the ones carved in situ, as the image dated tothe 7th century in the Shore Temple of Mahabalipuram.The meeting with the terrestrial world where the divinemanifestations take place is marked, above all, by aninalienable relationship between that form of the god andthe place where he appears. The earth takes the form ofgod; god gives her her form. Moreover this title “Raœga-natha” belonging to this manifestation of the god on theserpent deity is profoundly connected with ViÒnu restingon a serpent in the island town of Kriraœgam, “TheStage”, a divine “scene”, as it were, at the beginning ofthe delta that irrigates the region where the Kaiva templesof our corpus are situated. The name of this small islandappeared in VaiÒnava devotional poetry, many of whosepoems are dedicated to Kriraœgam. The concept of thegod of the raœga at Kriraœgam seems to be even older andpart of the mythical landscape which the isle is: a frag-ment of land emerging from the waters just as does, in aVaiÒnava framework, the whole created universe.24 Theprofile of Kriraœgam, washed by the Kauveri, evokes that

Fig. 14 – Pullamaœgai, Brahmapurikvara temple, 10th century, Ranganatha.(Photo: EFEO-Evenisse N 38 16).

Fig. 15 – Tirubhuvanai, Varadarajaperumal temple, 11th century, Raœganatha. (Photo: EFEO-Ravindran E 75 2).

N O T E S

1 The narrative friezes of the Melaikka†ampurtemple, dating from the beginning of the12th century (the first inscription is from 1113),are mainly consecrated to the Nayaqmars. Thesame Kaiva saints also appear in certain placeson the last frieze of the Tirumaœgalam temple,which seems to be later than the other two, andare found again on the pillars at Darasuram, dat-ing from the second half of the 12th century (thefirst dated inscription in this temple correspondsto the year 1167; for P.N. Srinivasan in L’Her-nault 1987: 16, the temple was completed beforethe 13th year of Rajaraja II, which means before1160). It is possible that this non-puranic Kaivacorpus had been exploited only from the11th century because of modifications in the ori-entation of religious movements it is impossibleto be more exact about at the moment.2 The pilasters of Copa temples often have a dec-orative band on the upper tiers. At Tirumalpurthis band bears the narrative representations

which appear on the bases in the rest of the cor-pus. On the pilasters below, note 8.3 On the dating of this temple see Balasubrah-manyam 1971: 56-58, 22.4 South Indian Inscriptions, Vol. XIII, no. 257,vol. XIX, no. 63, 138, 168, 188; Varalaru, vol. 2,no. 16 & vol. 3, no. 9.5 The epigraphic corpus of the Varadarajaperu-ma l of Tirubhuvanai is to be published. Some ofthe inscriptions are from the 10th century but itis not clear whether they belonged to the originalbuilding which was reconstructed in the thirties,as Pattabiramin (1948: 27) pointed out.6 South Indian Inscriptions, vol. XIII, no. 29-34,105, 132, 330, vol. III, no. 142 and 197, vol. XIX,no. 197, 220, 318, 319, 341, 445; EpigraphiaIndica 1906: 265-344.7 The only series situated on the base of exteriorwalls is found at Tirumaœgalam. Considered aslater by nearly a century than the first two seriesof representations, it is, with the unique series ofTirubhuvanai dated from the beginning of the11th century, the one which is placed in the most

visible position in the temple. It appears that theseries have a tendency to move upwards on thebuilding of the temple, finishing in the Hoysalacorpus where they are most often found at eyelevel. Elsewhere, the iconography of the narra-tive series is sometimes used on the pilasters ofa temple. This is certainly the placing chosen forthe representations of a cycle about KrÒna atTirumalpur, probably towards the middle of the10th century. At Puñjai, one of the pilasters atthe entrance to the shrine is decorated in thesame place with a representation of the Goddessdancing, which is found again in the secondseries of the temple. On the South face of thetemple of Pullamaœgai, one of the pilasters isdecorated with representations of KrÒnaepisodes. On the temples of Paccil and Tiru-maœgalam, various of the pilasters are orna-mented by representations of gods. AtPa††a∂akal, 7th century, the interior pillars ofsome temples bear narrative representations:perhaps these Copa pilasters are reviving a moreancient tradition. In the second half of the

6 3 4

Of Gods and Mortals: Lila C opa ❘ s c h m i d ❘

land encircled by water which ƒAn†al presents as submis-sive to the lord of Araœkam (Krirangam):

“Sovereign Lord who faultlessly rules / the heavenlyspheres, / the earth / encircled by billowing seas– / mylord of splendour / of scepters and crown / who reigns inholy Arankam– / surely my bangles can add naught to hisperfections.” (Nacciyar Tirumopi, 11, 3; trad. Dehejia1990: 115-116).25

