History of Indian and Eastern Architecture v1.c19.s4: Sikharas.
History of Indian and Eastern Architecture v1.c19.s4
- Language
- English
Sikharas.
v1.c19.s4.1
There is one other peculiarity common to both Hindû and Jaina architecture in the north of India that requires notice, before proceeding to describe particular examples. It is the form of the towers or spires called Sikharas or Vimânas, which invariably surmount the cells in which the images are placed. It is probably correct to assert that the images of the Tîrthankaras are invariably placed in oblong or square cells, and those of Hindû deities in square—generally cubical cells, of no great dimension, and that these cells receive their light from the doorway only. It seems also an invariable rule that the presence and position of the cell should be indicated externally by a tower or spire, and that these towers, though square or nearly so in plan, should have a curvilinear outline in elevation. If the tower at Bodh-Gayâ (*ante*, p. 78) retains unaltered the original form given to it when erected about the 5th or 6th century, this dictum would not apply to Buddhist architecture. As it is, however, the only Buddhist sikhara yet discovered it is hardly fair to draw any decided inference from one single example, while with Jaina or Hindû towers I know of no exception. Take, for instance, the tower represented in the following woodcut (No. 183), which purports to be an elevation of the celebrated Black Pagoda at Kanârak in Orissa, and may be looked upon as a typical example of the style, and of which it may be considered of a fair medium example. The upper part of the tower, to some extent, overhangs its base. It bends inward towards the summit, and is surmounted by what is called an Amalaka—a massive circular coping stone which supports a vase called *amritakalasa* or *amritakaraka*, *i.e.* "dew vessel."¹ Its peculiar corrugated form occurs frequently in old examples as a sort of blocking course dividing the sikharas horizontally into numerous small compartments, and it seems as if what is used there in a straight-lined form was employed as a circular ornament at the summit. It is a very beautiful architectural device, and was, as far as I can see, adopted only because it was so, and contrasted brilliantly with the flat ornaments with which it was employed. At present
v1.c19.s4.2
¹ The "âmalaka" has been popularly supposed to be derived from *âmalaka*—the *Phyllanthus emblica*, *Emblica officinalis* or *Emblica myrobalan*; but, though an article of Hindû materia medica, it is so insignificant a berry that it could hardly be thought of as an architectural model. May it not be from *amala*—"pure," "spotless"? Amalasilâ—"pure stone," is applied to this crowning member.—Beal, 'Buddhist Records,' vol. ii pp. 136-137; Foucher, 'L'Art Gréco-Bouddhique du Gandhâra,' tome i. p. 96*n*.
v1.c19.s4.3
183. Restored Elevation of the Sun-temple at Kanârak. (From a Drawing by the Author.) No Scale.
v1.c19.s4.4
we do not seem to be in a position to explain its origin, or that of a great many other details that are frequently met with in Hindû architecture.
v1.c19.s4.5
Whatever its origin, this amalaka is generally surmounted by a flat dome of reverse curvature, in the centre of which stands the *kalasa*, *karka*, or pinnacle, in the form of a vase, generally of very beautiful and graceful design.
v1.c19.s4.6
The great and at first sight puzzling question is, from what original is this curious combination of forms derived? It is like nothing found anywhere out of India, and like no utilitarian form in India, that we now know of. It cannot be derived from the dome-like forms of the topes. They are circular both in plan and elevation. The Sikharas are straight-lined in plan, and their section is never a segment of a circle; it is not derived from any many-storeyed buildings, as the sikharas or vimânas of the Dravidian architecture of the south of India, which seem certainly to have been copied from the many storeyed vihâras of the Buddhists, and we cannot fancy any class of domestic building which could have formed a model out of which they could have been elaborated. One curious thing we do know, which is that all the ancient roofs in India, whether represented in the bas-reliefs or copied in the caves, were invariably curvilinear—generally circular or rather ogee—having a ridge added externally to throw off the rain from that weakest part; but nothing on any bas-relief or painting gives us a hint of any building like these sikharas.
