Addition, Erasure, and Adaptation: Interventions in the ...vidyadehejia.com/wp...Davis-Addition-Erasure-2010.pdfAddition, Erasure, and Adaptation: Interventions in the Rock-Cut Monuments

Addition, Erasure, and Adaptation: Interventions in the Rock-CutMonuments of MÄ•mallapuram

Vidya DehejiaRichard Davis

Archives of Asian Art, Volume 60, 2010, pp. 1-18 (Article)

Published by University of Hawai'i PressDOI: 10.1353/aaa.2010.0001

For additional information about this article

Access provided by Columbia University (25 Mar 2013 06:44 GMT)

http://muse.jhu.edu/journals/aaa/summary/v060/60.dehejia.html

http://muse.jhu.edu/journals/aaa/summary/v060/60.dehejia.html

Addition, Erasure, and Adaptation: Interventions in theRock-Cut Monuments of Mamallapuram

vidya dehejia

Columbia University

richard davis

Bard College

The magnificent set of excavated caves, rock-cutmonoliths, sculpted stone tableaux, and structural

temples at the Pallava-period (ca. 575–728) port townof Mamallapuram (in present-day Tamil Nadu) hasattracted the attention of a wide range of scholars. Thefascination exerted by this unique site is demonstratedby the existence of two annotated bibliographies, onepublished in 1966 and a second barely fifteen yearslater in order to bring the scholarship into the year1980.1 Although scholars agree that the monuments atMamallapuram date between the late sixth and earlyeighth centuries, during the reign of the Pallava rulersof the Simhavis

˙n˙u line, unresolved questions proliferate

regarding their precise authorship, exact date, and in-tent; the incompleteness of many of the monuments addscomplexity to the conundrum they pose.

Most studies of Mamallapuram have focused onthe dominant monuments of the Pallava period, andfor good reason. But the structures commissioned bythe Pallava patrons continued in existence, and throughthe following centuries they attracted later visitors whosometimes imposed themselves upon the site throughtheir own material practices. These might include any-thing from the addition of small graffiti-like markingsonto existing monuments to the creation of larger newstructures that overshadowed the previous ones. Thingsmight be added, or sometimes subtracted, as in thedeliberate erasure of the identifying features of sculptedimages. Surprisingly little attention has been paid to thematter of physical change at Mamallapuram. Numerousclear signs of adaptation and elimination of imageryappear upon and within the rock-cut structures at thesite, but with the sole exception of Michael Lockwood’sstudies, art historians have shown a sort of benignneglect toward these changes.2 In this essay we proposeto attend more closely to the structures that show signsof alteration, erasure, and adaptation, as a preliminaryinquiry into change at this important site.

Material Changes and Religious Competition

The purpose of this study is to identify some of themarks of later intervention on the fabric of the stoneart and architecture at Mamallapuram, and to accountfor these as intentional activities. Through this effortwe hope to learn from these acts something about thelong-term history of Mamallapuram as a continuingsite of religious art.

It is not always easy to locate the times when theseinterventions took place. After all, iconoclasts do notusually sign their works of erasure.3 South Indian dis-cussions of later changes to early medieval sites oftenrefer to ‘‘the Muslims’’ and their alleged rampages ofthe fourteenth century as an all-purpose explanationfor all destructive alterations to Hindu religious struc-tures. But this reflexive narrative is clearly inadequate,as well as politically nefarious.4 In this study we shallsee that the most relevant religious conflict was notbetween Hinduism and Islam, but between votaries ofSiva and those of Vis

˙n˙u. That long-running competition

between followers of two primary Hindu deities, withthe balance of support and power shifting over time,accounts for subtle and not-so-subtle acts of materialintervention at Mamallapuram. In this study we havenot sought to explore another striking aspect of reli-gious competition in medieval South India, namely,that between the Saiva and Vais

˙n˙ava devotees and those

Jains and Buddhists they defined as Others. This is astriking theme in the songs of the south Indian poet-saints, the Saiva and Vais

˙n˙ava saints, known respec-

tively as the nayanmars and alvars, but we do not seematerial evidence of this conflict in the monuments ofMamallapuram.5

The fourteenth century saw a resurgence of Vais˙n˙a-

vism across south India, and the building at Mamalla-puram of the Sthala-sayana Temple to enshrine a stuccoimage of Vis

˙n˙u reclining (sayana) on the ground (sthala).

This temple was erected directly in front of the GreatPenance relief of the earlier Pallava monarchs, partlyobstructing the view of the cliff face with this greatsculpted tableau, which until then had been a focalpoint of the site.6 The new Vis

˙n˙u temple’s gateway

(gopuram) led through a stone pillared hall to the ocean,where Pallava monarch Rajasimha (r. 700–728) hadbuilt his Shore Temple; by erecting double shrines toSiva, he had enclosed the original rock-cut image ofVis

˙n˙u reclining on the seashore, thus relegating it to

secondary importance. This demeaning of Vis˙n˙u was

now reversed. During this Vijayanagara phase (ca.1336–1565) of Vais

˙n˙ava revival, certain rock-cut struc-

tures at Mamallapuram dedicated to Siva were sub-jected to a drastic erasure of their relief sculptures andin one instance even to the eradication of shrine walls,apparently with the intention of deflecting and reroutingsectarian affiliation in favor of Vis

˙n˙u.

As noted above, however, evidence exists of thereverse scenario during the Pallava period, when newepigraphs threatening the descent of divine wrath onthose who did not worship Siva were inscribed in anattempt to divert the original Vais

˙n˙ava affiliation of

certain structures. It seems possible too that the richlysculpted Pallava Mahis

˙a-mardinı Cave was originally

intended as a Vais˙n˙ava dedication and was altered soon

after its creation, during the Pallava period itself, toglorify Siva instead. Although the Pallava period in SouthIndian history has often been cited as a prime exampleof ‘‘Indian tolerance,’’ we need to qualify this somewhatin light of the physical remains at Mamallapuram.

