Archaeology

An initial investigation of the significance of the horns of dvarapalakas (door keepers) in early Pallava temples has led us to the surprising hypothesis that in the Mahishasuramardini Cave at Mahabalipuram, the main sanctum was originally planned for Vishnu not for the Somaskanda panel which we see today.

Let us consider the horns on dvarapalakas, There are several conflicting views among scholars as to the significance of these horns. Mr.P.R. Srinivasan says that they are a kind of mutation of the early Buddhist motif of Nagaraja as dvarapalaka, that is, the multi-headed snake hood of Nagaraja develops into two horns¹ If we remove the middle three hoods out of the five snake hoods from these Nagarajas they will, look exactly like the horned dvarapalakas of later times.¹ Others have explained the horns with reference to the practice of wearing horns by such primitive tribes as the Nagas and the Gonds. One further explanation, which is supported by Mr.K.R.Srinivasan is that the horned dvarapalaka represents a humanised form of the bull, Nandi.²

While photographing a dvarapalaka in the upper cave at Vallam (two miles east of Chingleput town), we were struck by the similarity between the horns of this dvarapalaka and the outer prongs of the trident or trisula as represented in Pallava sculpture elsewhere.³ These horns and the outer prongs of the trisula have the same peculiar compound curve at their base (see fig.1). Furthermore, the so-called horns in the Vallam example are not shown attached to the head or head-dress in a very realistic manner - -the right  horn is displaced quite unnaturally away from the head. We concluded,  therefore,  that the horns of the dvarapalaka along with his elongated makuta (as the central prong) did, in fact, represent the trisula, a Saivite emblem.

At Vallam, only the dvarapalaka to the proper right of the entrance has horns. However, we soon discovered that,  although the dvarapalaka to the left⁴ did not have horns, he did have an axe-blade projecting edge-forward from the front of his head-dress (see fig, 2), The axe is another Saivite emblem. The trisula "horns" and the axe-blade, then, must be clearly recognized as Saivite symbols which, along with certain other characteristics such as the snake-entwined club go to indicate quite unambiguously that these dvarapalakas are guarding a shrine for Siva.

Other examples of dvarapalakas with "horns" and axe-blades on their head-dress are to be found in the Kailasanatha temple at Kanchipuram, at the Atiranachanda cave-temple at Saluvankuppam, and at various shrines at Mahabalipuram. In this later Pallava treatment we find the head of a triple-bladed axe represented on the head-dress (see fig.3). In most of these cases, a knowledge of the significance of the trisula horns and the axe-blade is not necessary for an identification of the shrines as Saivite because within the shrines there is a lingam.⁵ However, consider the shrine on the western side of the second level of the Dharmaraja Ratha at Mahabalipuram. This sanctum is empty and unfinished,  and there is nothing inside it now that would indicate which god it was fashioned for. Therefore it is the horned guardian to the proper right of this shrine which reveals it was intended as Saivite. In the five celled Koneri Mandapa at Mahabalipuram, each of the three middle shrines has a horned dvarapalaka to the left. And to the right of each of these three shrines we find a dvarapalaka with an axe-blade represented on the head-dress. In one case, the axe-blade is shown edge forward at the top of the makuta In the other two cases, we find a single axe-blade in profile on the forehead band of the makuta. We conclude that at least these three out of the five cells were meant to be Saivite.

The trisula, as such, is represented on the makuta of the left dvarapalaka at the entrance of the Iswara temple (the old light-house), Mahabalipuram.

Another type of horned head though not of dvarapalakas -- must be mentioned. We refer to the type of head decorating the upper reaches of the Ganesh Ratha. The horns in this case are quite realistic, but the fact that they represent the Saivite trisula must now be evident.

The practice of showing the emblems of the deity on his guardians' head-dress is applied by the Pallavas to Vaishnavite shrines as well as Saivite. A clear example of this is found in the Varaha Cave at Mahabalipuram: the dvarapalaka immediately to the right (proper) of the sanctum's entrance has a discus represented edge-forward at the very top of his head-dress (see fig.4). The dvarapalaka to the left has a conch  placed at the top of his head-dress (see fig. 5). The discus and conch are Vishnu's emblems. That this Varaha Cave is a Vaishnavite temple is undisputed, and we find there the  Varaha, Trivikrama, and Gajalakshmi panels which are all Vaishnavite themes. But the discus and  conch emblems on the head-dress of the dvarapalakas give additional confirmation that the (now empty) sanctum was for Vishnu.

