28o ANCIENT SITES OF ENDERE [Chap. VII
adjoined by a sitting bench which had doubtless served as a specially warm corner. Otherwise these rooms were found entirely empty. Since the fire-places here, as at all other sites, lacked chimneys, the trouble from smoke must have been considerable and in curious contrast to the standards of comfort otherwise observed.
It deserves to be noted that most of the timber used in these quarters of the Endere fort, as well as the beams and posts found exposed in plenty about the courts and the gate (Fig. 78), were of the wood of the cultivated poplar or Eleagnus (Jigda). The use of Toghrak wood seemed rare. This is a plain proof that by the time when the fort was erected cultivation must have been resumed for some time past in this vicinity. Yet the remains of dead fruit- or garden-trees, as far as I could trace them, were distinctly scanty. On the much-eroded ground north of the Tang fort where, as we shall see, the main portion of the earlier settlement must be located, I found none. To the south of the fort traces of old arbours or orchards were also relatively rare and were found chiefly along the left bank of the ancient river-bed which was traced skirting the site on the east. This rarity of dead tree trunks can, I think, be accounted for by two circumstances. In the first place, those belonging to the more ancient settlement were bound to have been carried away for fuel when the abandoned oasis was partially reoccupied in Tang times. In the second place, a similar fate is likely to have befallen most of the trees planted during this later period as soon as the Tang fort and any homesteads near it were deserted in turn ; for the locality is likely to have remained for centuries a usual halting-place for caravans passing by the great route close by, and dead trees would always be in demand for camp-fires, etc.
SECTION III.—SURVEYS OF EARLIER REMAINS AT ENDERE
While the diggers under the supervision of Chiang Szû-yeh and Naik Ram Singh were kept hard at work clearing the remaining quarters within the Tang fort, I found time to survey remains farther away, both to the north and south, about which information had reached me, partly through brief notes of Prof. Huntington and partly through my new guides from Endere Tarim. Already on my first arrival at the site in 1901 I had noticed that about half a mile to the north-east of the large ruined Stûpa there were numerous shapeless mounds of clay rising between the low dunes. Misled by plain indications of the great depth to which the ground near the Stûpa had been eroded,' I had then taken them from a distance for mere erosion ` witnesses ', and want of time had kept me from any closer survey.
The photograph, reproduced in Fig. 8o, of one of these clay terraces which I examined on November 8 at about forty yards' distance to the south-east of the Stûpa, will help to explain the deception. Up to a height of over fifteen feet this mound consisted of nothing but natural loess eroded by the winds, and it needed very close examination to show that the little clay tower which is seen on the left rising about eight feet higher, is of artificial origin, probably the last remains of a small Stûpa. In view of the depth of erosion around, and of the size of some brick fragments showing the characteristic thickness of three and a half inches, it may be safely assumed that this tiny ruin, too, goes back to the time of the ` Tu-huo-lo ' settlement.
On proceeding to the cluster of clay mounds which Fig. 8o shows in the distant background, I soon ascertained that they had originally formed part of a large walled enclosure now half-buried amidst dunes and decayed by erosion almost beyond recognition. Tokhta Muhammad Khwaja, the intelligent settler of Endere Tarim, had been right in talking of these remains as a Sipil'. By