Sec. i] TO THE TARIM AND THE DELTA OF THE CHARCHAN-DARYA 453
tamarisk branches and by a framework of Toghrak beams. The width of the rampart was fourteen feet on the top, and at the base, now mostly covered by sand, about twenty-nine feet. The diameter of the little stronghold, measured from the centre line of the rampart, was approximately i 32 feet. The entrance must have been at the south ; for there a gate passage, six feet wide, was marked by two rows of massive Toghrak posts, which came to light on clearing the southern segment of the rampart. These posts, four on each side, must have served to carry the roof of the gateway, and probably also to secure the timber revetment of its sides.5 Their top portions for about three to four feet were in fair preservation, except for the charred ends, giving evidence of a fire ; but lower down the wood had rotted away owing to the effect of moisture. As shown by the position of the posts, the gateway passing through the rampart must have had a length of at least twenty-three feet.
The interior of the circumvallation was fairly clear of drift-sand, and near the centre some withered Toghrak beams were found just below the surface. But as on digging down beneath them to a depth of about four feet we came upon moist sand, it was clear that the remains of any structure which might once have stood here must have decayed completely. The north-east portion of the rampart was partly covered by the slope of a tamarisk-cone which rose some fifteen feet above its present top, as seen in Fig. 89. Referring to what has been explained above as to the tamarisk-cone overtopping the ancient fort found south of the Endere Site,6 we may safely assume that the fact of this cone at Merdek-tim having attained a height of at least twenty-five feet above the level on which the circumvallation was built in itself proves considerable age for the latter.
But a closer search of the rampart soon revealed more definite evidence of its antiquity. On clearing away the sand which covered the faces of the masonry portion, I found that it was composed of large sun-dried bricks very similar in dimensions to most of those which we had seen in the ruined structures of the Lou-lan Site. I distinguished two sizes of bricks, one measuring i8 by 11 inches, with a thickness of 4 inches, and another 14 by io inches, with a thickness of 31- inches. In view of such close agreement, approximately contemporary origin necessarily suggested itself, and this conclusion was strikingly confirmed by a succession of finds of Chinese copper coins which were picked up in various places on the top of the rampart. All belong to types current during Later Han times ; two bear the legend Huo-cle ian, first introduced by Wang Mang (A. D. 9-22), and the other four are Wu-chu pieces in a clipped condition.
None of the usual débris of perishable materials, within or around the fort, could have survived subsoil moisture. The same cause, vicinity of water, explained the complete absence of wind-eroded ground near the ruin and our consequent failure to trace any of the usual ancient débris of hard materials, such as potsherds, small metal objects, etc. Nor did the search made around that day and on the following morning reveal the survival of any other structural remains. Yet the evidence already secured left no reasonable doubt that this small fortified post must have been occupied during the early centuries of our era and perhaps down to the same period as the Lou-lan Site.
Insignificant as is the ruin itself, its date, as thus approximately determined, invests it with a distinct antiquarian and geographical interest. Its existence at this place proves that a branch of the Tarim—whether large or small we have no means to decide—must already, during the earliest period of Chinese control of the Tarim Basin, have flowed close to the present line of the
same period as its ruins.
For a similar, but more elaborate gateway, found at the fortified station or Sarai at Kara-dong, cf. Ancient Khotan,
i. pp. 447 sq. ; Fig. 53.
6 See above, pp. 283 sq.