Vis˙n˙u, Siva, and the Goddess at Mamallapuram

By and large early Pallava monuments reveal a remark-ably open and ecumenical approach to the worship of arange of Hindu deities. A consideration of the inscribedshrines that Mahendravarman Pallava (r. ca. 610–630)excavated across the Tamil country reveals three caveseach enshrining a linga; one cave dedicated to Vis

˙n˙u;

one cave with an unspecified, perhaps painted, image;one triple-shrined cave dedicated to Brahma, Vis

˙n˙u,

and Siva; a second triple-shrined cave of undeterminedaffiliation; as well as a cave with five shrines. Thevariety of enshrined deities suggests that Mahendraadhered to the sastric ideal of a monarch who, regard-less of his own personal affiliation, was expected tobuild and maintain temples to the entire range of Hindugods worshipped by his citizens, and to join withthem in their celebration of the varied festivals.7 TheArthasastra and the Sukranıtisara are among the texts

that enjoin this as a royal duty. This is parallel to thedevotional attitudes of the poet-saints of the period, theVais

˙n˙ava alvars and Saiva nayanmars, who reserved

their antipathy for the Buddhists and Jains while largelyaccepting the full pantheon of Hindu deities. As weshall see, however, not all Pallava rulers were equallyeven-handed in their patronage of the various Hindugods.

A similarly diverse group of dedications exists at thePallava port town of Mamallapuram, where the greaternumber of its monuments were created out of the mas-sive granite outcrop at the center of the site. Two cavesare dedicated to Vis

˙n˙u, as is one sculpted cliff-face that

portrays the Kr˙s˙na legend; three structural temples and

three rock-cut monuments are Saiva and house botha Somaskanda (Siva accompanied by wife Parvatı andson Skanda) panel carved into the shrine’s rear walland a linga at its center; one cave and one monolithare dedicated to Durga; and an unidentified shrine mustonce have housed the magnificent Camun

˙d˙a (epithet

of Kalı) image that is today cemented into a platformin the grounds of the local library. Several other monu-ments are of mixed dedication: these include the five-shrined Koneri Cave and the so-called Trimurti Cavecontaining images of Skanda, Siva, and Vis

˙n˙u, with an

additional niche for Durga sculpted along the facade.Indeed, even the magnificent sculpted relief of the GreatPenance, measuring roughly forty-five feet high byninety feet wide, appears to fall into this category;despite the story/stories revolving around the appear-ance of Siva, a place of prominence is given to a Vis

˙n˙u

shrine carved on the rock face directly below Siva.Several of the monoliths, including the so-called Arjunaratha, Bhıma ratha, and Nakula-Sahadeva ratha remainof uncertain affiliation because their deity images aremissing. (The names identifying these monoliths wereapplied in a much more recent period.) A number ofMamallapuram caves and monoliths were abandonedat an early stage of their creation, and afford but meagerclues to their original intended dedications.

Even from this admittedly incomplete listing, we cansee the remarkable diversity of affiliation with variousdeities at Mamallapuram. We should note here thatseveral caves, whether containing one, three, or fiveshrines, are missing their original enshrined images.The identification of the original image can be deducedin many instances from the character of the doorguardians flanking the shrines, as we will see below.For example, matted locks curving in to frame theneck, and in some instances a horned headdress, dis-tinguish Saiva door guardians; female door guardians

2 ARCHIVES OF ASIAN ART

indicate a Durga shrine; bearded guardians often sug-gest a Brahma image; and those wearing a cylindricalcrown are suggestive of Vis

˙n˙u.

An additional intriguing feature at Mamallapuramis the pairing of the goddesses Laks

˙mı and Durga,

consorts respectively of two very different deities, Vis˙n˙u

and Siva. One would not expect to find these twogoddesses together, but they flank the single, centralshrine in two major Vais

˙n˙ava caves, the Varaha and

the Adi-Varaha. Durga is identified across India withUma-Parvatı, consort of Siva, but only in south India isshe also considered to be the sister of Vis

˙n˙u.8 It is the

latter relationship that allows her an accepted positionwithin Vais

˙n˙ava caves. Roughly parallel is the iconogra-

phy in the Mahis˙a-mardinı Cave. On its left wall Vis

˙n˙u

reclines upon his serpent couch prior to the creation ofthe world, while on the opposite wall the lithe youngDurga, mounted upon her lion, battles the buffalo-demon Mahis

˙a, whose form, carved as a retreating

diagonal, signifies the ultimate victory of the goddess.It seems likely that Durga’s accepted position as sisterof Vis

˙n˙u may have been responsible for this unusual

pairing.

Dvarapala (‘‘door guardian’’) Types

Since the systematic erasure of imagery in certain rock-cut structures at Mamallapuram includes, and is occa-sionally restricted to, door guardians, it will be usefulhere to note that distinctive types of dvarapalas arefound flanking the shrines of early Pallava monuments.Further, the distinctive door-guardian types enable usto identify the affiliation of structures where the centralimage or icon is no longer present.

To demonstrate the sharp distinctions apparentamong dvarapala types, we shall briefly consider fourcaves, three housing triple shrines and one housing asingle shrine. The evidence is circumstantial, and seem-ingly circular in its argumentation and, in part at least,based on Mahendra’s inscription at Man

˙d˙agappat

˙t˙u,

which speaks specifically of its single cave having beendedicated to Brahma, Vis

˙n˙u, and Siva. In the above-

mentioned four caves guardians of a Siva shrine haveheavy curved jat

˙as (‘‘matted locks’’) hanging downward

to frame their faces, and each rests his hands on amassive club that is usually encircled by a serpent. Insome groupings one of the two Saiva guardians displayshorn-like protrusions behind his head, and sometimesthe other guardian has a small curved semicircle pro-truding from his headgear. Guardians of a Brahmashrine resemble sages; they tend to be bearded, often

with hair piled up on their heads, and to wear a lowergarment that reaches to their ankles. Guardians of aVis

˙n˙u shrine tend to be slender figures wearing a crown,

and frequently stand in exaggerated contrapposto. Whenwe turn to the rock-cut caves at Mamallapuram, weshall see that the erasure or alteration of door guardiansoften accompanied attempts to change the affiliation ofstructures dedicated to either Siva or Vis

