standing the proximity of the graves to ground that was occasionally flooded. From this and the good preservation of the grave mounds it seemed safe to conclude that the Muhammadan settlement to which this small cemetery belonged could scarcely date back more than a century or two. That I could get no information about it either from Abdulmalik or the men of Tikenlik was not surprising, considering the comparatively recent occupation both of this place and of Singer by people from other parts.
Before our first approach to the plateau at the mouth of the flood-bed we had already come upon potsherds, and had noticed an abundance of such ` Tati ' debris when we crossed the tract of bare clay towards the ruined circumvallation, situated about half a mile to the south-west of it (Pl. 36). Most of the pottery fragments and other small objects of stone, metal, and glass picked up here and near the ruined fort, of which specimens are described under Ying. I. o4-17, II. oI-16 (Pl. CX), 019, looked to me old ; but no definite indication as to date can be drawn from the few ornamented pieces. A Chinese coin which was picked up on the ` Tati ' to the east of the circumvallation shows the legend Wu-shu used during Han times. About half-way across this ` Tati ' there lies another small Muhammadan burial-place, already noticed by Dr. Hedin, with about twenty-three graves exactly resembling those above described. I t may here be mentioned that when crossing the bare ` Tati ' ground due east of the circumvallation on a subsequent visit, we noticed two roughly made wooden ploughshares closely resembling in shape those now used in oases of the Tarim basin and also two stone hand-mills. The slanting rays of the setting sun showed up low narrow embankments, such as divide Turkestan fields for the purposes of irrigation, suggesting that this flat open ground had been under cultivation in comparatively recent times, while a Muhammadan colony reoccupied the site.
The ruined circumvallation (Fig. 341) proved to be exactly circular, enclosing an area 194 yards in diameter within the inner foot of the rampart. This was built for the most part of stamped clay and irregular layers of tamarisk trunks and brushwood. But portions of it both to the south and north consisted only of stamped clay with a thick layer of matted tamarisk branches covering the top. I noticed no vertical posts or other timber bracing. The varying methods of construction which appeared to have been followed may, partly at least, be due to later repairs. The thickness of the rampart was about 24 feet at its base, and its height where fairly well preserved, as along a portion of the northern segment, over 18 feet. A thick layer of brushwood appears to have been used throughout to secure the top. There was found also an abundance of large stones, evidently intended for defence. The position of two gates to the west and east was marked by gaps about 3o feet wide exactly facing each other. Smaller openings on the north and south were manifestly mere breaches caused by minor branches of the flood-bed passing through the interior. The almost total absence of structural remains within the circumvallation can be accounted for by the periodical flooding to which the interior has evidently been subject since the site was abandoned. This would inevitably cause the rapid destruction of quarters built probably in most cases only of wattle and plaster. Only near the centre did I find debris of bricks and kisek, indicating the position of some structure. A live tamarisk-cone which had grown up over its remains appeared to have partially preserved them until recently. But they had been utterly disturbed by the same excavators who had been at work on the ruined shrine westwards to be presently mentioned. The ground within the little fort was covered with soft disintegrated clay. So it is scarcely surprising that the only object found here was the small silver pendant Ying. II. 05 (Pl. CX), having the shape of a ten-pointed star with five glass ` jewels ' set round a central boss. The work has an appearance of antiquity.
That the walled enclosure is of pre-Muhammadan origin may be considered certain. Its