ùndesirable subjects or otherwise useless captives. In that case we could easily understand the improvised nature of the quarters, their wholly unusual crowding, and finally that peculiar roughness of construction which struck me at the time, and which I do not remember to have seen anywhere in the Khotan region apart from mere shepherds' huts.
In conclusion a passing remark about that ever-present cause of physical change in these parts, desiccation. That it has been at work on the Endere River as elsewhere there can be no doubt, and Prof. Huntington has discussed the indications of it with his usual lucidity." But there seems no reason for seeking any connexion between this slow-working cause and the abandonment of the fortified village ; for, on the one hand, there is no proof that here cultivation was ever seriously started, and on the other, there is the fact that even at the present day a small agricultural settlement exists at Endere Tarim, some ten miles further down the river's present course. I could not spare time for a visit to this colony. But its principal settlers joined me while at Bilél-konghan and supplied interesting information about local conditions.
According to their statement the troubles against which thi§ modern representative of the terminal oasis of Endere struggles, arise from the frequent shiftings of the main channel. These are caused by the heavy ak-su or summer floods and make maintenance of their irrigation canal difficult. Confirmation of their statement was supplied when, on my subsequent march to the Endere River, as described in my Personal Narrative,12 I found that the river had completely reverted to the bed which in 1901 was known as the Kône-daryd.13 It was curious, too, to hear the settlers discuss an engineering project from which they hoped for full protection against all such troubles, but which they recognized to be wholly beyond their present resources. It was to keep the existing bed exclusively for the kara-su, or spring-fed water-supply of the river, and to divert the flood-water, or ak-sac, by means of a big dam built somewhere higher up in the vicinity of the Endere ruins into the wide plain about the abandoned old village. The amount of kara-su permanently available was declared to be ample for a much larger settlement than the present Endere Tarim. It was, in fact, estimated by my Niya men at ten Tâsh', i. e. sufficient to move ten millstones. Here, too, only prolonged and careful study by an irrigation expert can furnish definite guidance as to irrigation possibilities in the past and present.
SECTION II.—EXCAVATIONS AROUND AND WITHIN TANG FORT OF ENDERE
On the morning of November 8 I sighted once more the high Stûpa ruin which forms the conspicuous landmark of the ancient site of Endere first visited by me in 1901. Then want of time had obliged me to confine my exploration to the Stûpa and to the interior of the ruined fort about one mile to the south-east (Fig. 78),1 and even within the latter to leave uncleared some apartments which were too deeply covered with drift sand. So my ` archaeological conscience ' was eased when once more my camp was pitched by its side. A rapid inspection assured me that the remains of the little temple in the centre (see plan, Plate 2o), which had then disclosed interesting remains of manuscripts in Br .hmi script and Tibetan and a dated Chinese sgraffito of importance, had not suffered in the interval. My satisfaction was great when, on superficially clearing the 'sand from the north-west corner of the temple cella, the sgraffito again came to light, just as I had seen it before. Thus
l' See Pulse of Asia, pp. 213 sqq. That some of the archaeological and historical facts adduced have to be differently interpreted does not affect the main argument.
12 See Desert Cathay, i. pp. 307 sq.
13 Cf. Ancient Khotan, i. p. 420.
1 Cf. for previous work at the site, Ancient Khotan, i. pp. 42 x sqq.