Afghanistan caves were copied from those of India. In the Indian caves we find that this form began to be used as a decoration; the same as that which took place in Europe with the Greek pediment, which was also a gable, and has been largely applied for merely ornamental purposes. The Hindus adopted the circular gable as an ornament: in their hands it became decorative, and was made more ornamental, and you will scarcely find a temple in India where this form cannot be traced somewhere in its ornamentation. In the Dravidian, or Southern Indian, style it is to this day the predominating characteristic of the decoration, and it even yet affects some of the constructive details. I sketched this easily-recognized shape on the old topes of the Jelalabad Valley, where it had been carried and there applied as an architectural ornamentation. This form can be traced from Ceylon to the Hindu Kush—a wide space-over which it has spread, and to the inquiring mind, it calls for some explanation of its first origin. The oft-recurring question was, why did the early people of India construct this peculiar kind of roof? We know that all architectural forms had at first a reason for their existence, but in seeking for the source in this case no answer has yet been found.

While in Persia and the Afghan Frontier lately, I took much interest in the facility with which roofs, where wood is scarce, were there produced by means of sun-dried bricks. The dome is the usual method, but it was very common to find oblong houses covered with barrel roofs. Some of these had a semi-dome at one end, with the circular arch as the gable at the other. Now we know that the Chaitya halls had this semi-dome at the further end;-this, I confess, struck me very forcibly, for the one form is an exact repetition of the other; and I speculated on the possibility that I had found the origin of the Chaitya circular roof. There are certainly probabilities in favour of the theory: we know that there are forms common to both Indian and the Ancient Persian Architecture; mud-bricks were as common on the one side of the Indus as on the other, and barrel-roofs may have been the same. If such were the case, it might be possible

that this form had been copied in wood, where that material may have chanced to be more easily procured than bricks.

The suggestion produced by this Persian roof, although it is a very remarkable coincidence, I have entirely rejected, my reason being that I have what I consider to be now a better theory to offer.

Not long ago I chanced to pick up a book at a stall, called A Phrenologist among the Todas. To me the Phrenology of the Todas was the least important part of the book, but it is all interesting as an account of personal experience among these strange people, and parts are given with much humour. The illustrations are in Photography-the frontispiece caught my eye while buying the book-in it is a representation of a house, and the more I have looked at this peculiar structure, the more I am inclined to think that it gives the true origin of the early round roof of India. In a case of this kind there is no direct, or what might be called demonstrative, evidence; all that can be offered in favour of the idea is coincidence in form, with the highly possible chance that the peculiar manner of construction, belonging to what is supposed to be one of the Aboriginal races, dates back to an early period.

3

I turned up Fergusson, to see if he had chanced to light upon these houses, and I was delighted to find that he had, and his notice appears in a note which I had at the moment forgotten. He refers to the work of Mr. Breeks, and I find that his conclusions coincide exactly with my own. Of these structures he states that," Their roofs have precisely the same elliptical forms as the Chaitya with the ridge, giving the ogee form externally, and altogether, whether by accident or design, they are miniature Chaitya halls. Externally they are covered with short thatch, neatly laid on. Such forms may have existed in India two thousand years ago,

1 By William E. Marshall, Lieut.-Colonel of H.M. Bengal Staff Corps, 1873. 2 History of Indian and Eastern Architecture, p. 105.

3 An Account of the Primitive Tribes and Monuments of the Nilagiris, by the late James Wilkinson Breeks, of the Madras Civil Service, 1873. A work full of most valuable information; but so far as the Todas are concerned, I prefer Col. Marshall's book, as it deals with them alone, and its information regarding the one tribe is much more complete.

The TODA MAND or HUT.

From Colonel W. E. Marshall's "A Phrenologist amongst the Todas," frontispiece.