The cave itself is very small, only 31 ft. across and about 19 ft. deep, and it is a little uncertain whether the groups of figures at either end of the verandah are integral, or whether they may not have been added at some subsequent period. The inner groups, however, are of the age of the cave, and the architecture is unaltered, and thus becomes a fixed standing-point for comparison with other examples; and when we come to compare it with the groups known as the Indra Subha and Jaganât Subha at Ellora, we cannot hesitate to ascribe them to about the same age. Hitherto, the Jaina group at Ellora has been considered as the most modern there: an impression arising partly from the character of the sculptures themselves, which are neither purely Jaina nor purely Hindu--more, however, from the extreme difficulty of comparing rock-cut examples with structural ones. Our knowledge of the architecture of temples is, in nine cases out of ten, derived from their external forms, to which the interiors are quite subordinate. Cave-temples, however, have practically no exteriors, and at the utmost façades modified to admit more light than is usual in structural edifices, and then strengthened and modified so as to suit rock-cut architecture. As no ancient Jaina temple hitherto known had a dated inscription upon it, nor a tolerably authenticated history, it is no wonder that guesses might be wide of the truth. Now, however, that we know positively the age of one example, all this can be rectified, and there seems no doubt that all the Indra Subha group were finished before the cataclysm--say before A.D. 750. [Illustration: 147. Entrance to the Indra Subha Cave at Ellora. (From a Photograph.)] When with this new light we come to examine with care the architecture of these façades, we find the Ellora group exhibits an extraordinary affinity with the southern style. The little detached shrine in the courtyard of the Indra Subha, and the gateway shown in the above woodcut (No. 147), are as essentially Dravidian in style as the Kylas itself, and, like many of the details of these caves, so nearly identical that they cannot possibly be distant in date. May we, therefore, assume from this that the Chalukyan kingdom of Kalian, in the 7th century of our era, extended from Ellora on the north to Badami on the south, and that all these rock-cut examples, with the temple at Aiwulli (Woodcut No. 120), were excavated or erected under their auspices?