272 ANCIENT SITES OF ENDERE [Chap. VII
the night in the narrow waterless plain known as Yantak-chaval. Next day, instead of taking the shepherd track towards Endere Tarim followed by me in 1901, our guides led us eastwards for
about six miles until we struck the edge of the Bilé1-konghan depression. It presented itself here
as a wide area covered with high and isolated tamarisk-cones and fairly thick Toghrak jungle growing amidst low dunes. In my Personal Narrative I have described the trouble and time which
the search for the ruins cost us on this deceptive ground and how at last, as I led my big convoy for safety's sake towards the Endere River, I stumbled upon the ` Kône-shahr' quite close to the course we were steering.' Its position, as seen in sheet No. 40 of the map, proved subsequently to be about five miles to the west of the western terminal bed of the Endere River which in 1901 I had found dry, but to which the errant river had now returned again.
At first sight the appearance of the ruins was far more suggestive of an ` old town ' than that of the widely scattered structural remains or pottery-strewn ` Tati ' sites to which the term is indis-
criminately applied by all people in the Tarim Basin. Here in the midst of a plain comparatively
clear of tamarisk growth and overrun only by slight dunes, I found a roughly oval area enclosed by a clay rampart still traceable at most points and crowded with dwellings rudely constructed but on
the whole well preserved (see Fig. 73). Subsequent examination showed that the irregular oval within this circumvallation measured about 263 yards on its greatest axis, from north-east to southwest, and about 210 yards across where widest (see plan, Plate 19).
The houses, as clearly seen in the photographs (see also Fig. 76), generally showed roughly built walls of clay, either merely stamped or else with occasional layers of small coarse bricks. The
roofs were invariably constructed of rough Toghrak trunks overlaid with brushwood and earth, and
had survived in many cases more or less intact. Some three or four houses had walls constructed of reed bundles and plaster, with posts of unhewn Toghraks taking the place of timber framework.
Two small single-roomed structures, one of them burned down to within a few feet from the ground, were built wholly of Toghrak trunks left in the round and set up in palisade fashion. These rough dwellings, whether consisting of a single room or several grouped side by side, were scattered over the enclosed area without the slightest sign of arrangement, as the plan shows. But somehow the largest seemed to keep close to the encircling wall.
This circumvallation was as crude as the structures within or the ground-plan itself. It consisted of a mud rampart about sixteen feet broad at the bottom, carrying at a height of about eight feet a platform of rush layers fixed on roughly hewn beams. On the outside edge of the platform there rose a parapet, about one and a half feet thick, built of irregular lumps of clay. Portions of this, standing to a height of four feet, survived near the gate, as seen in the photograph (Fig. 84). This gate, the only one I could trace, stood on what may be considered the south-east face of the enclosure. It was eleven feet wide, seventeen in depth, and still retained its massive door folds of roughly hewn Toghrak beams in position. The outer face of the rampart showed stamped mud, with intervening irregular courses of shapeless clay lumps serving for bricks. The whole looked greatly inferior in constructive strength and appearance to the rampart of the Tang fort of the Endere Site. Almost the whole of this circumvallation could be traced with ease except on a portion of the north face where tamarisk-cones rising about fourteen to sixteen feet above the original level covered it. Wherever clear of sand the line of wall showed plain marks of having been subjected to fire. Charred fragments of wood lay on or near the rampart, and its clay was reddened by burning.
A rapid preliminary inspection sufficed to convince me that the remains of this deserted village' —for as such it could properly be designated in spite of its modest rampart—could not claim any
° See Desert Cathay, i. p. 303.