v1.c19.s4.7
184. Diagram Plan and Section of the Temple at Kanârak, designed to explain the construction of Hindû Temples.
v1.c19.s4.8
Another curious and perplexing circumstance regarding the sikharas is that when we first meet them, at Bhuvaneswar, for instance, on the Bay of Bengal, or at Pattadakal in the 7th century, near the west coast of India, the style is complete and settled in all its parts. There was no hesitation then, nor has there been any since. During the twelve or thirteen centuries that have elapsed since the erection of these earliest known examples, they have gone on becoming more and more attenuated, till they are almost as pointed as Gothic spires, and their degree of attenuation is no bad test of their age; but they never changed in any essential feature of the design. All the parts found in the oldest examples are retained in the most recent, and are easily recognisable in the buildings of the present time.
v1.c19.s4.9
The one hypothesis that occurs to me as sufficient to account for this peculiarity is to assume that it was a constructive necessity. If we take for instance an assumed section of the diagram (Woodcut No. 184, p. 324), it will be seen how easily a very tall pointed horizontal arch, like that of the Treasury at Mycenæ referred to above, p. 312, would fit its external form. In that case we might assume that the tower at Bodh-Gayâ took a straight-lined form like the doorway at Missolonghi and the 'Gate of Lions' at Mycenæ, while the Hindûs took the more graceful curvilinear shape, which certainly was more common in remote classical antiquity,¹ and as it is found in Persia may have reached India at a remote period.
v1.c19.s4.10
¹ See Woodcuts Nos. 102, 114, 124, 126, 129, 172, 177 and 178 of vol. i. of the author's 'History of Ancient and Mediæval Architecture,' 3rd edn.; and for the Missolonghi doorway and Mycenæ Gate of Lyons,—*Ibid.*, Nos. 130 and 131 on p. 247.
v1.c19.s4.11
This hypothesis does not account for the change from the square to the circular form in the upper part, nor for its peculiar ornamentation; but that may be owing to our having none of the earlier examples. When we first meet with the form, either in Dhârwâr or Orissa, it is complete in all its parts, and had evidently reached that state of perfection through long stages of tentative experience. The discovery of some earlier examples than we now know may one day tell us by what steps that degree of perfection was reached, but in the meanwhile I fear we must rest content with the theory just explained, which, on the whole, may be considered sufficient for present purposes at least.¹
v1.c19.s4.12
¹ In his work on the 'Antiquities of Orissa,' vol. i., Bâbu Râjendralâl Mitra suggests at page 31 something of this sort, but if his diagram were all that is to be said in favour of the hypothesis, I would feel inclined to reject it.
v1.c19.s4.13
DRAVIDIAN ROCK-CUT TEMPLES.
v1.c19.s4.14
Mâmallapuram—Kailâs, Elûrâ.
v1.c19.s4.15
ALTHOUGH it may not be possible to point out the origin of the Dravidian style, and trace its early history with the same precision as we can that of Buddhist architecture, there is nothing so mysterious about it as there is regarding the styles of northern India, nor does it burst on us full blown at once as is the case with the architecture of the Chalukyas. Hitherto, the great difficulty in the case has been, that the temples of southern India have almost all been found to be of so modern a date. The great building age there was the 16th and 17th centuries of our era. Some structural buildings, it is true, could be traced back to the 11th or 12th with certainty, but beyond that all was to a great extent conjecture; and if it were not for rock-cut examples, we could hardly go back much further with anything like certainty. Recent investigations, however, combined with improved knowledge and greater familiarity with the subject, have now altered this state of affairs to a great extent. It seems hardly doubtful now that the Kailâs at Elûrâ, and the great temples at Pattadakal, are anterior to the 10th century.¹ In fact, it has been ascertained that they date from the 8th, and the "raths," as they are called, at Mâmallapuram or the "Seven Pagodas" on the Madras coast, are as early as the 7th century, and are in reality the oldest examples of their class known, and the prototypes of the style.
v1.c19.s4.16
¹ Burgess, 'Report on Belgam and Kaladgi,' 1875, plates 39, 40.
v1.c19.s4.17
One circumstance which prevented the age of the Mâmallapuram raths being before detected is, that being all cut in granite and in single blocks, they show no sign of wearing or decay, which is so frequently a test of age in structural buildings, and being all in the same material produces a family likeness among them, which makes it at first sight difficult to discriminate between what is old and what new. More than this, they all possess the curious peculiarity of being unfinished, whether standing free, as the raths, or cut in the rock, as caves, or on its face, as the great bas-relief; they are all left with one-third or one-fourth merely blocked out, and in some instances with the intention merely indicated. It looks as if the workmen had been suddenly called off while the whole was in progress, and native traditions, which always are framed to account for what is otherwise most unintelligible, have seized on this peculiarity, and make it the prominent feature in their myths. Add to this that it is only of late we have acquired that knowledge of the subject and familiarity with its details, which enable us to check the vagaries of Indian speculation. From all these causes it is not difficult to understand how easily mistakes may be made in treating of such mysterious objects.