Another important example of Vaishnavite emblems on the head-dress of dvarapalakas is to be found in the Adivaraha Cave-temple at Mahabalipuram. Here the Varaha figure in the central shrine is under worship. The modern walls which enclose the front of this shrine hide parts of the dvarapalakas. However, one is still able to see the discus at the top of the head-dress of the right dvarapalaka and the conch similarly placed on the left dvarapalaka.

The discus appears at Mahabalipuram on the head-dress of several other figures who are not guarding doors. We may mention as an example the figure of Garuda (supporting Vishnu) found on the second level of the Dharmaraja Ratha, north side. In this example, the discus is shown edge-forward at the base of Garuda's makuta. A similar panel, but rather worn or unfinished, is represented on the north side of the Arjuna Ratha. A very clear treatment of the discus in this case at the top of the head-dress can be seen on the figure to the right,  just below the sleeping Vishnu in the Mahishasuramardini Cave.

In the case of the goddess, Durga, the female guards (dvarapalikas) who guard her shrines in Mahabalipuram are shown with a sword in hand (right guard) and with a bow (left guard). There are two Durga shrines at Mahabalipuram: the Draupadi Ratha and the Kotikal Mandapa. The two young fighting women accompanying the goddess in the Durga panel of the Adivaraha Cave are similarly armed and provide an analogous example, although, strictly speaking, they are not guarding a door here.

Our main conclusion so far, then, is that dvarapalakas are often shown with emblems or weapons which are characteristic of the deity they guard. It is tempting even to consider them as anthropomorphized emblems or weapons. In any case, if a sanctum itself is empty, these emblems or weapons of the dvarapalakas may be decisive in identifying the deity of that sanctum. We have found that in the cases of many Saivite shrines, one dvarapalaka has horns and the other an axe-head shown on the head-dress, and both may have clubs with snakes encircling. In the case of Vaishnavite shrines, we sometimes find the following arrangement: one dvarapalaka has a discus represented on his head-dress, and the other, a conch.

With these facts in mind, let us turn to the famous Mahishasuramardini Cave at Mahabalipuram. There are three sanctums in this cave-temple, the central one being given special prominence by having before it a raised porch with two lion pillars in front. But let us first consider the right (southern) sanctum. The dvarapalaka to the proper right of this sanctum has horns. The dvarapalaka to the left has a single axe-blade projecting edge-forward above his forehead. The right dvarapalaka has a club with snake around it. We conclude from these facts that the right sanctum of the Mahishasuramardini Cave is clearly Saivite.

Let us next consider the left (northern) sanctum. The two dvarapalakas are shown without any of the above Saivite emblems. And, furthermore, they both wear the long dress and the uttariya which are uncharacteristic of dvarapalakas of Saivite shrines.⁶ We conclude that the left sanctum of the Mahishasuramardini Cave is distinctly non-Saivite.

With a clearly Saivite sanctum to the right, with a distinctly non-Saivite sanctum (possibly for Brahma) to the left, and furthermore with a reclining Vishnu panel on the right wall, one would naturally expect the main central sanctum to be for Vishnu. But surprisingly, we find instead a large Somaskanda panel on the back wall of this main sanctum.

Now this surprise has led us to note, first, the obvious fact that this Somaskanda panel is different stylistically from the other two panels (of Vishnu and Mahishasuramardini) in this temple; and, secondly, that there is a striking similarity between this Somaskanda panel and similar panels found in the Kailasanatha temple, Kanchipuram. Is, therefore, the Somaskanda panel in this cave temple, a later addition, transforming what may have been originally planned as a Vaishnavite main shrine into a Saivite shrine?

To help us arrive at an answer to this question, let us consider the dvarapalakas guarding this central shrine. At first glance, both dvarapalakas seem to be Saivite: they both have clubs -- the club of the proper right dvarapalaka being encircled by a three-headed snake. The dvarapalaka to the right has horns (in light relief), and the dvarapalaka to the left has  a triple-bladed axe-head represented on the head-dress above his forehead.

But there are several puzzling aspects about the way in which these two dvarapalakas have been sculpted. In fact, it looks  as though these niches have been originally intended for dvarapalakas without clubs -- the kind of dvarapalakas one would expect  to be guarding a shrine for Vishnu. The reason we say this is that the clubs seem like an afterthought. The clubs are carved where the pilasters should be (see figs. 6 & 7), and completely break the orderly boundary of the rectangular niches (these boundaries are clearly established above the heads of the dvarapalakas). It would be interesting to know whether there is any other example in Pallava sculpture of such an extreme disregard of the rectangular boundary of the niche.

