In order to avoid repetition it will probably be more convenient to pass over far the present Arjuna's Ratha, which comes next in the series locally and to describe that in conjunction with the one bearing the name of Dharmarâja, which it resembles in every essential particular, the one being a copy of a three-storeyed, the other of a four-storeyed Buddhist Vihara. If this is, done the next will be that called Bhîma's Ratha, which is the largest of the group. It belongs to the same style as the Ganeśa temple just described, except that, as in the two last mentioned examples the conditions as to size are reversed; the smaller, the Ganeśa, is a three-storeyed, while Bhîma's is a two-storeyed Dharmaśâlâ or hall of assembly.
Its dimensions in plan are 48 feet in length by 25 in breadth, and it is about 26 feet in height. As will be seen from the annexed plan, it is a little difficult to say that what its disposition internally may have been intended to have, been it complete. The centre was occupied by a hall measuring 9 or 10 feet by 30, open certainly on one, probably on both sides, and probably intended to be closed at both ends.
It is, however, by no means clear that the eastern wall was intended to be removed and pillars substituted for it. In the account of the hall in which the first convocation was held, it is stated in the Mahawanso1, that the priest who read Bana, or the prayers, did so from a splendid pulpit at one end of the hall, but the president was seated in the centre of one side facing the assembly. The same disposition is described by Spence Hardy2 and M. Bigandet3, and would exactly suit such a hall as this, supposing the wall on one side to remain solid, but would be inconvenient and unlikely, if it were removed and pillars substituted. As the Mahawanso was probably describing (in the fifth century) some ordinary form of Buddhist ecclesia, or hall of assembly, it seems not unlikely that this was the type of those in use at that time, and consequently that the wall on one side was solid and not pierced, except, perhaps, by doors.
This central hall was surrounded by a verandah measuring 5 feet 3 inches in the clear on the sides, but only 3 feet at the ends. Assuming, however, that the hall was open on both sides, there would then be twelve pillars in the centre and two at each end. One of these is represented in the annexed woodcut, and they are all of the same pattern, which, in fact, with very slight modification, is universal at Mahâvallipur. They all have bases representing Yâlîs and conventional lions, with spreading capitals, and of proportions perfectly suited to a building of the dimensions of this one, if executed in wood. So little experience, however, had the Pallavas, or whoever undertook these works, in the material they were employing, that they actually set to work to copy literally a wooden building in granite. The consequence was, that even before they had nearly completed the excavation of the lower storey, the immense mass of material left above, settled and cracked the edifice in all directions and to such an extent as to necessitate the abandonment of the works, while they were in even a less finished state than those connected with the other Rathas. Not only is there a crack of some inches in width, right through the rock, but several of the little simulated cells have supped down for want of support, and give the hole a ruined, as well as an unfinished aspect.
The upper storey or clerestory, as we would call it in a Gothic building, with its five windows —one over each intercolumniation—is so nearly complete as to enable us to realise perfectly what was the structural form it was intended to imitate, but nothing to indicate with what materiel the roof of the original was covered. The most probable suggestion seems to be, that it was with thatch, though the thickness seems scarcely sufficient for that purpose, and metal could hardly have been laid on without rolls or something to indicate the joinings. On the other hand, it is hardly conceivable that they could frame carpentry so solidly a to admit of their roofs being coated with plaster or chunam, without cracking, to such an extent as to admit the rain. As represented here, it consisted of a solid mass, about a foot in thickness, formed into a pointed arch with barge boards at the ends. It may have been thatched, but judging from the construction simulated both at the ends and sides, the roof must have been strongly framed in timber, both longitudinally and transversely.5. Bharkut Stupa, p. 121, Plate XXXI.