v1.c19.s4.18
If we do not know all we would wish about the antiquities of Mâmallapuram,¹ it is not because attempts have not been made to supply the information. (Situated on an open beach, within 32 miles of Madras, it has been more visited and oftener described than any other place in India.) The first volume of the 'Asiatic Researches' (1788) contained an exhaustive paper on them by Wm. Chambers. This was followed in the fifth (1798) by another by Mr Goldingham. In the second volume of the 'Transactions of the Royal Asiatic Society' (1830) there appeared what was then considered a most successful attempt to decipher the inscriptions there, by Dr. Guy Babington, accompanied by views of most of the sculptures. Before this, however, in 1816, Colonel Colin Mackenzie had employed his staff to make detailed drawings of all the sculptures and architectural details, and he left a collection of about forty drawings, which are now, in manuscript, in the India Office. Like all such collections, without descriptive text, they are nearly useless for scientific purposes. The 'Madras Journal,' in 1844, contained a guide to the place by Lieutenant J. Braddock, with notes by the Rev. G. W. Mahon, the Rev. W. Taylor, and Sir Walter Elliot;² and almost every journal of every traveller in these parts contains some hint regarding them, or some attempt to describe and explain their peculiarities or beauties. With the exception of the Mackenzie MS.—the most of these were collected in a volume in 1869 by a Lieutenant Carr, and published at the expense of the Madras Government,¹ but, unfortunately, as too often happens, the editor selected had no general knowledge of the subject, nor had he apparently much local familiarity with the place. His work in consequence added nothing to our previous stores.²
v1.c19.s4.19
¹ The name of this place, among English writers, has been subject to various changes; a century ago it usually figured as Maha Balipuram, as in Southey's 'Curse of Kehama'; Dr. Babington stated that in the inscriptions it was called Mahâmallaipur; the Rev. W. Taylor made it Mâmallapuram, which is now accepted; other forms were Mavalivaram, Mahâvallipur, etc.
v1.c19.s4.20
² 'A Guide to the Sculptures, Excavations and other remarkable subjects at Mâmallaipur, generally known to Europeans as "the Seven Pagodas" by the late Lieut. John Braddock, etc.' in 'Madras Journal of Literature and Science,' vol. xiii. (June, 1844) pp. 1-56.
v1.c19.s4.21
¹ It included also a short account of the place written in Kanarese for Col. Mackenzie in 1803, with a translation. The publication was issued in two forms—in atlas folio, 96 pp., and also in octavo, 246 pp., with folding plates, and with the same errata.
v1.c19.s4.22
² A survey of the monuments at Mâmallapuram was made a considerable time ago, by the Archæological Survey of Madras.
v1.c19.s4.23
In 1883 the editor called attention to the temple of Kailâsanâthaswâmin at Conjivaram as a Pallava temple of probably about the 7th century, containing a number of early inscriptions.³ On examination by Dr. Hultzsch, these were found partly to belong to the same period as those of the raths at Mâmallapuram. The temple, now seemingly a Vaishnava shrine, was erected by the Pallava king Râjasimha
v1.c19.s4.24
³ Hultzsch, 'South Indian Inscriptions,' vol. i. pp. 8ff.
v1.c19.s4.25
185. Raths, Mâmallapuram. (From a Sketch by the Author.)
v1.c19.s4.26
*alias* Narasimhavarman II., towards the end of the 7th century and dedicated to Iswara or Siva; and the names on the Dharmarâja rath agree with those on this temple, confirming the date arrived at in 1880 for these works.⁴
v1.c19.s4.27
⁴ 'Cave Temples of India,' p. 110.
v1.c19.s4.28
The oldest and most interesting group of these monuments are the so-called five "Raths"¹ or monolithic temples standing on the seashore to the south of the other rock excavations. One of these, having an apsidal termination, appears in the centre of the preceding woodcut (No. 185) a little detached from the rest. The other four stand in a line running from north-north-east to south-south-west, and look as if they had been carved out of a single stone or rock, which originally, if that were so, must have been between 35 ft. and 40 ft. high at its southern end, sinking to half that height at its northern extremity, and its width diminishing in like proportion.