˙n˙u, the primary

deities.Maman

˙d˙ur Cave II, known from later inscriptions

as the Saiva Rudravalisvaram Cave, is located nearKancipuram, the inland Pallava capital. It is a typicalMahendra-style excavation with two rows of columns,each row with two columns in full round and twoattached to their respective side walls, each cut squareabove and below and octagonal in the middle. Cut intothe rear wall of the cave are three shrines, with eachdoorway flanked by door guardians. Those guard-ing the southernmost shrine are bearded and appearsage-like, with hair massed on top of their heads. Theywear long lower garments and sacred threads, and eachholds a lotus in the hand nearest the doorway. Suchdoor guardians would have most likely flanked theentrance to a Brahma shrine. The central shrine, dedi-cated to Siva, has door guardians leaning on heavyclubs, their hair in curved, matted locks, or jat

˙as, on

either side of the neck, and large rounded earrings(patra kun

˙d˙alas). The dvarapalas of the northern shrine,

presumably intended to house a Vis˙n˙u image, stand in

graceful contrapposto (tribhanga), with one hand rest-ing on the thigh and the other gesturing toward theshrine. Though no Mahendra inscription exists to con-firm that the Maman

˙d˙ur Cave II was intended to house

shrines to Siva, Brahma, and Vis˙n˙u, the dvarapala types

are suggestive of such an intention. We may recall thatthe inscription in Mahendra’s cave at Man

˙d˙agappat

˙t˙u

specifies exactly such a dedication; unfortunately for ourstudy, the Man

˙d˙agappat

˙t˙u Cave, probably Mahendra’s

very first excavation, displays guardians only along itsfacade and not flanking its three shrines, as was soonto become standard.

A second Mahendra-style excavation, also withthree shrines in its original formulation, is the Kalman-dakam at Kuranganilmut

˙t˙am, only two miles away from

Maman˙d˙ur. The two guardians flanking the southern

shrine each wear a sacred thread, differently arranged,and stand facing forward in slight tribhanga with outerhand on the hip and the inner holding an indistinctobject that is probably a lotus. They have neither beardsnor massed chignons, and yet it would seem that theshrine must be dedicated to Brahma. The door guardians

VIDYA DEHEJIA and RICHARD DAVIS � Addition, Erasure, and Adaptation 3

of the central cave are typically Saiva, each resting hisinner hand on a heavy club wound around with a serpent.The southern image of the pair is in profile, while hiscompanion is frontal. Both wear a sacred thread in theform of knotted snakes, and both wear a crown abovethe curved jat

˙as that frame the neck. Additionally, the

southern one of the pair sports a large curved horn onthe side of his face turned toward the doorway. Theguardians of the northern shrine, presumably Vais

˙n˙ava,

stand in exaggerated contrapposto, with one hand onhip and the other gesturing toward the shrine; bothwear a slightly tapering cylindrical headdress and asacred thread.

The third triple-shrined excavation to be consideredin this context, the Trimurti Cave at Mamallapuram,appears to provide a windfall, since it preserves all threeshrine images as well as its three sets of door guardians.Yet identifying their icons is not as simple as it mighthave been. The guardians of the first shrine are beardedfigures, each wearing garments extending to his ankles,a sacred thread, and long hair piled on his head,and holding a flower in the hand nearest the doorwaywhile resting the other hand on his hip; one of themadditionally holds a water vessel. The imagery isstrongly suggestive of Brahma guardians. The single-faced rock-cut image within, however, belies a strictlyBrahma identification in favor of Subrahman

˙ya (Skanda)

as Brahma-Sasta, a form he assumed to put down thepride of Brahma.9 Here Subrahman

˙ya appears to have

taken over Brahma’s door guardians at the same timethat he has appropriated Brahma’s physical appearance.In the central shrine the rear wall is carved with animage of standing Siva, and a linga is inserted into thesocket in its floor. One flanking guardian rests a handon a club, the other holds a tall spear, and both resttheir other hands on their hips. Both wear heavy roundearrings and a crown that holds their jat

˙as in place. The

guardians of the third shrine, carved with an image ofstanding Vis

˙n˙u, resemble those flanking the Siva shrine;

one hand seemingly gestures toward the image withinthe shrine, the other appears to be gesticulating inwonder (vismaya).

The final cave to be considered in this brief discus-sion on dvarapala types is Mahendra’s single-shrined,inscribed Saiva cave at Siyamangalam, known in itsinscriptions as the Avanibhajana-Pallavesvaram.10 Bothdvarapalas are heavyset, bearing weight on one leg withthe other resting lightly on the ground, and both placeone arm on a serpent-wound heavy club carved along-side the door jambs. Both sport the curved heavy jat

˙as

that frame the face, with a crown holding the locksin place, and both wear large rounded earrings. The

guardian to the left, facing frontally, has large hornsthat curve upward toward the top of his crown; hiscompanion figure lacks these. This seems to be astandard variation in paired Saiva door guardians.Lockwood’s suggestion that both dvarapalas may beviewed as personifications of Siva’s weapons seemsplausible;11 the ‘‘hornlike protrusions’’ stand for curvedprongs that represent the trident, and the frontal view ofthe curved edge of the axe that adorns a few of theheaddresses represents Siva’s battle-axe.

In Pallava rock-cut architecture, then, the dvarapalasflanking a shrine enable one to make largely justifiedidentifications of the deity enshrined within.

Pallava Interventions, 1: The Saivite Curse

The small but richly carved Pallava-period Adi-VarahaCave, dedicated to the glory of Vis

˙n˙u, and enshrining an

image of Varaha holding up the goddess Earth, survivestoday in its Vijayanagara-period (ca. 1336 c.–1565)enlargement through the addition of a pillared androofed man

˙d˙apa hall erected in front. It retains intact

its original seventh-century carvings, including elegantlysculpted images of Durga and Laks

˙mı, Vis

˙n˙u, Harihara,

Siva Gangadhara, and Brahma, as well as royal por-traits of the Pallava monarchs Simhavis

˙n˙u (r. 560–580)

and Mahendra (r. 580–630), accompanied by theirqueens. The sculpted walls of this Vais

˙n˙ava cave were

never subjected to any type of adaptation or intrusion,but cut into its floor is the so-called Saivite curse. Inlarge, well-executed letters of the Pallava grantha scriptcurrent in the seventh or eighth century, is an inscrip-tion in two lines running from viewer’s right of theshrine to the cave’s proper right side wall.

