Sec. iii] SURVEYS OF EARLIER REMAINS AT ENDERE 281
closely observing the alignment of the fragments of wall still traceable here and there it was possible Ancient cir-
approximately to determine the area once enclosed, an oblong about 540 feet from north to south cumvallationnortheast
and about 34o feet across. The circumvallation could best be followed on the north and west ; on of Stupa.
the latter face, of which Fig. 81 shows the northern portion, the line of the wall could be measured for a distance of about 44o feet by the continuous chain of fragments which the process of erosion had not yet destroyed. On the east face the fragments of wall were much fewer, though extending a little further south than on the opposite face ; the wall to the south had practically disappeared altogether. In view of the observations subsequently made as to the effect of wind erosion on the walled enclosures of sites surveyed further east, at ` Lou-lan', An-hsi, Chlao-tzû, etc., I think it safe to conclude from the above conditions that the destructive force of the wind and wind-driven sand must here have made itself felt with particular intensity and continuity from the north-east.
Wind erosion had evidently first breached the outer walls and then scoured the remains of what- Effects of
ever buildings the interior once contained. In the south-east corner of the walled area remains wind
of massive walls seemed to indicate the former existence of a separate enclosure, measuring about 170 feet from east to west and 110 feet across. Whether these interior walls belonged to a large set of quarters as found in E. 111 or to a kind of citadel could no longer be ascertained. The only other structural remains traceable in the interior consisted of a small enclosure of which some remnants of the wall survived close to the highest fragment of the eastern wall face, seen in Fig. 83. It was significant that what had helped to protect them was a layer of consolidated sheep-dung, the relic of a later period when, perhaps, the least ruined structure within the abandoned ` old town ' had served as a shelter for herdsmen. Elsewhere the ground within, where not covered by broad dunes, showed nothing but small pottery débris, mostly black or dark brown.
The enclosing walls were formed at their base by a rampart of stamped clay, about thirty to Constructhirty-five feet thick. This rampart appeared to have borne a superstructure built with courses Lion of en-
of large sun-dried bricks set in thick layers of clay, as seen in Fig. 83. But there was little walls. closing
regularity in the construction, or else repairs had been frequent. Usually the courses of bricks were single, but in some places two or three successive courses had been used in the same way. Frequently the place of proper bricks was taken by shapeless lumps of fine clay. What is important to note is that the bricks throughout showed a fairly uniform size, 19 to 20 inches by 13 to 14, with a thickness of 32 to 4 inches. The same pattern was found in the ruins E. vt, v1I, which were proved by epigraphical finds to belong to the earlier settlement, as well as in the southernmost ruins to be described below. But the bricks used in the buildings and the wall parapet of the Tang fort differed distinctly in size.'
Owing to the lesser or greater effect of wind erosion in different positions the height of the Remains of actual remains of the wall above the pottery-strewn level of the interior varied greatly, from about ten earlier town. to twenty-six feet. With the high tamarisk-cones rising above the north front of the circumvallation and the broad dunes filling a considerable portion of the interior, the whole made up a weird picture of desolation which even the openness of the vistas across the bare eroded areas all round did not help to relieve. For excavation there seemed no scope left here. But that these ruins, like those described south of the Tang fort, went back to the period of the earlier settlement and were among those Tu-huo-lo towns ' which Hsüan-tsang had found completely deserted, I could not reasonably doubt. Apart from the very definite indication furnished by the size of the bricks, there was striking evidence in the condition of the walls. Though much more massive in construction than those of the rang fort, their decay had advanced so much further as to be explicable only on