The Kauveri represents oceanic turbulences as well asthat eternal sea symbolized by the serpent, on which reststhe god, qualified to create the world wherein tinkle thebracelets of ƒA n † a l. The black divinity sleeping on the ser-pent is the Lord of the stage whose manifestation mustneeds precede that of the avatara. He is the god whobrings into being the earthly scenes. The evocation, in theP a r i pa †a l, of the city of Madurai/Mathura draws forth theimage of that scene which first appears in the navel of thegod:

“The renowned town is comparable to a lotus flower/ blossoming on the umbilical cord of the Dark one. Herstreets are the same / as the petals of the flower; and thesame as the precious fruit / which is in the middle of thesepetals, the palace of the Lord” (Paripa†al, fragmentno. 8).26 At the centre of the world stage, the temple,where the gods play, opens its petals.

The narrative model gives the shape to the bases ofthese temples. It is the first level of interpretation, in thesame way as is the mythology recited to an attentive audi-ence who may sometimes perhaps have seen it playingbefore their very eyes just as it is shown on the base of thebuildings. The deities it brings onto the scene introducethe people of earth to the worlds of the gods in an inter-mediary space – earth and sky, interior and exterior – ofthe temples which facilitated the meeting of several uni-verses, human, divine and religious. Images and texts alsomeet in this space. The correspondences, both estab-lished and impossible to establish, between these repre-sentations and the texts to which we have access, make usaware of a number of narrative traditions. The concept ofa v a tar a, as it appears in certain texts, and especially in theHarivamka, allows us to explain, at least to a certainextent, the important place that Kaiva temple complexesdevoted to VaiÒnava representations. The base of thetemple reveals itself as a space where gods, demons andmen meet in the midst of the prominent musicians anddancers: the theatre of the world. The succession ofscenes occupying the panels evokes a theatrical action,which in the Harivamka corresponds to the concept ofavatarana, the appearance of God on earth, conceived asa scene from a drama where men and gods mingle: thenthe story begins.

6 3 5

12th century, the pillars of Darasuram (seeabove note 2) are covered with representationsfrom bottom to top.8 On the VaiÒnava representations of theVaradarajaperumal see Schmid 2002.9 See the study of this text by Gérard Colas1983: 6.10 See for example Sanford 1987: 279-280 whoemphasizes the epigraphical comparisons ofCopa kings with Rama. There is without doubt astrong connection between Rama and the royalfunction but Rama is above all a divine form, agod descended here below and mixing withhuman beings: an avatara.1 1 The site of Paccil also seems weak as a linkbetween representations of Ramay ana and theking: three temples, two Kaiva and one V a iÒna v a,form a Copa group on the site. The Kaiva t e m p l e ,other than that carrying narrative cycles, bearsseveral inscriptions linking it to the royal familyeven though no single inscription qualifiying asroyal appears in the Tirrumex xa†i kvara. In a con-ference entitled “Mapping Movement in SacredSpaces in Medieval Tamil Nadu”, presented atthe “36th International Congress of Asian andNorth African Studies” in year 2000, at Mon-treal, Leslie Orr raised the question of the impor-tance of the link between the Copa kings and thetemples. The examples considered by the authorcorrespond to famous foundations such as Chidambaram and Tiruvannamalai as well as to lesser known temples such as those of K a t t u m aq qa r g u d i .The network of relationships between these tem-ples gives little place to the king (his capital, hisfoundations, etc.). In the case of temples bearingnarrative cycles it would certainly seem that thelink with the king is in fact a very distant one.12 civamayam ramaceyam, Srinivasan 1990; aninscription on a copper plate dated from the 17thcentury and including Kaiva representations.13 This type of example suggests that there existsa very close relationship, at least in the icono-graphical field, between the first Copa templesdecorated with narrative cycles. It is possible thatthe same teams of craftsmen worked on thesetemples. The style of the panels however issometimes different from one temple to another(on occasion even in the same temple) and thechronology would rather indicate that teams ofcraftsmen transmitted the models, perhaps fromfather to son, and we may also imagine icono-graphical models circulating between teams ofsculptors.14 There are two very similar representations ofthe boar avatara, above p. 8.15 Kiva is unarguably a Vedic divinity; he has aperipheral status in the Vedas however and issometimes contrasted with ViÒnu who is consid-ered to be more orthodox.16 See for example Rocher 1984 who, bringingup the question of the sectarian character of thePuranas, emphasizes the elements of the reci-procity between Kaiva and VaiÒnava movement