Dhik tes˙am dhik tes

˙am punarapi dhig dhig

dhigastu dhik tes˙am Yesan na vasati hr

˙daye

kupathagat˙i-vimoks

˙ako rudrah

˙

In translation: ‘‘Cursed be those, cursed be those, onceagain cursed, cursed, cursed, cursed be those in whoseheart does not dwell Rudra [Siva], deliverer from walk-ing the evil path.’’ Who might have added this vitriolicverse, so out of keeping with the tenor of the originalVais

˙n˙ava dedication?

We postulate that its author and instigator mighthave been the Pallava monarch Narasimhavarman II,called Rajasimha (r. 700–728). Rajasimha is known tohave built three structural temples to Siva, all containinghis dedicatory inscriptions: the Rajasimhesvara alongthe shore at Mamallapuram (popularly called the ShoreTemple), the Kailasanatha in the inland capital of


Kancipuram, and a second Rajasimhesvara Templeupon a hill at Panamalai. All his inscriptions attest hisardent Saivism. In his foundation inscriptions at bothPanamalai and Kancipuram, Rajasimha compares hisown birth to that of Guha (Subrahman

˙ya or Skanda),

begot by the supreme Siva;12 the Udayendiram copper-plate inscription of the later Pallava ruler Nandivarmanrefer to Rajasimha as a devout worshiper of Mahes-vara;13 and the Velurpalaiyam copperplate inscriptionof Nandivarman III (829–853) proclaims the greatnessof Rajasimha in having built a temple to Siva comparableto Mount Kailasa.14 It should be noted, however, thatRajasimha’s own Reyuru copperplate inscription, consi-derably more ecletic, speaks of him as a staunch followerof Bhagavan (Vis

˙n˙u), Mahesvara, and Subrahman

˙ya.15

The immediate question is why the Vais˙n˙ava custo-

dian of the Adi-Varaha Cave did not obliterate this rec-ord, either during the Pallava period itself or during thelater Vijayanagara period, when Vais

˙n˙avism thrived. An

answer to the Pallava part of this query presumably liesin the power wielded during his reign by Rajasimha,which made it perilous to erase any inscription thathe had commissioned. We know that his death wasfollowed by the collapse of the dynastic line and theestablishment of a new, collateral line of Pallava rulers,who had little interest in Mamallapuram and its monu-ments. The seaside town may have continued to serve asa port, but their center of interest and activity appears tohave shifted entirely to the inland capital of Kancipuram,leaving Mamallapuram to languish in its incompletestate. Even a Saiva curse inscribed into the floor of asplendid Vais

˙n˙ava cave was ignored, and is still extant

as evidence of a powerful monarch’s strong espousal ofSaivism.16 The original Vais

˙n˙ava dedication consisted of

three inscriptions on the walls of the cave above thesculptures. Two are labels identifying the Pallava royalportraits; the third is a verse listing the ten avatars ofVis

˙n˙u: Matsya, Kurma, Varaha, Narasimha, Vamana,

Rama (Parasurama), Rama, Rama (Balarama), Buddha,and Kalki. The Saiva curse, by contrast, is engraved intothe floor of the cave, where it may not have beenviewed, either by the monarch who ordered its additionor, indeed, by the cave’s Vais

˙n˙ava custodians, as an

unacceptably blatant intervention. If indeed it wasRajasimha who had the curse carved, his admirationand respect for the sculptures, coupled with the fear ofdivine or temporal retribution, may have been to himsufficient reason to leave them intact; perhaps theaddition of an image of Siva Gangadhara and one ofHarihara (conjoint Vis

˙n˙u-Siva) facilitated his restraint.

We can find help in resolving the authorship ofthis Saiva curse from its repetition within the text of a

lengthy inscription in two other monuments. The firstis the Gan

˙esa ratha, a completed monolith that does

not contain a rock-cut image within its shrine, butcarries an inscription which describes it as the Siva shrineof a ruler titled Atyantakama Pallava (AtyantakamaPallavesvara gr

˙ham), an honorific of Rajasimha. The

Ganesa ratha was not altered or tampered with in anyway during the later Vais

˙n˙ava phase. Its inscription,

engraved on the wall of its veranda to the right of theshrine doorway, concludes with verses numbered eightto eleven, reading thus:

Just as in a large lake filled with water which is fitfor bathing, and covered with various lotus flowers,handsome Sam

˙kara (Siva) abides on the large head,

sprinkled with the water of coronation and coveredwith bright jewels of the illustrious Atyantakamawho deprives his enemies of their pride, who is areceptacle of wealth, who possesses the charm ofcupid, and who assiduously worships Hara (Siva).

He, desiring to attain the glory of Sam˙kara (Siva),

caused to be made this lofty dwelling of Dhurjati(Siva) in order to procure the fulfillment of hissubjects’ desires.

Six times cursed be those in whose hearts does notdwell Rudra (Siva), the deliverer from walking onthe evil path.17

This last eleventh verse is the exact Saiva curse (in avarying English translation) encountered in the Adi-Varaha Cave; here the curse is followed by the name ofthe shrine. An earlier verse (number five) in the Gan

˙esa

ratha speaks of King Atyantakama, famed by the nameRan

˙ajaya, who commissioned this lofty house of Sambhu

(Siva). Ran˙ajaya, a confirmed title of Rajasimha, occurs

in his foundation inscriptions in both the Shore Templeat Mamallapuram and the Kailasanatha Temple at theinland capital of Kancipuram. These two temples con-tain an identical verse that reads:

May Rajasimha, Ran˙ajaya, Srıbhara,

Citrakarmuka, Ekavıra, Sivacud˙aman

˙i, for a long

time protect this earth.18

Both foundation inscriptions also contain his title ofAtyantakama. The evidence thus strongly associates theSaivite curse with Rajasimha.

