reconquest of the latter was completed by the subjection of Yen-ch`i, or Kara-shahr, in the north-east.17
During my four days' stay at Yarkand I was as unsuccessful as on my previous visits in
obtaining information as to any old sites. The position of ` the old city' of Yarkand, from which Aba Bakr the tyrant, was, according to Mirza I,laidar's testimony, believed to have excavated great riches,18 still remains unidentified. The intensive cultivation proceeding all over this great and flourishing oasis is likely to have buried all ancient remains under deep layers of alluvial deposit, as explained in my former Report with reference to Yötkan, the site of the old Khotan capital.l9 But the probability that the present Yarkand city is itself built on ground occupied by
the ancient capital of So-ch'ê, or near it, is strengthened by a large find of coins (which was made some time before my stay) near the Ya-mên of the District Magistrate in the ` new city ' of Yarkand. From it Pên Ta jên, the learned Amban then in charge of the District, was kind enough to offer me a representative set of ten specimens. These comprise, as the inventory list prepared by Mr. J. Allan shows,2Ö one copper coin with the legend K`ai-yüan as issued by Kao Tsu, the first emperor of the Tang dynasty (A. D. 618-27), and by his successors for over a century, and various issues of the Sung dynasty, ranging from A. D. 990 to I I It. The composition of this hoard seems to bear evidence that the site of the present Yarkand city was already occupied towards the close of the pre-Muhammadan period.
For my move from Yarkand to Karghalik, the next étape southward, I used on this journey a new and somewhat devious route which took me along a hitherto unsurveyed portion of the Tiznaf River. In Chapter XI of my personal narrative I have fully described it. Its choice was due to the wish to satisfy my archaeological conscience by a visit to the site of Kizil jai, from which about two years earlier some fairly well-preserved manuscript records in Uigur script had been brought to Sir G. Macartney. They have since been under examination by Dr. E. Denison Ross, C.I.E.,21 who was able to show me the originals on my passage through Calcutta in December, 1908. The antiquarian results of the excursion which I made on July 4 to the site from Bagh jigda, a small village on newly settled land near the right bank of the Tiznaf River (Map No. i i), proved scanty. After passing for about five miles northward through recently reclaimed cultivation, we entered an area of luxuriant scrub and jungle extending between the Tiznaf River and the moving sands on the east, and after another four miles or so reached a spot marked by a group of large wild poplars and known as 1 Kizil-jai Mazar'. About half a mile north-west of it the discoverer of the Uigur records, a Bagh-jigda tenant, Ibrahim by name, pointed out as their provenance one of those curious tamarisk-covered sand-cones which are a typical feature on the edge of the Taklamakan desert.22 On closer examination I ascertained that more than ten years had passed between the discovery and the presentation of the documents to Sir G. Macartney. Ibrahim declared that he had come upon one small packet on the top close to the surface, while searching for dead wood or kötek, and then upon two more when clearing the sand some two yards lower, the total height of the sand-cone above the adjacent ground being about fourteen feet. How the several documents, all apparently of the same period, were found at greatly varying depths, seems difficult to explain, considering that these sand-cones are of relatively slow growth,23 and that the difference in levels he named would indicate probably an
17 See Chavannes, Trois glnlraux, pp. 25 sqq. (T`oungpao, 1906, p. 233).
18 Cf. Tàrikh-i-Rashidi, pp. 256 n., 257, 296.
'V See Ancient Khotan, i. pp. 196 sqq. 28 Cf. App. B below.
21 Dr. Ross has in preparation a full account of them.
22 As to the origin and growth of these interesting formations, cf. Ancient Khotan, i. pp. 1 zo, 458.
23 For definite archaeological evidence on this point, furnished by a ruined shrine at the Farhad Beg Site, see below.