of the class. This temple, though entering in the side, was never intended to be pierced through, but always to contain a cell. The large oblong rath, on the contrary, was intended to be open all round, and whether, consequently, we should consider it as a choultrie or a gopura is not quite clear. One thing, at all events, seems certain--and it is what interests us most here--that the square raths are copies of Buddhist viharas, and are the originals from which all the vimanas in southern India were copied, and continued to be copied nearly unchanged to a very late period. Woodcut No. 183, for instance, represents one from Mádura, 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 twelve 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 porticoes with the Buddhists, and became the gopuras or gateways which are frequently--indeed generally--more important parts of Dravidian temples than the vimanas themselves. They, too, like the vimanas, retain their original 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 Manaar (Woodcut No. 184). To all this, however, we shall have frequent opportunities of referring in the sequel, and it will become much plainer as we proceed. [Illustration: 184. Entrance to a Hindu Temple, Colombo. (From Sir J. E. Tennent’s ‘Ceylon.’)]