462 KARA-DONG AND THE SEARCH FOR HSÜAN-TSANG'S P`I-MO [Chap. XIII
of the previous day, who had been allowed to keep away during the night for the sake of fresh efforts, turned up after midday with the news that he had found U lûgh-Ziârat at last. So accompanied by him and old Turdi I started back again to the point where the caravan had been halted on the previous day. Ulûgh-Ziârat proved to be within only two miles of it north-westwards, but effectually hidden behind a series of high tamarisk-covered sand-cones. It was a small eroded plain about half a mile long from north to south, and slightly less broad, resembling in all its features the Uzun-Tati site, and covered like it with plentiful débris of undecorated but well-made old pottery. About its middle rose a loess-bank with what are supposed to be tombs of fifteen or sixteen saints. On its top a small collection of staves, rudely fashioned of tamarisk, clearly marked the spot as still receiving worship. I could not spare time for closer examination of these alleged tombs, as I was anxious to visit before nightfall the Sipil' or fortification reported further ahead. After crossing dunes for an approximate direct distance of it miles to the east-north-east, I reached it in the middle of a relatively open sandy plain covered with tamarisk bushes and numerous young Toghrak trees.
The ` Sipil ' proved to consist of the comparatively well-preserved remains of a small fort, built in the form of an oval, having its longer axis from north to south. The greatest length of the interior was about 48o ft., its maximum breadth about 348 ft. The wall, about 7I ft. thick at the base, was constructed of stamped loess with layers of rushes placed horizontally at varying levels to give it consistency. At a height of about 9 ft. from the present ground-level it bore a parapet which seemed to have originally been about 5 ft. 8 in. high, with a thickness of 3 ft. This, however, was broken in numerous places ; behind it ran a platform about 5 ft. broad. No loop-holes were visible. The interior of this small fort, approximately approaching in size that of the Endere Fort, showed no remains whatever of buildings, though the sand within it could only be from 3 to 6 ft. high, and in several places left the foot of the walls clear. Except for a gap in the east face, which probably marks the position of the gate, the wall could be followed all round, no portion of it having completely fallen. I noted neither inside nor outside any pottery fragments. In the absence of other remains I was unable to form any definite opinion as to the date of this circumvallation, but its relatively good preservation, its shape different from that of the old forts of Endere and Ak-sipil4, and the total absence of brickwork seemed all to point to its having been constructed during Muhammadan times. Seeing that the position it occupies lies nearer to the present limits of cultivation, and must even now have subsoil water at a relatively small depth, it appears very improbable that the fort could have been abandoned earlier than Uzun-Tati. I left this ruin at nightfall, and was met on the way while returning to camp by the Beg of Gulakhma. A close examination of the men he had brought with him showed that the remains I had already examined were all that were known to them.
That these remains of Uzun-Tati and Ulûgh-Ziârat have a strong claim to be recognized as marking the position of Hsüan-tsang's Fi-mo and Sung Yün's Han-mo can, I believe, be proved by several convincing arguments. In the first place, we find their position in exact agreement with the distance of 33o li, i. e. a little over three marches, and with the eastern bearing from Yôtkan which the Hsi yii-chi indicates for P`i-mo. The coins and porcelain fragments found by me at Uzun-Tati make it quite certain that the site was occupied for centuries after Hsüan-tsang's time, while, on the other hand, there is nothing to preclude us from believing that it existed already long before the pilgrim's visit.