the Farhad-Beg-yailaki Site were determined. I may add that a decorated wooden comb found at this ruin, F. II. ooi, is also of a Western type.
In clearing the open hall or court immediately adjoining room i on the south-west (Fig. 31 7) we came upon a wooden boarding, 4 feet square, with posts at the corners. Cross-bars of wood between these carried a rough piece of matting, plastered on the upper surface, at a height of about 2 feet from the floor. The purpose of this boarding remains uncertain, no other finds having been made within or near it. But it curiously recalled the wooden boarding found within the large refuse-heap of N. xry at the Niya Site and once used to store fodder before it became a dustbin.' In the end room, ii, of the same wing, the only object discovered was the well-preserved wooden double-bracket, F. it. ii. oI (Plate XVII). Its shape and well-carved ornamentation are of special interest because, on the one hand, they show close descent from the carved double-brackets found at the Niya Site,B and, on the other, clearly indicate a transition towards the double-bracket brought to light at Khadalik.° I have already had occasion, when discussing the stucco pilasters and cantilevers of.the Miran temple M. II, to trace step by step the development which this architectural motif has undergone in the art of the Tarim Basin since its original model was borrowed from the Persepolitan columns and cantilevers of Gandhara sculpture.10 Referring for a detailed description of this fine piece of wood-carving to the List below, it will suffice to point out that in conjunction with collateral evidence it has its value for the dating of the Farhad-Bag=yailaki Site. I need scarcely add that this carving, like all the other timber used in the structures of this site, is of the wood of the cultivated poplar.
The rooms adjoining i in the north-east wing contained well-preserved fireplaces and sitting-platforms by their side, but furnished no objects. Beyond them to the east and on a slightly higher level stood a temple cella, iii, 18 by 14 feet within, enclosed on three sides by a circumambulatory passage about 6 feet wide. The sand covering it was less than 3 feet deep, and consequently what remained of the plastered walls retained only the scantiest traces of .their original fresco decoration. Of the stucco relievos once adorning the cella only fragments survived, which mostly appeared to have belonged to one vesica. The largest piece, F. Ii. iii. ooi. a (Plate CXXXIx), shows well-modelled small appliqué figures of seated Buddhas within a floral border which in its motifs differs from the corresponding stucco wall ornaments both of Rawak and of Dandân-oilik. The square image base within the cella was broken to within a foot or so of the ground, and had evidently been burrowed into for ` treasure '. But close behind it there were discovered two painted panels, F. II. iii. 1, 2, and along the•foot of the adjoining cella wall five more.
They all lay obviously in the place where they had been deposited as votive offerings by the last worshippers, and show only too clearly the damage they had undergone through corroding sand and exposure before they were finally buried. A detailed description of them all will be found in the List below. Two among them, F. II. iii. 5, 6, show Buddha figures of rough workmanship. But others, in spite of poor preservation, are of artistic interest. Thus in F. it. iii. 2 (Plate Cxxv) we see the very well-proportioned and gracefully posed figure of a standing Bodhisattva (?), with designs of seated Buddhas, a bird, etc., painted on the nude portions of the body. F. H. iii. oo2 (Plate CxXV), 17 inches high, displays on one side the richly draped figure of a standing Buddha with a kneeling donor at his feet. The heads of this Buddha and the Bodhisattva in F. n. iii. 2 have features of a very pronounced Indian type. The same is the case also with the elaborately dressed armed figure seen on the reverse of F. II. iii. 002 (Plate CXXV) riding a camel. His mount