The second instance of the occurrence of theSaivite curse within a lengthy inscription occurs in theDharmaraja man

˙d˙apa. This modest-sized excavated

cave has two rows of simple columns, square above


and below and octagonal between, that are reminiscentof Mahendra’s caves. Apparently the central shrinewas dedicated to Siva, since it was once flanked bySaiva-style door guardians, as attested by the inscrip-tion. This shrine extends into the cave and is flankedby two recessed side shrines, all three of them sharing asingle base molding. Each shrine has a semicircular stepat the base, with further steps leading up to the floorlevel of the shrines, which are all empty. An eleven-verseinscription, identical to that engraved in the Gan

˙esa

ratha, is encountered immediately upon entering thecave, along its left end wall. The only minor variationis that the name of the cave, Atyantakama Pallavesvaragr˙ham, here precedes the Saivite curse rather than fol-

lowing it as in the Gan˙esa ratha. The inscription rein-

forces the connection of Rajasimha with the Saivitecurse. Significantly, however, whereas no recutting isapparent in the Gan

˙esa ratha, with its Rajasimha in-

scription incorporating the Saivite curse, the identicalinscription in the Dharmaraja Man

˙d˙apa seems to have

aroused the ire of later Vais˙n˙avas, as we shall see.

Pallava Interventions, 2: Alteration of theMahis

˙a-mardinı Cave

A more dramatic Pallava adaptation may be foundin the altered affiliation of the uninscribed Mahis

˙a-

mardinı Cave. Here a reclining Vis˙n˙u occupies one side

Fig. 1. Mahis˙a-mardinı Cave. Mamalla-

puram. Dvarapala. To left of left shrine.Ca. 7th c. Note curved prongs of trident.


wall of the cave, and Durga’s battle with the buffalo-demon Mahis

˙a is carved on the opposite wall. Pride of

place is given to its central shrine, which is fronted bya raised porch with pillars carrying seated lions. Therecarving of its dvarapalas is unobtrusive and escapescasual detection. The two flanking shrines are merelycut into the cave’s rear wall.

As it stands today this central shrine has on its rearwall an exceptionally large bas-relief of Somaskanda,extending all the way to the floor, with a socket cut intothe center of the floor to hold a linga. The dvarapalasflanking this central shrine, however, are not typical

Saivite guardians. They appear to have originally beenVais

˙n˙ava guardians, subsequently altered to Saivite ones

through the addition of clubs.19 In fact, the doorjambshave been carved away to accommodate such recutting,so that the normal jamb, measuring some six inches, isreduced in places to a mere inch. The dvarapalas of theflanking shrine to the left are a pair of typical Saivadvarapalas; the left-hand dvarapala has trident prongsalongside his head as well as a serpent-wrapped club(Fig. 1), the one to the right sports the curved edgeof the battle-axe in his headgear (Fig. 2). Flanking theshrine to the right are typical Brahma-type guardians.

Fig. 2. Mahis˙a-mardinı Cave. Mamalla-

puram. Dvarapala. To right of left shrine.Note curved edge of battle-axe in head-gear. Ca. 7th c.


This occurrence of a pair of typical Saiva guardiansflanking the shrine to the left, together with the alteredVais

˙n˙ava guardians of the main shrine, makes it neces-

sary to reconsider the original intention of the shrine. Itseems likely that the cave, with its reclining Vis

˙n˙u on

one wall and Durga fighting the buffalo-demon on theother, was originally intended to venerate Vis

˙n˙u in its

central shrine, with a Siva shrine to the left and aBrahma shrine to the right. The cave’s dedication seemsto have been altered to the worship of Siva, probablysoon after it was created, by cutting a Somaskandapanel into the wall of the central shrine, adding asocket for a linga, and converting the central Vais

˙n˙ava

dvarapalas into Saiva guardians. Once again, we sug-gest the possibility of Rajasimha being the author ofthis alteration, particularly in view of Rajasimha’s well-

known predilection for placing both a Somaskandapanel and a linga within a single shrine, as evidenced,for instance, at the double shrines of the Shore Temple,which were indubitably his commission.

As we can see, relations among worshipers of thevarious Hindu gods were not always harmonious, evenduring Pallava times. With the exception of the proba-ble alterations to the shrine of the Mahis

˙a-mardinı

Cave, the religious tension that existed during the Pal-lava period is evident at Mamallapuram primarily ininscriptions and is thus relatively unobtrusive. It wouldbe apparent only to those who could read the script,whether in ancient times or today. These interventionshave left behind intriguing puzzles of intent and pur-pose, however, and we hope that scholars and interestedviewers will explore them more fully.

Fig. 3. Ramanuja Cave. Mamallapuram. Ca. 7th c.


Vijayanagara-Period Erasures and Adaptations,1: The Ramanuja Cave

Earlier we emphasized the varied affiliations of rock-cutstructures across Pallava territory and at Mamalla-puram itself during the heyday of the early Pallavarulers, and spoke of the inclusion of imagery relating toDurga in Vis

˙n˙u caves. We now turn to structures that

suggest a period of greater tension between followers ofSiva and those of Vis

˙n˙u, or at any rate between the

sponsors of some structures dedicated to these deities.The most striking instance of such discord is providedby the Ramanuja Cave, a modest structure cut into themain granite hillside at Mamallapuram, not far fromthe Mahis

˙a-mardinı Cave. The structural columns and

lintels seen before the cave belong to the Vijayanagaraperiod (Fig. 3), when similar additions were placedin front of other structures, including the Adi-Varaha