v1.c19.s4.29
¹ *Ratha* has much the same meaning as *Vimâna*—a chariot or covered car.
v1.c19.s4.30
The first on the north is the Draupadî Rath—a mere pansâlâ or cell 11 ft. square externally, and with a curvilinear roof rising to about 18 ft. high (Woodcut No. 186). Apparently it was once crowned by a finial of some sort, but its form cannot now be ascertained. This rath is the most completely finished of the five, and is now unique of its kind, but must have belonged to an extensive class of buildings when it was executed, and their form consequently becomes important in the history of the style. The cell inside measures 6 ft. 6 in. in depth by 4 ft. 6 in. across, on the back wall of which is a four-armed Saktî or female divinity, probably Lakshmî, with some attendants: the dwârpâlas also are females, as are the figures on the north, east, and south sides.
v1.c19.s4.31
186. Draupadî's Rath. (From a Photograph.)
v1.c19.s4.32
The next is known as Arjuna's rath, and is a small copy of Dharmarâja's—the last to the southwards—the only difference being that Arjuna's is very much smaller than the other, measuring 11 ft. 6 in. by 16 ft. in plan, and 20 ft. in height. A cell has been excavated inside, only 4 ft. 6 in. by 5 ft., but it contains no image; the figures on the outside walls, however, seem to show that it was dedicated to Siva. It is cracked from top to bottom, and part of the finial has fallen off. The roofs of the lower and first stories are ornamented with those ranges of little simulated cells which became the distinguishing characteristics of Dravidian architecture, and it is surmounted by an octagonal dome which is an equally universal feature of the style.
v1.c19.s4.33
187. Plan of Bhîma's Rath. (From a Plan by Mr. R. F. Chisholm.) Scale 20 ft. to 1 in.[^1]
v1.c19.s4.34
The third—Bhîma's rath—seen partially in the Woodcut No. 185, is very remarkable: it is an oblong building having a curvilinear shaped roof with a straight ridge. Its dimensions are nearly 48 ft. long, 25 ft. wide, and 25 ft. high. Externally, it seems to have been completely carved, but internally only partially excavated, the works being apparently stopped by some accident. It is cracked completely through, so that daylight can be seen through it, and several masses of the rock have fallen to the ground—this has been ascribed to an earthquake and other causes. My impression is, the explanation is not far to seek, but arose from unskilfulness on the part of workmen employed in a first attempt. Having completed the exterior, they set to work to excavate the interior so as to make it resemble a structural building of the same class, leaving only such pillars and supports as were sufficient to support a wooden roof of the ordinary construction (Woodcut No. 187). In this instance it was a mass of solid granite which, had the excavation been completed, would certainly have crushed the lower storey to powder. As it was, the builders seem to have taken the hint of the crack and stopped the further progress of the works.
v1.c19.s4.35
¹ The dotted hatching on this plan represents a suggested mode in which the rath might have been completed if finished as intended.
v1.c19.s4.36
It is a little difficult to say how it was intended to have been completed. The centre was occupied by a hall about 9 or 10 ft. wide by 30 ft. open on one or probably both sides, and intended to be closed at both ends. This central hall was surrounded by a verandah measuring 5 ft. 3 in. in the clear on the sides, but only 3 ft. at the ends. There would then have been ten or twelve pillars in the centre and two at each end. One of these is represented in the annexed woodcut (No 188), and they are all of the same pattern, which, in fact, with very slight modifications, is universal at Mâmallapuram. They all have bases representing Vyâlîs or conventional lions, with spreading capitals, and of proportions perfectly suited to a building of the dimensions of this one, if executed in wood.
v1.c19.s4.37
188. Pillar in Bhîma's Rath. (From a Drawing by Mr. R. F. Chisholm.)
v1.c19.s4.38
The fourth and most southerly, however — Dharmarâja's rath — is the finest and most interesting of the group. A view of it has already been given (Woodcut No. 89) and it is shown on the right hand of Woodcut No. 185. As will be seen from the annexed plan (No. 189), its dimensions are 26 ft. 9 in., by 28 ft. 8 in., and its height is rather more than 35 ft. It is consequently much larger than Arjuna's rath, but even with these dimensions it can only be considered as a model. The three upper storeys are ornamented with those little simulated cells mentioned above, and which are so universal in the south of India; the front of each of these cells with their connecting links, is adorned with a representation of one of those semi-circular dormer windows which are so usual in Buddhist architecture. Here each has a human head looking outwards. Behind these cells the walls are divided by slender pilasters into narrow compartments, and in each is placed the statue of a deity of the Hindû Pantheon, among which are found representations of [Figure: 189. Plan of Dharmarâja Rath. (From a Drawing by Mr. R. F. Chisholm.) Scale 20 ft. to 1 in.]
