The third, the Das Avatar, is another two-storeyed cave, very similar in all its architectural details to the two preceding, but the sculptures are all Brahmanical. At first sight, it seems as if the excavation had been made by the Buddhists, and appropriated and finished by their successors. This may be true to a certain extent, but on a more careful examination it appears more probable that we owe it entirely to the Brahmans. It is evidently the earliest Brahmanical temple here, and it is natural to suppose that when the Saivites first attempted to rival their antagonists in cave-temples they should follow the only models that existed, merely appropriating them to their own worship. The circumstance, however, that makes this most probable, is the existence of a pseudo-structural mantapa, or shrine of Nundi, in the courtyard; this evidently must have been a part of the original design, or the rock would not have been left here for it, and it is a model of the usual structural building found in Saiva temples in different parts of India. This is a piece of bad grammar the Buddhists never were guilty of; their excavations always are caves, whilst the great characteristic of Brahmanical excavations, as distinguished from that of their predecessors, is that they generally copied structural buildings: a system that rose to its greatest height in the Kylas, to be described further on. The Buddhist excavations, on the contrary, were always caves and nothing else. It is not easy, in the present state of our knowledge, to determine whether the Ellora Buddhist group is later or earlier than those of Dhumnar and Kholvi. It is certainly finer than either, and conforms more closely with the traditions of the style in its palmiest days; but that may be owing to local circumstances, of which we have no precise knowledge. The manner, however, in which it fades into the Hindu group is in itself sufficient to prove how late it is. If we take A.D. 600 as the medium date for the Viswakarma and its surroundings, and A.D. 750 as a time when the last trace of Buddhism had disappeared from western India, we shall probably not err to any great extent; but we must wait for some inscriptions or more precise data before attempting to speak with precision on the subject. A great deal more requires to be done before this great cartoon can be filled up with anything like completeness; but in the meanwhile it is satisfactory to know that in these “rock-cut temples,” eked out by the few structural examples that exist, we have a complete history of the arts and liturgies of the Buddhists for the thousand years that ranged from B.C. 250 to A.D. 750; and that, when any one with zeal and intelligence enough for the purpose will devote himself to the task, he will be able to give us a more vivid and far more authentic account of this remarkable form of faith than can be gathered from any books whose existence is now known to us. JUNIR.