of pre-Muhammadan origin, is likely to have been occupied intermittently also during later periods.19 A large cave, the position of which was pointed out to nie at the foot of the hills over-
looking the debouchure of the Ushak-tal stream, about three miles to the north-west of the village, was said to have been examined by Professor Grünwedel. The short winter day left no time to see it on my way to Chokkur.
The ruin reported at the last-named place (Map No. 48. D. 3) proved to be a small walled enclosure of a type similar to that at Chong-köl. Towers of stamped clay, mostly about 25 feet by 19,
strengthened the curtains at short intervals. The walls of the latter seemed to have had a thickness of only 7 feet, covered now on both slopes with masses of fallen clay. No structural remains bad
survived in the interior, and diggings carried on for manuring soil showed that whatever buildings it
might once have contained had crumbled away into mere earth. A mound of earth, about 5o feet in diameter, seemed to mark the position of some central structure. The ruined fort stands in the
middle of a small but well-irrigated area of cultivation, and around it stretch luxuriant Toghrak jungle and grazing. It was through ground like this, or equally luxuriant reed-beds where the shores of Lake Baghrash lay nearer, that the thirty miles' march led me to the town of Kara-shahr on December 8.
From this place I visited the ruins of Baghdad-shahri, the only old site in the neighbourhood of which information was obtainable, apart from the ` Ming-oi ' near Shbrchuk. The way led along
the high road towards Korla which crosses the wide river-bed about half a mile from the town and
then passes a wide steppe with fertile soil and scattered patches of new cultivation. Water for more canals is available in plenty, and only want of population prevents an oasis being created on the
river's right bank far bigger than the one adjoining the present town. The Baghdad-shahri site, about nine miles distant from the latter, lies close to the eastern edge of a long but narrow stretch of cultivation of which the roadside station of Danzil is the centre (Map No. 49. B. 1). Its remains consist of a large oblong circumvallation, which undoubtedly marks the position of a town of importance.
The corners of the town walls are approximately orientated ; the north-west face measures about 1,030 yards, that to the south-west about 935 yards. The walls, everywhere badly decayed
except near the west corner, appear to have been built throughout of layers of stamped clay 3 inches in
height and to have had a thickness of aboùt 9 feet. They rest on a broad earth rampart, rising 12-15 feet above the adjoining ground. Apart from a large mound of stamped clay of uncertain
character, rising within the north corner of the circumvallation to a height of 25 feet or so and about
20 yards across on its flat top, the interior of the town retains no structural remains whatsoever. The whole of it is occupied by salt-encrusted low ridges and hillocks, with a large earth mound of the
same shapeless appearance a short distance off the middle of the south-west face. Remains of a square wall, apparently of late date, were traceable-on its top. There was found on the surface a well-preserved coin with the nien-hao Chien-chung (A.D. 780-4), together with plenty of coarse pottery on the slopes. Fragments of a Tang coin, apparently a Kai ygan piece, were picked up on the top of the previously mentioned mound.
A look at the salt-permeated soil sufficed to show that percolation from below and subaerial moisture must have destroyed here all remains except those of the hardest water-resisting materials.
For systematic excavations such a site could hold out little hope, and the mere fact that no ` treasure-
seeking ' was practised at the site by any of the inhabitants of the hamlets close by confirmed this impression. But even in the absence of direct archaeological evidence there is much to support the
belief that the walled town of Baghdad-shahri marks the site of the Kara-shahr capital, at least as it existed in Tang times. From the itinerary of the T. ang Annals discussed above we have seen that
19 For specimens of pottery, including one obviously Chinese, with transparent celadon green, see the List below.