v1.c19.s4.39
Siva, Brahma, and Vishnu in various characters as Arddhanâri Narasimha, Varâha, etc. — none of them with more than four arms.¹ Over these figures are seventeen short epigraphs containing epithets that were long misunderstood; but from a comparison of them with the very similar series round the inside of the enclosure of the Kailâsanâtha temple at Kânchîpuram, a clue is obtained that enables us to fix the date of these monuments. Stated briefly, it stands thus:—On the east side of the third storey of this rath is the epigraph—"The temple of the holy Atyantakâma-Pallavesvara: Ranajaya," and this Atyantakâma, in the other storeys is styled Narasimha Srînidhi, Srîbhara, etc. On the monolithic temple of Ganesa, to be noticed below, and in the Dharmarâja Mantapa cave are identical inscriptions of the same Atyantakâma-Srînidhi-Srîbhara; and at Sâluvankuppam cave is an inscription of King Atiranachanda with the names of Atyantakâma, Srîbhara, Ranajaya, Kâlakâla, etc.—which are also epithets of Râjasimha in the Kânchî inscriptions. Now we learn from copperplate grants that the Pallava king, Râjasimha, bore the names of Kâlakâla, Narasimhavishnu and Narasimhavarman, and must have reigned about the last quarter of the 7th century. We can hardly escape the conclusion, then, that Râjasimha, Atyantakâma, Atiranachanda, etc., are all names (or *birudas*) of one king, the son and successor of Ugradanda-Lokâditya or Paramesvaravarman.² His dedications are all Saiva, and their occurrence on so many of the Mâmallapuram shrines supports the testimony, previously founded on the style alone, that they belong to one time, and were all excavated within a short period about A.D. 670 to 700.
v1.c19.s4.40
¹ 'Transactions of the Royal Asiatic Society,' vol. ii. plates 16, 17; or Carr's compilation.
v1.c19.s4.41
² Hultzsch, 'South Indian Inscriptions,' vol. i. pp. 1-24; and Fleet, in 'Bombay Gazetteer,' vol i. pt. i. pp. 322-326.
v1.c19.s4.42
As stated above, we have on this rath many of the gods of the Hindû Pantheon, but in forms more subdued than are to be found elsewhere. The one extravagance is that they generally have four arms—never more—to distinguish them from mortals; but none of those combinations or extravagances we find in the caves at Elûrâ, Elephanta, and elsewhere. It is the soberest and most reasonable version of the Pantheon yet discovered, and, consequently, one of the most interesting, as well probably, as the earliest.¹
v1.c19.s4.43
¹ These figures, properly drawn or photographed, would be almost indispensable for the illustration of Hindû mythology.
v1.c19.s4.44
The upper three storeys are entirely finished externally with
v1.c19.s4.45
190. Elevation of Dharmarâja Rath. Scale 13 ft. to 1 in. (From a Drawing by Mr. R. F. Chisholm.)
v1.c19.s4.46
their sculptures, the lower one merely blocked out; and it is difficult to say how far it was intended to excavate the interior. A cell was excavated to a depth of 5 ft. in the third storey, and it may have been intended to enlarge it. A similar attempt has been made in the second storey, but only to the depth of 4 ft. From there being six pilasters on the outside of the third storey, we may gather that in a structural building its roof would have been supported by thirty-six wooden posts, and in like manner that the second storey would have had sixty-four supports, but, of course, with some of these omitted in the centre. From its extreme irregularity it is not easy to suggest what may
v1.c19.s4.47
191. Section of Dharmarâja Rath (through A.B. fig. 189), with the suggested internal arrangements dotted in. Scale 10 ft. to 1 in.
