edges of robe, and borders alternately of halo and vesica. At bottom is larger similar Buddha, having dark red robe, white flesh, copper-green halo, and vesica of flames (?) in buff, red, and green, with bead border.
.Along top are petals of larger Padmâsana. Rough work. i' i or x i r °.
F. xir. ooio. Fresco fr. from frieze at foot of wall in south corner. Lower part shows red and white, and blue and white, trefoil flowers with white tendrils scattered on maroon ground; also a lotus bud, red outlined black. Above, a green and blue lotus bud on red ground. 161- x r 2". For position in situ, see Fig. 314.
SECTION III.—FROM DOMOKO TO KHOTAN
From Farhâd-Bag-yailaki I moved on March 13 south past the present oasis of Domoko in order to examine remains which ROze and his companions had traced at a point known as Karayantak in the area of scrub-covered low sand hillocks stretching east of .the stream of Domoko. It proved to be situated less than a mile and a half due east of Mazâr-toghrak, the site explored in October, 1906, and about the same distance to the south-east of the great dyke at the head of the Domoko-yâr.1 The remains were those of a completely destroyed Buddhist shrine, which in plan and decoration must have shown the closest resemblance to the main temple of Khâdalik, though probably smaller. The clearing of the sand which covered the remains to a height of 9-10 feet in most places (Fig. 319) had not proceeded very far before it became clear that here, just as Khâdalik, the ruined structure had been worked as a quarry for timber. With the exception of a few feet length of foundation beams and the fragments of posts, etc., seen in Fig. 319, all pieces of wood which could be of use had been removed, obviously before heavy sand had accumulated at the ruin. Small pieces of painted wood chipped off from posts, etc., were plentiful. On the other hand, there was no sign of destruction by fire.
Apart from pieces of frescoed wall plaster, K.Y. 002-3 ; 1. oo18, showing a diaper of small seated Buddha figures and part of a large vesica, nothing remained of the cella walls, no doubt built of timber and wattle. About 10 feet east of where these plaster pieces cropped out there were found remnants of what evidently had been a central image base in stucco. Near this, amidst -débris of shattered wood and plaster, were found small relievo fragments in stucco including the heads of Gandharvi-like figures (K.Y. I. 001-2, Plate cxxxix), closely allied in style to those recovered at Khâdalik ; a wooden Pahl. board with a faded inscription in cursive Central-Asian Brâhmi, and seven clay impressions from an intaglio stamp showing a Bodhisattva seated on a lotus throne and modelled in purely Indian style (for a specimen see K.Y. I. oo10, Plate CXxxix). A clay mould, struck from the relief of one in this series, evidently for the sake of producing more of these votive offerings, is seen in K.Y. 1. oo16 (Plate Cxxxix). In the same place were found the fragment of a Buddha statuette in wood, K.Y. I. 0020 (Plate CxXXVI1I), measuring a little over one foot across the knees, and the badly effaced painted panel, K.Y. I. 0021. Of a small chapel, which seems to have adjoined on the south the approach to the passage round the cella, there survived two image bases in stucco. Remains of a frescoed band about 6 inches high at the foot of one of them showed kneeling figures of a family group, evidently of the donor, the whole carefully painted but much injured by ` Shbr'. From this was recovered the fragment K.Y. 11. ooi.
The conclusion drawn from these scanty art remains as to the date of this shrine, and probably of its abandonment also, which is approximately the same as that determined for the Khâdalik temples, received striking confirmation by the discovery on the floor of a well-preserved coin of the Chien ytian period (A. D. 758-9), showing no sign of wear. As in the case of Khâdalik and Mazârtoghrak, it appears very probable that this shrine, too, and the settlement likely to have existed around it were deserted about the close of the eighth century. The antiquarian and geographical interest presented by this simultaneous abandonment has already been discussed.2
I See above, pp. 203, 205 sq. 2 See above, pp. 207 sqq.