One possibility we are suggesting is that work had begun on these niches at a time when the main sanctum was intended for Vishnu. At this time, the boundaries of the niches and the general pose of the dvarapalakas were established. For one reason or another, the work was not completed. At a later date, when Saivism was in the ascendancy, the details of the dvarapalakas were finished as Saivite, including the horns in very shallow relief on one guard and the axe-head on the other's head-dress. The clubs had to be added in a most unusual place, where the pilasters normally would come. To accomplish this addition of the clubs, the rock area for the pilasters and all the rest of the architectural ornamentation of the main shrines facade had to be removed. This refacing of the rock has left only a plain wall for us to see today.⁷

Another possibility we would like to suggest is that the original Vishnu shrine may have been planned to project forward in the area which is now the inner porch. At a later date this shrine was then converted into the present porch and another cell excavated in the rear wall for the Somaskanda panel. Such a haphazard development  would explain why the dvarapalakas of the main, central shrine are considerably smaller than those of the side shrines, and why they have been carved in such a make-shift manner.

There is one further puzzling matter which we must mention at this point. It concerns what remains of an animal sculpted in the solid rock just in front of the Mahishasuramardini Cave and in line with the main sanctum. The upper portion of this animal is it? Mr. R. Nagaswamy identifies it as a Nandi.⁸ A close examination, however, clearly shows the feet of an elephant. As all four feet are squarely on the ground, we conclude that the elephant must have been in a standing posture. And as the toe nails of the elephant are seen on the side of the feet away from the Cave, we conclude that the elephant was facing directly away from the Cave.

The elephant has been associated with several deities. And it is also associated, with Vishnu. Specifically we may mention a panel of the Kailasanatha temple, Kanchipuram, which shows Vishnu seated on an elephant throne. Or, alternately, perhaps this elephant was sculpted as Skanda's vehicle, at the time the Somaskanda panel was done.

But let us now return to the Mahishasuramardini Cave and a careful study of the panels with the object of showing that the Somaskanda panel is a decidedly later work than the other two panels in this cave. As a basis for our argument, we would like to mention at this place certain general observations we have made about the dress and ornaments of Pallava-sculpted figures.⁹ In early Pallava sculpture (around the period of the great Penance Panel and the Rathas usually ascribed to Narasimhavarman I), men do not wear any leg ornaments and are shown with only one diagonal band¹⁰ across the body. In the early period, women do not wear any diagonal band and have only a single anklet on each leg. In the later Pallava sculpture (at the time the Kailasanatha and Shore temples were built), we notice that men now have leg ornaments and often have more than one diagonal band. In this later period, women are now seen wearing the diagonal band; they frequently have more than one leg ornament; the halter (or vertical straps) for the breast-band is introduced; and a hair-style which looks like a thick garland around the base of a tall crown, and slightly pinched in the middle, is seen for the first time.

On the basis of these general observations, we have analyzed the panels of the Mahishasuramardini Cave and have concluded that the Somaskanda panel is of a decidedly later period than the other two panels in this cave.

First, we wish to establish that the Somaskanda panel of this cave temple belongs to the later (Kailasanatha) period.

To begin with, we may mention that, there is an extremely close similarity between the figures of the Somaskanda panel of this cave temple and the figures of similar panels¹¹ in the Kailasanatha temple. In both places we find these earmarks of the later period:
1) Parvati wears a diagonal band;¹²
2) she also wears both the kinkini on her ankles as well as the silambu;
3) she has the peculiar garland-like hair-style with the tall headdress.¹³ The fact that the faces of Siva and Parvati in the Cave temple are so strikingly similar to what we find in Kailasanatha would make us believe that they were created by the same school of sculptors.¹⁴

Next, we wish to point out that the Somaskanda panel of the Mahishasuramardini Cave is quite different stylistically from the early Somaskanda panel of the Dharmaraja Ratha. The following characteristics of the later period all of which are found in the Mahishasuramardini Cave panel, are absent in the Dharmaraja Ratha panel:
1) Parvati has the peculiar garland-like hair style in the Cave panel whereas she does not, in the Ratha panel;
2) Parvati wears a diagonal band in the Cave panel whereas she has none in the Ratha panel;
3) Parvati wears two types of anklets in the Cave panel whereas she has only one type in the Ratha panel;
4) Parvati's pose in the Cave panel is in common with her pose in the numerous Somaskanda panels of the Kailasanatha temple, the Shore temple, and the Atiranachanda temple whereas her profile pose in the Ratha panel is very different from all the others;
5) Siva wears three different diagonal bands in the Cave panel whereas he has only one in the Ratha panel;
6) The discus and conch of Vishnu are depicted with flames (generally accepted as a later characteristic) in the Cave panel whereas there are no flames shown in the Ratha panel.