v1.c19.s4.48
have been the intended arrangement of the lowest storey; but, from the wider spacing of the pillars externally, it is evident that in a structural building stone, and not wood, would have been employed. From the arrangement of the exterior we gather that there would have been four free-standing pillars in the centre, as shown in dotted lines in the plan and section (Nos. 189 and 191). It is not clear, however, how many of the eight piers that surrounded these four were free-standing or attached as pilasters to thick external walls. What stopped the completion of this and the other raths, we shall never learn. They are certainly very like Buddhist buildings, as we learn to know them from the early caves, and it seems hardly to admit of doubt that we have here petrifactions of the later forms of Buddhist architecture,¹ and of the first forms of that of the Dravidians.
v1.c19.s4.49
¹ Among the remains found at Bharaut is a bas-relief representing a building so exactly like the long rath here, that there can be no doubt that such buildings were used in the north of India two centuries at least before Christ, but to what purpose they were applied is not so clear. The one at Bharaut seems to have contained the thrones or altars of the four last Buddhas.
v1.c19.s4.50
The want of interiors in these raths makes it sometimes difficult to make this so clear as it might be. We cannot, for instance, tell whether the apsidal rath, called Sahadeva and Nakula’s, to the west of the line of the others, and forming the fifth of the group, was intended to reproduce a chaitya hall or a vihâra like that in Woodcut No. 63. Though small, it is one of the most interesting of the whole; but like the others, it is very unfinished, especially on the east side. Its dimensions are 18 ft. in length—north and south, by 11 ft. across, and about 16 ft. in height. It faces north, on which side there is a small projecting portico (Woodcut No. 193), supported by two pillars, and within is a small empty cell. Externally the back end is apsidal, and so perhaps, if on a larger scale, its interior might have been; as it is, it is too small, and the square form is more convenient in such an apartment.
v1.c19.s4.51
192. Plan of Sahadeva's Rath. (From a Drawing by Mr. R. F. Chisholm.) Scale 20 ft. to 1 in.
v1.c19.s4.52
The interest of this rath lies in the fact that it represented, on a small scale, the exterior of one of those chaitya halls, which form so important a feature in the western groups of Buddhist caves, of which, until the discovery of the Chezarla and Têr structures, we had no other instance from which to judge of what the external appearance may have been of the structural chaityas, from which the cave-temple examples were copied. But this rath being in several storeys, and the whole so conventionalised by the different uses to which they are applied for the purposes of a different religion, that we must not stretch analogies too far.²
v1.c19.s4.53
² Among the sculptures of the Gandhâra monasteries are several representing façades of buildings. They may be cells or chaitya halls, but, at all events, they are almost exact reproductions of the façade of this rath (see Woodcut No. 123, page 216). Being used as frameworks for sculpture, the northern examples are, of course, conventionalised; but it is impossible to mistake the identity of intention.
v1.c19.s4.54
The sixth—the Ganesa rath—situated at a distance of three-
v1.c19.s4.55
193. View of Sahadeva's Rath. (From a Photograph.)
v1.c19.s4.56
quarters of a mile north of the others—is represented in the woodcut (No. 194), which, strange to say, is the most nearly finished of any, and gives a fair idea of the form these oblong temples took. Though small, it is a singularly elegant little temple. In plan, its dimensions are 19 ft. by 11 ft. 3 in., and 28 ft. in height. It is in three storeys with very elegant details, and of a form very common afterwards in Dravidian architecture for gopurams or gateways. The roof has a straight ridge, adorned at the ends by Saiva trisulas, and similar emblems
v1.c19.s4.57
194. Ganesa Rath, Mâmallapuram. (From a Photograph.)
v1.c19.s4.58
crowned the dormer windows. The ridge was ornamented by nine small pinnacles, which also continue to be employed. Though entering in the side, this temple was never intended to be pierced through.
v1.c19.s4.59
On the back wall of the verandah is an inscription in old and very florid characters, dedicating the shrine to Siva by King Atyantakâma-Ranajaya,¹ now identified with Râjasimha Pallava of the end of the 7th century.
v1.c19.s4.60
¹ Hultzsch's 'South Indian Inscriptions,' vol. i. pp. 4-8.
v1.c19.s4.61
What interests us most here, however, is that the square raths are the originals from which all the vimânas in southern India were copied, and continued to be copied nearly unchanged to a very late period. Woodcut No. 195, for instance, represents one from Madurâ, erected in the 18th century. It is changed, it is true, and the cells and some of the earlier features are hardly recognisable; but the wonder rather is that eleven centuries should not have more completely obliterated all traces of the original. There is nothing, however, in it which cannot be easily recognised in intermediate examples, and their gradual transformation detected by any one familiar with the subject. On the other hand, the oblong raths were halls or porticos with the Buddhists, and became the gateways or gopurams which are frequently — indeed generally — more important parts of later Dravidian temples than the vimânas themselves. They, too, like the vimânas, retain their original
v1.c19.s4.62
195. Perumâl Temple at Madurâ. (From a MS. Drawing in the possession of the late General Monteith, Madras Engineers.) No scale.
