fragments came to be buried beneath the slope of the sand-cone. The cone rose now fully sixteen feet above the level indicated by the débris. It is reasonable to assume that the origin of the sand-cone goes back to the very period, about the close of the eighth century A. D., when these settlements about Khâdalik were deserted. The rate of growth thus deduced subsequently received striking confirmation by the exactly corresponding observation made at the Buddhist temple, F. x11, at the ruined site of Farhâd Bég-yailaki about nine miles to the north-west of Khâdalik (see Map No. 31). This temple had originally been built on the top of an old tamarisk-cone which has grown just
eighteen feet above the height it showed at the time when the shrine was abandoned, probably
about the close of the eighth century A. D.5
Very instructive, too, was the visit which on September 3o I paid to some minor remains south Site of Kökof Khâdalik. The site of Kök jigda, thus called from an old and still living Jigda or Eleagnus-tree, jigda.
proved to be about a mile and a half distant to the south-east. In the middle of an open and slightly eroded area, measuring about 200 yards from east to west and about 120 yards across at its broadest, rose a' low mound covered with broken pieces of coarse wall plaster, reed wattle, stable refuse, and other marks of old occupation. The remains extended over a space about thirty yards in diameter. A trial trench failed to show the exact position of walls, but revealed the plastered flooring at a depth of nearly four feet and proved that the wattled walls had been constructed with reeds fixed horizontally as in the shrines and better-class dwellings of Dandan-oilik. Since this system of wattle construction is never met with in remains of the Muhammadan period, it seems safe to assume that this site, too, was abandoned about the same time as Khâdalik.
More positive proof of such simultaneous abandonment came to light at Darabzan-dong, as Mullah Khwâja called the southernmost of his small sites. I found it to lie due south at a direct distance of two and a half miles from Khâdalik and close to a winding reed-filled depression which clearly marked an earlier course of the Domoko stream. Near the left bank of this, and just south of a track connecting the northernmost part of Domoko with the oasis of Achma, there rose a small plateau which presented the characteristic features of a ` witness ', as compared with the wind-eroded ground immediately adjoining. Its northern portion was higher, rising about twenty feet above the depression, and measured about two hundred feet from east to west, and a hundred and ten feet at its broadest. Its top revealed no structural remains ; but the numerous cuttings made into it, evidently by ` treasure-seekers ', and the layers of straw and farmyard refuse which cropped out on the edges, were evidence that these had once existed.
At its south-eastern end this plateau was adjoined by a terrace of which the top, about seven feet lower and thirty-four feet in diameter, still retained a thin layer of débris less than a foot in height. Here a few sand-worn pieces of plaster attracted my attention, and on clearing the ground we came upon a number of much-decayed pieces of stucco evidently from a frescoed temple wall. One of the fragments (Dar. oo4 in list below) showed a small Buddha head painted in the style of the Khâdalik frescoes. On two other fragments of stuccoed wall surface a few characters of what looked like very cursive Brâhmi script could just be made out. Among sculptural remains were two stucco heads modelled in high relief, of which one, Dar. 008, still retaining plentiful remains of paint, is reproduced in Plate CxXXVIII. These heads as well as the fragment of a lotus-petal vesica border (Dar. 003) showed close resemblance in style to the work of the Khâdalik shrines. A painted panel and the portion of a Brâhmi Pahl containing at least twenty leaves, which also turned up from under the scanty cover of sand, had unfortunately suffered badly by rotting. Yet they, too, helped towards the approximate dating of the ruin.