In passing, we point out that Brahma and Vishnu, in the Cave panel, appear as small figures (relative to Siva) and are placed behind Siva's throne, just above his upper hands (this treatment is in common with the later period Somaskanda panels of the Kailasanatha, Shore, and Atiranachanda temples) whereas, in the Ratha panel, Brahma and Vishnu stand on either side of the main figures and are almost the same size as Siva.

In regard to this last point, we take the more equalitarian treatment of Brahma and Vishnu in the Ratha panel to be an important indication of this panel's early date a period in which the three gods were more uniformly represented in a shrine dedicated to only one of them.¹⁵

We have now compared the Somaskanda panel of the Mahishasuramardini Cave with other Somaskanda panels of the later Pallava period in which the Kailasanatha temple was built and have found much in common.

We have contrasted the Somaskanda panel of the Mahishasuramardini Cave with the earlier one of the Dharmaraja Ratha and have found significant differences.

We conclude, therefore, that the Somaskanda panel of the Mahishasuramardini Cave was done around the period in which the Kailasanatha temple was built.

Next, we wish to establish that the other two panels of the Mahishasuramardini Cave (the Vishnu and Durga panels) were done during an earlier period say around the time the great penance Panel and the Rathas were sculpted.

Considering first the reclining Vishnu panel in this Cave, we find the following early characteristics:
1) no leg ornaments are found on men;
2) no man wears more than one diagonal band;
3) the women have no diagonal bands;
4) there is no halter on the breast-band of the woman seen above Vishnu;
5) the women have only single anklets; and 
6) they do not have the garland-like hair-style with tall makuta.

Then, we contrast the reclining Vishnu panel of this Cave (as an early example) with the much smaller reclining Vishnu panel¹⁶ of the Kailasanatha temple (as a later example). The coating of lime plaster on this latter panel makes any job of detailed comparison difficult. However, we mention the following differences between the two panels; 
1) in the Cave panel, the breast-band is depicted without any halter -- whereas in the Kailasanatha panel, the woman kneeling at Vishnu's feet wears a halter on her breast-band.
2) the five heads of the serpent Ananta in the Cave panel are depicted in a naturalistic manner whereas in the Kailasanatha panel, the five heads of Ananta are curved as horned Yali-heads.

In passing, we note that the Kailasanatha panel has a small figure of Brahma seated on a lotus, just above Vishnu's left hand-whereas there is no Brahma figure to be found in the Cave panel.

We have now listed the early characteristics of the reclining Vishnu panel of the Mahishasuramardini Cave and have contrasted this panel with the reclining Vishnu panel of the Kailasanatha temple. From our analysis, we conclude that the Cave panel was done during a period decidedly earlier than that of Kailasanatha.

Considering next the Durga panel of the Mahishasuramardini Cave, we contrast it (as an early work) with the Saluvankuppam Durga panel and the Kailasanatha Durga panels (as later works).
1) Durga has no garland-like hair-stylewith tall makuta in the Cave panel whereas in the Saluvankuppam and Kailasanatha panels, she does have this characteristic hair-style;
2) Durga has no diagonal band in the Cave panel whereas she has a diagonal band in the Kailasanatha panel;
3) Durga has no halter on her breast-band in the Cave panel whereas she does have a halter in the other panels;
4) Durga has only a single anklet in the Cave panel whereas she has more than one on each leg in the other panels;
5) Durga's lion has no saddle strap in the Cave panel whereas the lion has a saddle with strap in the Saluvankuppam and Kailasanatha panels;
6) the buffalo demon has only one diagonal band in the Cave temple whereas he is depicted with two diagonal bands in the Saluvankuppam panel;
7) the buffalo demon has no leg ornaments in the Cave panel -- whereas he has prominent ones in the Saluvankuppam panel.

We conclude, therefore, that the Durga panel of the Mahishasuramardini Cave was done during a period decidedly earlier than that of Kailasanatha.

To summarize our arguments based on stylistic analysis, then:

First, we have tried to show that the Somaskanda panel of the Mahishasuramardini Cave is a relatively later work by comparing it with similar panels of the Kailasanatha period, and contrasting it with the Somaskanda panel of the Dharmaraja Ratha. Second, we have then tried to show that the other two panels of the Mahishasuramardini Cave are earlier works by listing the early characteristics we find in them, and then contrasting them with similar panels of the Kailasanatha period.

