side, and near it, on the east, a mass of timber débris which seemed to mark the place once occupied by a structure at least sixty feet long. The heavy but badly-fissured beams lay now on hard clay, swept absolutely clear of any other remains. The little mound rested on a foundation of tamarisk fascines, and its brickwork was reinforced by two layers of reeds embedded at an interval of one foot. About forty-five yards to the NW. there was another plot of ground strewn with the débris of big pieces of timber, which must have once belonged to the framework of a large dwelling built of timber and wattle. Now they lay on the hard bare clay soil which erosion had pared down to a level about ten feet lower than that marked by the foundation of the mound close by.
It was on returning to camp from this last small Tati' that I particularly noticed the unmistakable old river-bed we had to cross, marked by the dead Toghraks and tamarisk-cones lining the banks. The winding bed was from fifteen to twenty feet deep, with a width of about 150 to 180 feet. The banks were very steep and remarkably well preserved. The fact that the general direction of the bed lay towards the north-east, as seen in Plate 22, and thus nearly coincided with that of the prevailing wind, may have helped to protect it. It was curious to note, as I repeatedly observed elsewhere, that the fairly level bottom of the bed showed scarcely any signs of wind-erosion, though the ground on either bank was cut up into a perfect maze of Yardangs and trenches. The bed seemed to connect with the one we had crossed when approaching the ruined area from the south on December 17.
Between the scanty remains just described and the ruined station south, as well as on the other sides of the latter for a distance of about a mile, the ground showed here and there patches covered with potsherds of the same general type as prevailed within the walled enclosure. Seeing how even substantial structures had fared under the constant scouring by wind-driven sand, it was easy to understand why humbler abodes, probably built with mud walls such as I had last seen at Charkhlik, or perhaps consisting of mere reed-huts like those of Abdal, had completely disappeared without leaving any trace but the pottery débris of these little ` Tatis '. But of the life once led here and of the traffic once passing over this ground the bare eroded surface had retained plentiful relics in the shape of small objects in stone, metal, and other hard materials. It was a happy hunting-ground for honest Jasvant Singh, the Surveyor's cook, and the labourers, when they were not kept busy by digging, and the ` finds ' they brought me made up the considerable collection shown in the descriptive list below (L.A. ooi-ooi 77). Of some of the more interesting finds brief notice still remains to be taken here.
Among the specimens of potsherds the grey, mat-marked pieces, L.A. 002,'00146 (Plate XXXVI), may be mentioned as corresponding exactly to Chinese ware of Han times found along the Tun-huang Limes.8 The fragments retaining a deep blue-green glaze, L.A. oo3, 007, 00100, are more likely to represent local ware, as are certainly the specimens of the predominant poorly levigated pottery, L.A. 0o1, 009, Coto, including the lamp L.A. oo6 (Plate xxxVI). The abundance of small objects in bronze at this site is very striking, and far greater than comparison with the contemporary Niya Site might have led one to expect.9 I think that we may recognize in it, as well as in the disproportionately greater number of coins found, an indication of the busy traffic which the great Chinese trade-route brought to the ancient station. The bronze arrow-heads, L.A. 0017, 0069, 0082, illustrated in Plate xxtX, are certainly of the same type as the ancient Chinese ammunition' found at the watch-stations of the Limes. The numerous finds of pieces from Chinese bronze mirrors, L.A. 0027-29 (Plate xXIX), 0078, 0084, etc., point to easy import from the East. The winged dragon shown in relievo on L.A. 00113 (Plate XXIX) closely resembles the ornamentation of N. 009 (Ancient Khotan, ii. Plate LXXIV). Bronze rings, L.A.
e See above, p. 384. 9 Cf. Ancient Khotan, i. pp. 414 sqq.