in the same texts, rather than considering the tolerance and intolerance of the two movementstowards one another. The iconography of theVaiÒnava temple at Tirubhuvanei may corre-spond to a more sectarian tendency (vediclyinspired?) among other tendencies.17 The Kaiva narrative representations from theGupta period are very scarce: a terracotta fromR aœg-Mahal in Raj a s t han, representing thedescent of the Ganges, another which belongs tothe site of Ahichattra in Uttar Pradesh and rep-resents the destruction of the sacrifice of DakÒa,(NMD 62.238, Kreisel 1986: A 27), and a sand-stone bas-relief from Mathura of the demonRavana shaking mount Kailasa (MM 2577, Id.:160, 238-239, pl. 109).18 One of the themes from the Mahabharata,which is found in three different temples,(Tiruchennampun∂i, Kumbakonam and Puñ-jai) is that of the Kiratar j u niya which puts on thescene a VaiÒnava hero in the framework of Kaivadevotion. This must without doubt be consid-ered in the same category of representation asthe adoration of the liœga by Ravana at Kum-bakonam. It is a matter of the integration of aVaiÒnava corpus used as a narrative corpus inthe Kaiva totality of the temple. In the Hoysalatemples this myth is also often found, as in thetemples of Javaggal, photos Evenisse-EFEO N45 8-10.19 Inscription no. 25 of Rajaraja I found on theinterior gopuram of the Brhadikvara of Tañjorementions the donations given to the musiciansbeating their drums at festivals dedicated to Kiva(South Indian Inscriptions, vol. II: 125-126.Two of the six Copa inscriptions of the temple atPaccil (see above p. 6) evoke the same world;one is a donation to a singer and the other adonation made for the dances (maima††i) andthe drums (kara†ikai).2 0 b a r h ina s t a b a k a d hatupallavair baddhamallapari-b a r h a v i∂a m b aÌ / karhi cit sabala ali sa gopair ga Ìs a mahvayati yatra mukundam //. The translationgiven in the text corresponds with the first half ofthe verse only.21 A. Couture (2001: 322) offers a quotationwhich is particularly clarifying in this respect:tavavatarane viÒno kamsaÌ sa vinakiÒyati, “ OViÒnu, at the moment of your descent (orentrance) [on the stage at Mathura], Kamsa willperish”, Harivamka 44, 82, this may apply to thebirth of KrÒna at the same time as to his appear-ance on the stage where he must fight against thewrestlers in the pay of his uncle.22 It is noticeable that the animal avataras,Varaha and Narasimha, who often appeared onCopa temples are always represented in a semi-human form. This is, it is true, the most com-mon form but the Varaha avatara may equallyappear in an entirely animal form as sometimesoccurs when it is the boar who is searching forone of the extremities of the liœga of Kiva.23 See for example the Periyatirumopi of Tiru-maœkai Apvar, 3, 1 and the Nacciyar Tirumopi of

ƒAn†al 11, 3; and there are many other examples.24 T.V. Gopal Iyer and V. Desikan of the centreof the EFEO in Pondicherry gave me clarifyingindications on the appearance of the term raœga,tamil a r aœg a m: it appears in the C i l a p p a t i k kar a min circumstances which seem to indicate that in this text it was already linked to the town ofKriraœgam as a place suitable for the manifesta-tion of Raœganatha. In the 10th canto of theP u kar k kan †a m, the word appears several times inits Sanskrit meaning of stage but one of theseoccurrences is an equivalent of the tamil turutti,little island. The description of a grove is pre-sented thus (though I have no direct access toTamil texts, I quote the text itself for the benefitof specialists): “axxu vi arankattu vixxu vixxaki”which may be translated as “[the grove] majesti-cally situated in a unique manner on an tinyisland which divides the river in two”. We mayalso translate this as “the stage which divides theriver in two” but a stage which would be specif-ically related to that which forms the isle wherethe divinity of Krir aœgam resides. It seems that inthis tradition the Kriraœgam deity appears as thedivinity par excellence of the stage, to the pointof conferring on a term of Sanskrit origin ameaning derived from the geographical positionoccupied by the temple of the god. A fragmentof earth which emerges from the cosmogonicwater flowing all around it, the isle is an image ofthe world created by the god and, as such, a rep-resentation of the stage where he manifests.25 ponkotam cupnta pavaqiyum vinnulakum / ankatuñ corame yapkiqxa vemperumaq / cenko lu†aiyatiru varankac celvaqar / enkol vapaiya li†artirvarakate.26 mayoq koppup malarnta kamaraip / puvera†upuraiyuñ cirup puv iq / itapakat taqaiya teruvamitapakat / tarumpotu††aqaittai yaqqal koyil.

Acknowledgements: Heartfelt thanks to FranckEvenisse whose photographs illustrate this articleas well as to everyone at the centre of the Écolefrançaise d’Extrême-Orient in Pondicherry andespecially to V. Desikan, N. Ramaswamy(Babu), Shanti Rayapoulle and R. Satya-narayanan who helped me to collect, organizeand interpret the documentation.