v1.c19.s4.63
features very little changed to the present day, as may be seen from the annexed example from a modern Tamil temple on the opposite shore of the Gulf of Manâr (Woodcut No. 196). To all this, however, we shall have frequent opportunities of referring in the sequel, and it will become much plainer as we proceed.
v1.c19.s4.64
The other antiquities at Mâmallapuram, though very interesting in themselves, are not nearly so important for our history as the raths just described. The caves are generally small, and fail architecturally, from the feebleness and tenuity of their supports. The southern cave-diggers had evidently
v1.c19.s4.65
196. Entrance to a Hindû Temple, Colombo. (From Sir J. E. Tennent's 'Ceylon.')
v1.c19.s4.66
not been grounded in the art, like their northern compeers, by the Buddhists. The long experience of the latter in the art taught them that ponderous masses were not only necessary to support their roofs, but for architectural effect; and neither they nor the Hindûs who succeeded them in the north ever hesitated to use pillars of two or three diameters in height, or to crowd them together to any required extent. In the south, on the contrary, the cave-diggers tried to copy literally the structural pillars used to support wooden roofs. Hence, I believe, the accident to the long rath, and hence certainly the poor and modern look of these southern caves, which has long proved such a stumbling-block to all who have tried to guess their age. Their sculpture is better, and some of their best designs rank with those of Elûrâ and Elephanta, to which they were anterior. The Bâdâmi sculptures, executed in the 6th century (A.D. 579), are so similar in style with the best examples in the Mâmallapuram caves, that we had concluded they could not be far distant in date, and must be placed in the preceding century; and this has since been supported by the contents of the inscriptions in the Dharmarâja mantapam at the Sâluvankuppam cave.
v1.c19.s4.67
197. Head of the Nâga figure at Mâmallapuram.
v1.c19.s4.68
The great bas-relief on the rock, 90 ft. by 30 ft., is perhaps the most remarkable thing of its class in India. It is close to the Ganesa ratha, and is locally known as 'Arjuna's Penance,' but what it was meant to represent is still a puzzle. It is in two sections divided by a crack or split in the rock, in which are placed a great Nâga wearing a crown surrounded by a seven-fold hood (Woodcut No. 197); under him is a Nâginî with the usual triple hood; and below this a large cobra's head. The figures on both sides are directed towards this recess. On the left there is a small shrine below and a devotee worshipping at it, whilst behind are several wild beasts; above the shrine is a yogi standing on one foot before a deva, four armed, with a sort of sceptre and attendants, and behind and above him all the figures are in pairs—male and female—with only two arms each. On the
v1.c19.s4.69
198. Cave at Sâluvankuppam. (From a Photograph.)
v1.c19.s4.70
rock to the right, there are two large and some smaller elephants, and above them three tiers of figures, represented as floating through the air, all two-armed, mostly in pairs, some with birds' legs and wings (Gandharvas), and wild beasts behind. There seems nothing to enable us to fix its age with absolute certainty;
v1.c19.s4.71
it can hardly, however, be doubted that it belongs to about the 8th century.
v1.c19.s4.72
There is one other antiquity at a place called Sâluvankuppam, half a mile north of Mâmallapuram village, which deserves notice as a descendant of the tiger cave at Udayagiri near Katak (Woodcut No. 272). Here, not one, but a dozen tiger or lion heads welcome the anchorite to his abode, or rather, the devotee to his shrine (Woodcut No. 198). Here, too, they are conventionalised as we always find them in Chalukyan art; and this example serves, like every other, to show how the Hindû imagination in art runs wild when once freed from the trammels of sober imitation of natural things, which we found to be its characteristic in the early stages of Buddhist art.
v1.c19.s4.73
Here is an inscription in two different alphabets of King Atiranachanda, who has also the *birudas* or epithets of Atyantakâma, etc., from which he appears to be identical with the Râjasimha-Narasimha who executed the Dharmarâja rath and probably most of the excavated shrines at Mâmallapuram.