Cave and the Kr˙s˙na sculpted tableau that once stood in

the open air like the adjoining Great Penance cliff.The verandah of the Ramanuja Cave has two columns

rising from the heads of seated lions who face outward,along with two pilasters rising from lions who face eachother across the width of the verandah. The lions arefinely carved, with fangs cut entirely in the round, suchthat a visitor may encircle each fang with a finger. Theoverhanging stone eaves are adorned with Pallava-stylearches that end in a spade-shaped top (kud

˙u), and

the recessed area above carries a row of barrel-vaultedshrines (sala). Carved onto the lower side of the eave isa series of curved stone beams with a row of dwarfishgan

˙a figures below. Flanking the cave on either side,

beyond the eave, is a finely detailed relief model of thetypical aedicular kut

˙a shrine that adorns the very top

of a south Indian temple or which, in miniature, adornsthe corners of its various storeys. Each aedicular shrineextends to cover the space roughly from ceiling to floorlevel of the cave facade; its interior is deeply cut andcould accommodate a small portable image. Betweenthese shrines and the attached lion columns is the firstsign of later intervention: the two dvarapalas alongthe facade have been systematically chiselled away.Their discolored stone outlines suggest that they wereoriginally Saiva-style guardians.20

Inside the cave one discovers shocking structuralalterations and the drastic removal of relief sculptures.Once the cave had three shrines cut into the rear wall,each with a shallow rectangular niche cut into its ownrear wall. Most likely the central shrine was dedicatedto Siva, with Brahma and Vis

˙n˙u in the two side shrines.

In a niche on the rear wall of the original central shrinewas once a Somaskanda panel, which has been sys-tematically removed. Discolorations, however, clearlyreveal a front-facing image of Siva, with Parvatı seatedin profile, and a flying celestial at each upper corner ofthe panel; the image of Skanda is no longer recognizable(Fig. 4). Although Somaskanda panels were popularwithin shrines of the Pallava period, it is noteworthythat this panel differs from that in the Mahis

˙a-mardinı

Cave, the Atiran˙acan

˙d˙a Cave, or the shrines of the Shore

Temple. In the Ramanuja Cave the positioning of thefigures and their more slender outlines are reminiscentof the Somaskanda panel in the upper shrine of theDharmaraja ratha, which appears to predate Rajasimha.The walls enclosing all three shrines of the RamanujaCave, together with the door guardians that must haveflanked their entrances, have been drastically cut awayin an attempt to create a single long rectangular space,apparently considered more appropriate for its altered

Fig. 4. Ramanuja Cave. Mamallapuram. Erased Somaskandawithin rectangular panel on rear wall of what was once thecentral shrine. Ca. 7th c., with 14th/15th–c. alterations.


use (Fig. 5). The space alignment suggests a possibleintention to accommodate a reclining image of Vis

˙n˙u,

presumably of stucco or wood; if such a plan wascarried out, it has left no visible traces.

This thorough and rather brutal alteration appearsto have occurred in the fourteenth or fifteenth century,when Vais

˙n˙avism made its mark at the site. Those re-

sponsible for the erasure and adaptation added a large,lightly inscribed discus (cakra) and conch shell (sankha)along the rock face that forms the left and right exten-sions of the cave facade, beyond the kut

˙a shrine models,

thereby alerting visitors to the cave’s altered Vais˙n˙ava

dedication. It is noteworthy that fifteenth-century com-mentaries on the hymns of the alvars, especially on thefamous Tiruvaymoli of Nammalvar, indicate an increas-ing distrust and suspicion of worshippers of Siva. Inmore than one context these late commentators suggest

Fig. 5. Ramanuja Cave. Mamallapuram. Single elongated space created against back of cave by removing walls that separated theoriginal three shrines. Ca. 7th c., with 14th/15th–c. alterations.

Fig. 6. Ramanuja Cave. Mamallapuram. Erased Durga panel.Ca. 7th c.


Fig. 7. Adi Varaha Cave. Mamallapuram. Durga panel. Ca. 7th c.

Fig. 8. Ramanuja Cave, Mamallapuram.Erased panel on right wall. Ca. 7th c., with14th/15th–c. alterations.


Fig. 9. Ramanuja Cave. Mamallapuram.Saivite curse engraved into floor atentrance. Ca. 7th c.

Fig. 10. Dharmaraja Man˙d˙apa. Mamallapuram. Ca. 7th c.


that opponents of Vais˙n˙avism are to be throttled. The

Tamil phrase used is nirgun˙an˙enpar mid

˙air˙r˙aipid

˙it˙t˙ar

pole, meaning ‘‘like throttling [to death] those whohold that the supreme is nirgun

˙a [without gun

˙as],’’

probably referring to Saiva Advaitin ideals.21

Additionally, the large relief panels adorning thetwo side walls of the Ramanuja Cave in its originalconfiguration have been totally chiselled away. Theobliteration of the imagery on the left wall is notcomplete, and sufficient discoloration remains to enableits clear identification as a standing image of Durga,flanked by female attendants, with kneeling devotees ather feet (Fig. 6). The panel closely parallels the imageryseen in the Adi-Varaha Cave (Fig. 7). Durga stands inthe same elegant stance, with the heel of one leg raisedupon what can only be the obliterated head of thebuffalo. The two flanking attendants are poised just

as they are in the Adi-Varaha Cave; the figure on theleft gracefully crosses one leg behind the other, whilethe attendant at the right is poised to hold a tall bow.The erasure of the imagery in the Ramanuja Cave didnot completely extend to the lowest areas so that theraised heel of the female attendant at the left stillremains relatively intact. Similarly, two flying celestialshover at each upper corner, and the largely obliteratedfigures below suggest kneeling devotees similar tothose seen in the Adi-Varaha Cave. The panel alsodisplays similar pilasters at each end of the wall toframe the niche, and kud

˙u arches above featuring male

heads.The panel on the opposite right wall, framed in

identical fashion, presents problems in deciphering itsimagery (Fig. 8). The outline of the centrally placed,chiselled-away portion suggests an image seated upon a

Fig. 11. Dharmaraja Man˙d˙apa. Mamallapuram. Erased dvarapalas flanking shrine. Ca. 7th c., with 14th/15th–c. alterations.


lotus-like throne; the exceedingly broad outline of thechiselled-away head and the surviving discolorationforce one to confront the possibility of a triple head.Remnants of the standing figure to the extreme leftof the panel indicate a standing male with tiger feet. Isit possible that this wall featured an image of seatedSadasiva with three faces, with the standing sageVyagrapada (‘‘Tiger-footed One’’) to one side? Clearlyidentifiable are erased images of flying celestials flankingthe main image, and also intact portions of cloud-likesemicircles.