From this analysis of style, our main conclusion is that the Somaskanda panel of the Mahishasuramardini Cave is a decidedly later work than the other two panels of the same cave.

This conclusion (from an analysis of style) strengthens the claim we have made earlier (on the basis of an examination of the Cave's dvarapalakas) that the Mahishasuramardini Cave has had an erratic history of development and that there are reasonable grounds to suppose that the main, central shrine was originally planned for Vishnu.

NOTES
¹ P.R.Srinivasan, " Beginnings of the Traditions of South Indian Temple Architecture", a Bulletin of the Madras Government Museum, New Series-General Section, Vol. VII,No.4, 1959,p.34.
² K.R.Srinivasan, Cave-Temples of the Pallavas, Architectural Survey of Temples, Series No.1, New Delhi: Archaeological Survey of India,1964, p. 36.
³One of the clearer examples of the Pallava trisula is found on the Dharmaraja Ratha, second level, north side, accompanying a figure of Siva.
⁴Throughout this paper our use of 'left' and 'right' will mean the proper left and proper right, respectively.
⁵As the lingam, itself, was probably a later addition to Pallava temples, even its presence is not conclusive proof that any given shrine was originally Saivaite. See K.R.Srinivasan's concise argument in support of the "later addition" thesis in his paper "Some aspects of religion as Revealed by early Monuments and Literature of the South" Madras: University of Madras, 1950,p. 61ff. An elaboration of this position is found in K.V.Soundara Rajan's paper, "Cult in the Pallava Temples", in the Transactions of the Archaeological Society of South India, 1962-65, Madras, 1969.                                           ⁶The right dvarapalaka wears the uttarya over his right shoulder, and the left dvarapalaka wears it over his left shoulder, a fact overlooked by some observers.
⁷Surprisingly, the facades of the other two shrines seem to have been refaced in a similar way. In doing this job of recessing the walls, the feet of the dvarapalakas of the left shrine have been sheared off. In the case of the right dvarapalaka of the right shrine, his right foot remains projecting out beyond the wall's surface in a most unusual manner. While refacing the wall, a portion of the rock was left underneath his foot to give some sort of support to it.

We must mention , in passing, two other puzzling aspects:(1) the dvarapalakas of the main, central sanctum are noticeably smaller than the dvarapalakas of the other two sanctums; (2) the entrances of the two side shrines are in poor alignment with the stairways provided with them.

8 In his book, Mamallai (in Tamil), Madras State Department of Archaeology,1969, p. 90.
⁹Some of these observations have been discussed in "Mahabalipuram: Costumes and Jewellery", by Gift Siromoney, Madas Christian College Magazine,1970, pp.78-83.
¹⁰As there is much confusion in the application to early sculpture of the term "sacred thread", we have deliberately coined the more general term "diagonal band " which we intend to include the thread as well as other similarly worn items.
¹¹Specifically, we refer to the sub shrine centrally located on the northern side of the main shrine of the Kailasanatha temple.
¹²This characteristic is also found in the Somaskanda panels of the Shore temple and the Atiranachanda temple.
¹³ Also in the Shore Temple and the Atiranachanda temple.
¹⁴We must disagree with K.R.Srinivasan (Some aspects of Religion...... p. 51) who says that possibly the Somaskanda panel of the Mahishasuramardini Cave includes a unique representation of Siva as Chandesanugrahamurti.The so-called Chandesa is actually a woman; and the snake which Siva holds by the tail in his upper right hand has been mistaken for a garland. Neither is the Nandi reclining in front of the throne a unique feature in a Somaskanda panel as there is a similar representation (but badly worn) in one of the Somaskanda panels of the Atiranachanda temple.
¹⁵To help substantiate this claim, we may point out the Orukal Cave (Saivite) temple, Tirukalukunram, and the Adivaraha Cave (Vaishnavite ) temple, Mahabalipuram, as early monuments exemplifying the more uniform representation of the three gods within a temple which is chiefly dedicated to only one of them.
¹⁶This panel is found directly above the entrance to the Somaskanda sub-shrine already discussed  (the one centrally located on the northern side of the main shrine...)

 

Fig 1. Horned Dvarapalaka -- Vallam	Fig 2. Axe-Blade -- Vallam
Fig 3. Triple Bladed Axe-head,   Atiranachanda	Fig 4. Discus of Dvarapalaka near  Varaha cave
Fig 5. Conch -- Varaha cave - II	Fig 6. Right Dvarapalaka -- central shrine, Durga cave
Fig 7. Left Dvarapalaka -- central shrine, Durga cave