THROUGH THE TAKLAMAKAN TO KARA-DONG 1239
of it. Here, as in the case of the ancient oases south of the Taklamakân, the question of ` desiccation ' is bound to force itself upon the attention of the antiquarian student.
On January 25 I left Kucha town, and on the following day reached Shahyar, the furthest settlement to the south and now the head-quarters of a separate district. A day's halt had to suffice for the last preparations, which included the raising of a month and a half's supplies for my relatively large party, and after moving south across the frozen Tarim we left the last shepherds' huts behind on January 30. The trying and, as experience showed, distinctly hazardous journey across high dunes for a marching distance of close on i 8o miles has been fully described in my Personal Narrative.2 The detailed account given there of the physical features of the true ` Sea of Sand' we traversed renders it unnecessary to record here the manifold observations of geographical interest which this desert crossing enabled me to make, and which the Maps Nos. 3o, 35, 36 will help to illustrate in their topographical aspects. It must suffice here to indicate those few points which have a bearing, direct or indirect, on the question whether this vast dune-covered area has been visited by man during historical or prehistorical times.3
From the Achchik-darya (Map No. 35. B. 2), where we crossed the last traceable dry bed occasionally receiving water from the Tarim, to our Camp 3i 7 (Map No. 35. B. 4), a direct distance of about 28 miles, the east-west bearing of the high sand ridges, or ` Dawans', crossed in succession, lay approximately parallel to the Tarim. This furnished a clear indication that the system of drift-sand formations on this ground was mainly influenced by the great riverine belt that it borders. The patches of bare clay crossed here in places, however, did not yield any stone-age or other relics such as would mark ground occupied by small temporary settlements of hunters or herdsmen in prehistoric times.
From Camp 317 to the south, a distinct change made itself noticeable in the configuration of Change in
the high ridges of sand as distinguished from individual dunes. The ' Dawans ', too distant from direction of
our track for any safe estimate of height, but certainly rising in places up to 150 feet or more and continuous for miles, now invariably bore from north-north-east to south-south-west. This is, as the general map will show at a glance, exactly the general bearing of the course of the lower Keriya River, which itself is determined by the configuration of the slope contours in this part of the great desert basin. This direction of the Dawans is an unmistakable sign that the Keriya River once reached so far ; for it is a constant observation made everywhere, both in the Taklamakân and in the Lop region, that the drift-sand near river-beds, whether dead or still carrying water, is heaped up in ` Dawans ' parallel to the latter.'
A striking confirmation is supplied by the fact that in the groups of living Toghraks which were met with at short intervals for a direct distance of over 13 miles south of Camp 317, as seen in Map No. i5. B. 4, the trees were found everywhere growing in lines roughly directed from northeast to south-west or north to south. The tendency of wild poplars and other trees in the riverine jungle belts of the Tarim Basin to range themselves close to the river banks or parallel to them is
a well-established fact.' I may add that the identical bearing was observed also in the rows of
Toghraks, living or more frequently dead, that were met with at rare intervals on the three marches
south of Camp 318. Water, which up to this halting-place was obtained from shallow wells dug in
Interest of desert crossing.
parallel to Tarim course.
2 See Deserl Cathay, ii. pp. 379-405.
3 I may conveniently note here that about two miles to the south of the last shepherds' huts at Samsak-daryasi (Map No. 35. B. 5) I found traces of old fields, probably dating from some intermittent cultivation carried on many years ago on ground reached by occasional floods from the Tarim.
About 2A miles to the south-east of these huts I was shown a small mound known as Kizil-dong where some debris of rough pottery indicated occupation of uncertain date.
4 Cf. above, pp. 455 sq. ; Hedin, Central Asia, i. p. 363.
5 Cf. e.g. Hedin, Reisen in Z.A., pp. 54 sq.; above,
PP. 355, 452.