The Ramanuja Cave poses one further perplexingquestion. It also contains the ‘‘Saivite curse’’ cut intothe floor of the cave in large, well-executed Pallavacharacters between the two central facade columns, sothat it is the first thing one sees when entering the shrine

(Fig. 9). ‘‘Cursed is he, cursed is he, once again cursed,cursed, cursed, cursed is he in whose heart does notdwell Siva, deliverer from the evil path.’’ We mayassume that this verse was inscribed into the floorof the cave during its original Saiva phase, when aSomaskanda panel and linga graced the central shrine,with Durga on the left wall and a second relief on itsright wall that may have featured Sadasiva. If such anassumption is valid, then the original cave was probablya Rajasimha commission. The question remains as towhy those who so drastically interfered with the cave’soriginal Saiva program did not erase this ‘‘six-timescursed’’ Saiva inscription, engraved in beautiful Pallavagrantha letters of the early eighth century. Perhaps thePallava script was no longer familiar to fifteenth-centuryviewers or readers. Being accustomed to the Tamil

Fig. 12. Koneri Cave. Mamallapuram. Ca. 7th c.


script, which was well established by this date, theyignored the ancient, now illegible curse and left ituntouched.

Vijayanagara Erasures and Adaptations,2: Dharmaraja Man

˙d˙apa and Koneri Cave

The small, simple rock-cut cave known as DharmarajaMan

˙d˙apa, with blocky Mahendra-style columns, ex-

amined in connection with the Saivite curse, is a triple-shrined structure (Fig. 10). We have seen that thecentral shrine was apparently intended to house a linga;it extends farther into the cave than the two flanking

shrines, all three of which share a single base molding.The doorway of the central shrine was originally flankedby Saiva-style guardians; these have been systematicallyremoved, leaving only the stone silhouettes to attesttheir one-time existence (Fig. 11). We postulate thatsuch removal occurred during the Vijayanagara phaseof the site’s occupation and alteration; certainly thelarge conch shell and discus incised into the two centralcolumns of the facade must have been added at thistime. Why the planned alteration was never completedposes yet another conundrum.

One final example of the removal of relief carvingscomes from the five-shrined Koneri Cave on the oppo-

Fig. 13. Koneri Cave. Mamallapuram.Dvarapala of shrine 3. Ca. 7th c.


site side of the massive central rocky scarp at Mamalla-puram (Fig. 12). Along its facade are a set of foursimple, slender columns, square above and below andoctagonal between, with the two end columns attachedto the cave walls; farther inside the cave are four fullyfluted slender columns, which taper slightly toward theirbulbous capitals and are adorned with two bands offloral designs. Against the rear wall of the cave fiveshrines share the same raised base molding; the centralshrine and the two end shrines project slightly into thecave and have steps leading up to them, whereas thetwo between are less recessed and without steps. Cutinto the center of the rear wall of each shrine is a plainrectangular niche that may have been intended fora painted image of the deity. It has been suggestedthat the Koneri Cave may have been designed to housethe five aspects of Siva as Isana, Tatpurus

˙a, Aghora,

Vamadeva, and Sadyojata.22 The dvarapalas of fourof the five shrines are male, with fangs, bulging eyes,and fiercely curved, joined eyebrows, all suggestive of aSaiva affiliation (Fig. 13). In addition, in shrines twoand three, one door guardian has the curved prongs ofSiva’s trident alongside his headdress (Fig. 14). Reinforc-ing the probability of the cave’s dedication to the five-featured aspect of Siva is the fact that the northernmostshrine, the first encountered by the visitor, has femaledoor guardians who would have flanked either Durgaherself or Vamadeva, the female aspect of the five-featured Siva.

This northernmost shrine, however, has not re-mained intact. The female door guardian to the lefthas been almost entirely cut away, and her companionon the right has had substantial portions of her bodyremoved (Fig. 15). Apparently, this removal too datesfrom the fourteenth-century phase when the Vais

˙n˙ava

emblems of the sankha and cakra were incised on thetwo flanking projections of the excavation in front ofthe facade. Once again, however, apparent plans for aVais

˙n˙ava adaptation of the cave were never fully put

into execution. This leaves a series of questions forfurther exploration. Is it possible to reconstruct morefully the history of these Vijayanagara-period Vais

˙n˙ava

interventions at Mamallapuram? Who initiated them?Why were some of them carried out fully and othersleft incomplete?

Concluding Remarks

This brief essay raises a variety of unresolved ques-tions concerning the addition, erasure, adaptation,and intended reuse of certain rock-cut monuments atMamallapuram in the context of exclusive and un-wavering devotion to either Vis

˙n˙u or Siva. The devo-

tional alvar and nayanmar poet-saints of the Pallavaperiod, for instance, sang of Vis

˙n˙u and Siva as equal

in greatness and glory, and reserved their ire for theirBuddhist and Jain competitors, but rivalry betweenVais

˙n˙avites and Saivites has also raised its head at

Fig. 14. Koneri Cave. Mamallapuram.Dvarapala of shrine 2. Ca. 7th c.


various times. The Saivite curse of the Pallava era testifiesto an ardent espousal of the worship of Siva to the exclu-sion of all else, at least among some Pallava patrons.