6 3 6

Of Gods and Mortals: Lila C opa ❘ s c h m i d ❘

R E F E R E N C E S

Annual Report on South Indian Epigraphy for1929-1930, Madras.

Annual Report for Indian Epigraphy for 1992-1993 (1998) Archaeological Survey of India.New Delhi.

Balasubrahmanyam, S.R. (1971) Early CholaTemples. New Delhi.

Barrett, D. (1974) Early Chola Architecture andSculpture. London.

Le Bhagavata-purana, traduit et publié par E.Burnouf, IV par M.-L. Hauvette-Besnault.Paris, 1884.

Cilappatikkaram, ed. non accr. Tirunelveli &Chennai, 1942.

Colas, G. (1986) Le temple selon Marici, Extraitsde la Marici-samhita étudiés, édités et traduits.Pondichéry.

Couture, A. (2001) From ViÒnu’s Deeds toViÒnu’s Play, or Observations on the WordA v a tara as a Designation for the Manifestationof ViÒnu. Journal of Indian Philosophy 29.

Dehejia, V. (1990) ƒAn†al and Her Path of Love,Poems of a Woman Saint from South India. NewYork.

Epigraphia Indica (1966), Archaeological Surveyof India. Calcutta/Delhi, 1966.

Evans, K. (1997) Epic Narratives in the HoysalaTemples, the Ramay ana, Mahab harata andBhagavata Purana in Halebid, Belur and Amrt-apura. Leiden.

Hardy, F.(1983) Viraha-Bhakti, The Early His-tory of KrÒna Devotion in South India. Delhi.

H a r i v amka, critical edition by P.L. Vaidya.Poona, 1969-1971.

Harle, J.C. (1958) The Brahmapurisvara Templeat Pullamangai. Bombay.

Hawley, J.S. (1983) Krishna, the Butter-thief.Princeton.

L’Hernault, F. in collaboration with Srinivasan,P.R. & Dumarçay, J. (1987) Darasuram, Epi-graphical study, Étude architecturale, Étudeiconographique. Paris.

Hoekveld-Meier, G. (1981) Koyils in the Cola-mandalam: Typology and Development of EarlyCola Temples. Amsterdam.

Kaimal, P. (1999) The Problem of Portraiture inSouth India, circa 870-970 A.D. Artibus AsiaeLIX, 1/2.

Kreisel, G. (1986) Die Kiva-Bildwerke derMathura Kunst: Ein Beitrag zur frühhinduistis-chen Ikonographie. Stuttgart.

M a hab har a t a, critical edition by V.S. Suk-thankar. Poona, 1969-1971.

M a h a l i n g a m , T.V. (1967) The Nag ekv a r a s vam iTemple. Journal of Indian History. vol. XLV/1.

Nalvayir Tivvyar Pirapantam, Tiruveenka. tat-taq Tirumaqxam. Chennai, 1973.

Le P a r i p a†a l, texte tamoul, Introduction, traduc-tion et notes par F. Gros, Pondichéry, 1968.

Pattabiramin, P.Z. (1948) Quatre vieux templesdes environs de Pondichéry. La Revue His-torique de l’Inde française 7.

Rangaracharya, V. (1919) A Topographical List ofthe Inscriptions of the Madras Presidency (col-lected till 1915). 3 vol., Madras.

Rocher, L. (1984) Kiva in the Puranas. Dis-courses on Kiva, Proceedings of a Symposium onthe Nature of Religious Imagery, ed. M.W.Meister. Bombay.

Sanford, D.T. (1974) Early Temples BearingRamayana Relief Cycles in the Chola Area: AComparative Study. Ph.D. dissertation. LosAngeles.

Sanford, D.T. (1987) Miniature Relief Sculp-tures at the Pullamangai Siva Temple, withSpecial Reference to the Ramayana Sequence.Kusumañjali, ed. M. S. Nagaraja Rao, II.Delhi.

Schmid, C. (1999) Représentations anciennesde KrÒna luttant contre le cheval Kekin sur deshaltères : l’avatara de ViÒ nu et le dieu duMahabharata. BEFEO 86.

Schmid, C. (2002) Divines aventures de KrÒna:la lila et les traditions narratives co pa, Arts Asia-tiques t. 57, pp.33-50.

South Indian Inscription, vol. II, III, VII, XIII,XIX, 1891-1967, Archaeological Survey ofIndia. Delhi.

Srinivasan, P.R.(1990) Tiruvannamalai, a kaivasacred complex of South India, 1.2, Inscriptions.Pondichéry.

Le Tiruppavai d’ƒAn†al, Un texte tamoul dedévotion vishnouïte, J. Filliozat. Pondichéry,1971.