Before leaving this neighbourhood there is still one small cave that is worth mentioning as the only other known of the same age as those of Barabar and Rajgir.1 It consists of a chamber rectangular in plan, and measuring 15 feet 9 inches, by 11 feet 3 inches, which is hollowed out of an isolated granite boulder lying detached by itself, and not near any other rocks. Inside it is as carefully polished as any of those at Barabar, except the inner wall where the surface has peeled off.2 Its principal interest, however, resides in its section (woodcut, No. 11), which is that of a pointed arch rising from the floor without any perpendicular sides, which are universally to be found in the other caves here. The jambs of the doorway also slope inwards nearly in the ratio of 3 to 4, from both which peculiarities I would infer that this may be the oldest cave in the neighbourhood. We must however have a more extended series of examples before we can form a reliable sequence in this direction, but it is only by quoting new examples as they turn up that we can hope to arrive at such a chronological scale; in the meantime, however, we may feel sure that this hermitage belongs to the great Mauryan age, but whether before or after Aśoka's time must be left at present undetermined; my impression at present is that it is the oldest thing of its class yet discovered in India.

On the banks of the Sona river, above Rohatsgarh, there are several excavations, some of them apparently of considerable extent, but they have never yet been examined, so far at least as I can learn, by anyone who could say what they were, nor of what age. We must consequently wait for further information before attempting to describe them. Further up, in the valley of the same river, at a place called Harchoka, there are some very extensive excavations, regarding which it would be very desirable some more information could be obtained. The place is situated in latitude 23° 51' 31", longitude 81° 45' 34", as nearly as maybe 110 miles due south from Allahabad, and as it is only 70 miles south-east from Bharhut, it seems a pity it was not visited by General Cunningham, or one of his assistants, while exploring that country in search for fragments of that celebrated stupa. What we know of it is derived from a paper by Captain Samuells in Vol. XL. of the Journal of the Royal Asiatic Society, p. 177 et seq., which is accompanied by a plan and section very carefully drawn, but the latter unfortunately on so small a scale that its details are undistinguishable. As Captain Samuells does not profess to be an archaeologist his text does not afford us much information, either as to the age of this excavation, nor as to the religion to which it was dedicated. If an opinion may be hazarded, from the imperfect data available, I would suggest that this cave is contemporary with the late Brahmanical caves at Elurâ, and consequently belongs to the 7th or 8th century, and that the religion to which it was dedicated was that of Siva.3

It may at first sight appear, that more has been said in the preceding pages, with reference to these Behar caves than their importance justifies. Looked at from an architectural point of view, this is undoubtedly the case, but from their being the oldest caves known, and their dates being ascertained with all desirable precision, a knowledge of their peculiarities forms a basis for what follows, without which our knowledge would still rest on a very unstable foundation.

From the experience gained by our examination of these caves we gather, first:—

That all the caves with which Buddha's name or actions are associated were mere natural caverns unimproved by art, except in so far as some of them have been partially lined with brickwork, but in no instance are they entitled to be called rock-cut.

Secondly. That the earliest rock-cut examples were, even internally, plain unornamented chambers with polished walls, their roofs imitating the form of woodwork, or it may be that of bambu huts.4 That what ornament was attempted externally, as in the Lomas Bishi cave, was a mere copy of a wooden construction, and that any extension that was required, as in the Son Bhandar cave, was actually executed in wood.

Thirdly. That all the jambs of the doorways slope inward, following the lines of the posts supporting the circular roofs, which were made to lean inwards to counteract the thrust inherent in that form of construction.5

Lastly. That all the rude unknown caves may be considered as anterior to the age of Chandragupta, and all those, in Behar at least, with sloping jambs may be assumed to be comprehended within the duration of the Mauryan dynasty, which ended about 180 B.C.; the angle of rake being probably the best index yet obtained for their relative antiquity.

• 1. It is situated at a place called Sita Marhi, 14 miles south of Rajgir, and 24 east from Gâya, as nearly as I can make out from the map attached to Mr. Beglar's report, but the spot is not marked, though the name is.
• 2. Mr. Beglar, from whose report (viii. p. 106) these particulars are taken, mentions some pieces of sculpture as existing, and now worshipped in the cave, but whether they are cut in the rock or detached is not mentioned, and is of very little consequence, as they are evidently quite modern.
• 3. In the year 1794 Captain Blunt visited two extensive sets of caves at a place called Mara, in the neighbourhood, and described them in the seventh volume of the Asiatic Researches. Captain Samuells seems also to have visited them, but as he does not describe them he probably thought them of less importance than those at Harchoka.
• 4. In no instance is it possible to conceive that they were copies of constructions either on stone or brick.
• 5. I shall be very much surprised if it is not found that the walls in the Barabar caves do also lean inwards; but they have not yet been observed with sufficient accuracy to detect such a peculiarity.