Sec. in THE REMAINS OF AK-SIPIL AND KIGHILLIK 477
valley-like depression between two ridges of sand about 6o ft. high, and running approximately from north-west to south-east. The length of the Nullah between them was about 17o yards, and the average breadth of the flat sandy bottom about 8o ft. From a huge heap of consolidated dung and other refuse, referred to hereafter, the spot has received among the treasure-seeking fraternity the designation of Kiglaillik ` the dunghill '. Close to the southern end of the Nullah I found a low mound, about 24 ft. in diameter, of what looked like loess earth mixed up with decomposed wood. It indicated the place of a structure, the original form and dimensions of which could no longer be made out. The mound rose only 2 to 3 ft. above the pottery-strewn ground near by ; apart from the thin cover of drift-sand its material proved to consist mainly of decayed plaster such as was used in the walls of the ancient buildings described at Endere, but here quite soft, and often showing a reddish colour. The small pieces of charred wood which also turned up in it suggest fire as the cause of the latter.
That this débris heap had already been searched through again and again by ` treasure-seekers ' was well known to Turdi, and clearly proved by the complete confusion in which I found mixed up drift-sand, decayed plaster, decomposed timber, and small relief fragments in stucco. The latter, of which close on sixty pieces were recovered in the course of a careful search, were found both in the centre of the little mound and within some yards of it. The exceptional hardness of these small fragments, and the peculiar fissured and discoloured appearance of their surface, which in some cases bore clear marks of having been scorched, at once attracted my attention. The assumption that these stucco pieces received their present appearance in a fire that consumed the structure naturally suggested itself. It was confirmed when the closer examination made subsequently by Mr. Andrews showed that the stucco in every one of these fragments, as described under A. 01-057 in the list below, was either vitrified or burnt. But whether this accidental burning would also account for their exceptional hardness was a question that long remained doubtful, as well as the determination of the materials in which the reliefs had been worked.
These questions have now been definitely settled by the result of the valuable chemical analysis to which Professor A. H. Church, F.R.S., has subjected specimens of these reliefs. As his Appendix (F) shows, the material was essentially ` plaster of Paris ', as also in the DandânUiliq stucco reliefs, ` the native gypsum having been prepared in the usual way by moderate heating (called " burning ") and then mixed with water, just as is now done in making plaster casts.' But in the case of the plaster of Paris of the Kighillik' reliefs three peculiar features, revealed by the chemical analysis, prove ` that it has been subjected, after completion, to a high temperature and to an atmosphere charged with what chemists call " reducing matters"'. With reference to these peculiar features, among which the low percentage of water present deserves special mention, Professor Church has subsequently been kind enough, in a letter dated November 23, 1904, to state his opinion ` that an accidental fire in the presence of vegetable matter has in all probability caused them. It seems a case of chemical reduction hardly contemplated by the stucco-makers '. Thus the fact that the shrine which these stucco-reliefs decorated had been destroyed by fire may now be considered as definitely established.
That the structure was indeed a Buddhist shrine, and one adorned in a fashion resembling that of the small Dandan-Uiliq temples, is clearly shown by the reliefs themselves. Fragmentary as they are they could all be recognized as parts of a wall decoration which consisted of a series of small Buddha figures in relief arranged within a large aureole. The edges of the latter were formed by overlapping lotus-leaves in relief with rows of overlapping flame-tongues ranged on the outside. A comparison of the descriptions given of all the pieces, and of the