Sec. iv] DESERT SITES NORTH OF TUMSHUK AND MARAL-BASHI 1307
circumvallation (Fig. 338). The clay rampart enclosing it measured about 110 yards on each face, and still rose on the west side, where it was best preserved, about 12 feet above the nearest bare ground. But the close approach of dunes on all sides made it difficult to ascertain the original ground-level. The rampart, badly broken as it was in most places by wind-erosion, yet showed that its stamped clay was strengthened by layers of brushwood. The watch-tower, though much decayed too, stood to a height of about 35 feet from what seemed the ground-level, as marked by the eastern rampart. From measurements at the east face of the tower, where its foot was less encumbered by débris than elsewhere, the base appears to have been about 25 feet square. The tower was built of sun-dried bricks measuring either i 1" x 10" x 3" or 14" x 8" x 4". The masonry looked rough, with layers of mud plaster but no brushwood or the like between the courses.
The general appearance of these remains and of the eroded ground all round left no doubt about
the antiquity of the ruined fort, and the finds of small objects in its immediate vicinity fully confirmed this. Just outside the north face of the enclosure there were picked up eleven much-worn Wu-chu and ` goose-eye ' coins lying close together, evidently as they had dropped from a string. Among other Chinese coins found near the circumvallation three are Wu-ciiu pieces and one a Tang coin with the legend ICai yüan. Their evidence, indicating occupation from an early period down to Tang times, was supported by that of the coins which were brought to me at Kelpin, avowedly as having been found at the Chong-tim site on the ` treasure-seekers" latest visit. Of these, twelve were Tang pieces, the latest bearing the nien-hao Ta-li (A. D. 766-80), and one a Huo-ch`üan coin of Wang Mang (A. D. 14-19).' The pottery fragments which lay thickly on all patches of eroded ground about the ruined fort were mostly of fine red clay and, as seen from the specimens described in the List below, bore a distinctly old look. One piece, Chong-tim. 002, with mottled green glaze, is considered by Mr. Hobson to be of Chinese Han type ; another, Chong-tim. 005, with floral ornament in low relief under green glaze, is held to be akin to Near-Eastern, probàbly Persian, ware dating from the second to the ninth century A. D. Among the miscellaneous small relics in bronze, glass, paste, and wood (Chong-tim. 006-47) picked up at the site there is none that would necessitate a lower chronological limit for the abandonment of the site than that of late Tang times.
Within the circumvallation no structural remains of any sort survived. But inside the west
rampart there stretched refuse-heaps, from 3 to 4 feet. high, for a distance of some 25 yards. These I had completely cleared to the ground, but their yield was very scanty. Within the masses of horse and cattle dung, straw, and brushwood, there turned up only a few rags of cotton fabrics (Chong-tim. 0049. a, b), bits of felt, rope, and the like. I may note that not a single piece of paper was unearthed. It was, however, significant to find numerous small pieces of sulphur scattered at different points of the enclosure. They furnish conclusive proof that the sulphur mine on the eastern slope of the Kudughun Peak (Map No. 14. D. 5), which is now regularly worked by Kelpin people, or some similar deposit in the hills northward was already being exploited during the occupation of the little fort. Its character, and particularly the high and solid watch-tower, seemed
to point distinctly to the ruin being that of a station on an ancient route line leading from Ak-su to Kâshgar. It appears only natural that such a roadside station, lying within 10 miles or so of the sulphur mine in a straight line, should have formed a regular place of transit for its products.
I may explain here at once that the Chong-tim site has proved to be situated on the most direct line connecting Chilan, where the present road from Ak-su leaves the foot of the hills, with what my explorations of October, 1913, have shown to be its ancient continuation leading north of Marâlbâshi and along the foot of the mountains to Kâshgar. It would be impossible here to give the reasons upon which this statement is based without fully discussing the results of my topographical ' Cf. Appendix B for all numismatic details.