The more drastic erasure and alteration of shrinewalls and sculptures during the fourteenth or fifteenthcentury testify to later Vais

˙n˙ava interventions directed

at Saiva shrines. We can see the heated competitionbetween followers of Siva and of Vis

˙n˙u during this later

period reflected in new mythological and iconographicinnovations in South India. Concerned about the increas-ing esteem and recognition achieved by Vais

˙n˙avism, and

specifically the expanding popularity of the Narasimhaman-lion avatar of Vis

˙n˙u, Saivas created a new com-

posite Saiva figure known as Sarabesa, an eaglelike crea-ture who tramples upon Narasimha.23 The Vais

˙n˙ava

response was to create an even fiercer As˙t˙a-mukha

Gan˙d˙a-Bherun

˙d˙a Narasimha, a mythical bird with the

heads of eight different animals of which the prime onewas leonine, with fierce long talons on its eight legs,who, in turn, tore into Sarabesa.24

The conclusions here must remain tentative, and wehope to inspire further exploration of these issues. Wehave attempted here to identify later interventions intothe monuments of Mamallapuram, and to place theseacts into a historical context of sporadic religious con-flict between Saivas and Vais

˙n˙avas in southern India.

Fig. 15. Koneri Cave. Mamallapuram.Erased dvarapalalika of shrine 1.Ca. 7th c., with 14th/15th–c.alterations.


Material evidence, such as alterations, erasures, and newiconography, can be just as powerful as any literary dia-tribe in revealing religious competition, challenge, andenmity.

Notes

* Throughout this essay, ‘‘left’’ and ‘‘right’’ refer toviewer’s left and right. All photographs in this essay byVidya Dehejia.

1. William Willetts, An Illustrated Annotated AnnualBibliography of Mahabalipuram on the CoromandelCoast of India, 1582–1962 (Kuala Lumpur: University ofMalaya, 1966); N. S. Ramaswamy, Mamallapuram, anAnnotated Bibliography (Madras: New Era Press, 1980).

2. Michael Lockwood, Pallava Art (Tambaram:Tambaram Research Associates, 2001). This volume isa revised and enlarged edition of two earlier works:Mahabalipuram Studies (Madras: The Christian Litera-ture Society, 1974) and Mamallapuram and the Pallavas(Madras: The Christian Literature Society, 1982).

3. For an inscription that does relate an act of royaltheft, see Richard H. Davis, ‘‘Trophies of War: The Caseof the Calukya Intruder,’’ in Perceptions of India’s VisualPast, eds. C. B. Asher and T. Metcalf (New Delhi:American Institute of Indian Studies, 1994), pp. 161–77.

4. For a valuable recent collection of essays, see DavidGilmartin and Bruce B. Lawrence, Beyond Turk andHindu: Rethinking Religious Identities in South Asia(Gainesville: University Press of Florida, 2000).

5. Indira V. Peterson, ‘‘Sraman˙as against the Tamil

Way: Jains as Others in Tamil Saiva Literature,’’ in OpenBoundaries, ed. John Cort (Albany: State University ofNew York Press, 1998), pp. 163–85.

6. Padma Kaimal, ‘‘Playful Ambiguity and PoliticalAuthority in the Large Relief at Mamallapuram,’’ ArsOrientalis, vol. 24 (1994), pp. 1–24.

7. We see this at many other sites too. One instance isthe Chandella capital of Khajuraho (present-day MadhyaPradesh), where Yasovarman built a temple to Vis

˙n˙u,

his son Dhanga dedicated one to Siva in the form of anemerald linga, and Dhanga’s friend built a temple to theJina Parsvanatha.

8. Vidya Dehejia, The Sensuous and the Sacred (NewYork: American Federation of the Arts, 2002), p. 134.

9. K. R. Srinivasan, Cave-Temples of the Pallavas(New Delhi: Archaeological Survey of India, 1964), p. 159.

10. Ibid., p. 91.11. Michael Lockwood, Pallava Art, p. 7ff.12. T. V. Mahalingam, Inscriptions of the Pallavas

(New Delhi: Agam Prakashan, 1988), pp. 173–75; pp.200–202.

13. Ibid., pp. 227–39.14. Ibid., pp. 372–79.15. Ibid., pp. 169–72.16. Dr. Nagaswamy would see this inscription as an

affirmation that Vis˙n˙u and Siva are one and the same,

and I am willing to agree with him that textual interpre-tation can be found to support such a philosophic theory.I do believe, however, that the engraving of the Saivacurse does not affirm that Vis

˙n˙u and Siva are equal and

identical.17. R. Nagaswamy, ‘‘New Light on Mamallapuram,’’

Transactions of the Archaeological Society of South India,Silver Jubilee Volume (1962), p. 48. Translation mar-ginally altered from original.

18. See also Michael Lockwood, Pallava Art, p. 223.19. Ibid., p. 12.20. Remnants of a set of frontal feet make one wonder

if there was yet another change in affiliation. Is it possiblethat there was an original Vais

˙n˙ava cave, changed to

Saiva, and changed back to Vais˙n˙ava?

21. I am grateful to Dr. R. Nagaswamy for alerting meto these references. See commentary ‘‘Idu’’ or BhagavadVishayam on the first two Tiruvaymolis of Nammalvar byAdiaya valainjan jiyar, commentary 6000, 12000, 24000,and 36000, published by Krishnaswami Aiyangar (Sudar-sanam) (Puttur Agraharam, Thiruchi, 1995), p. 227. Thesame expression is repeated on p. 215. Dr. Nagaswamypoints out (personal correspondence) that it is the com-mentators, not the alvars, who are responsible for suchstatements and misinterpretations of original statements.In the very last verse of his Tiruvaymoli, Nammalvar statesthat he achieved liberation by worshipping Hari, Hara,and Ayan (Vis

˙n˙u, Siva, and Brahma); commentators twist

this to suggest that Nammalvar meant to say Hari, towhom Hara and Ayan are slaves.

22. K. R. Srinivasan, Cave-Temples of the Pallavas,p. 141.

23. The very first images of Sarabesa appear on thelate Chola temples such as Darasuram and Tribhuvanam.

24. I thank Raju Kalidoss for drawing my attention tothis material.


Documents

Addition, Erasure, and Adaptation: Interventions in the ...vidyadehejia.com/wp...Davis-Addition-Erasure-2010.pdfAddition, Erasure, and Adaptation: Interventions in the Rock-Cut Monuments