Sec. iii] A HILL RANGE IN THE TAKLAMAKAN 85
along their tops with a view to save the constant ups and downs so trying to laden camels and thus to reduce strain at the cost of distance would in any case have been useless, since those tops were invariably crowned by huge dunes very steep-edged on their lee side and consequently such as camels could not possibly skirt or ` traverse '. Nor was relief to be sought in the depressions or valleys ; for in proportion as the ` Dawâ ns ' rose to greater height these depressions at their bottom became more and more smothered with dunes. At first there were still a few little patches of eroded level ground at the bottom of the valleys that we crossed after four and five and a half miles, respectively, from Camp xxvi, and near them a little living tamarisk scrub (Fig. 87). But beyond this point there was no flat ground to be found even in the valleys, and progress with the heavily laden camels became painfully slow. Careful levels taken along our track a couple of miles before reaching Camp xxvii showed an aggregate ascent of more than 35o feet over a single mile's marching distance as measured by cyclometer, with corresponding descents even more trying to the heavily laden animals. Measurements similarly taken over different average stretches on that day's march proved that the windings involved by the only practicable route line always added from 3o to 4o per cent. of marching distance to the actual progress in a straight line as marked on the map. No wonder that the hired camels already began to show serious signs of exhaustion towards the close of that day.
But an interesting discovery awaited us at its end. As we descended into a valley running as usual ENE.-WSW., and whose bottom lay some 25o feet below the point where we had crossed the adjoining ` Dawân ', I noticed the edge of a wind-eroded clay terrace protruding from the side of a dune. By the steep Yardang-like side of this terrace there ran a narrow trench, dug out by wind-erosion to a depth of about 5o feet and its lowest point showing a few small living plants. Here Kâsim Akhûn, returning during the night, succeeded in striking scanty water after digging down about five feet. Narrow patches of the same hard grey clay reappeared farther on and induced me to follow the valley bottom for about a mile or so eastward (Fig. 88). Soon we picked up, wherever the bare ground showed, small pieces of stone, some roughly worked, some mere lumps. The former seemed to resemble the small stone implements which had been met with in plenty in the Lop desert on my march to the Lou-lan Site in 1906.9 Among the specimens, which are described in the List below, I thought I could recognize some blades of the pygmy' type, and a few cores, the rest being coarse irregular flakes struck off by man but not themselves utilized.s'
The presence of these remains of the Stone Age, scattered on the surface of whatever small patches of eroded ground appeared between the dunes, clearly proves occupation by a palaeolithic settlement of what is now utterly lifeless desert, nearly thirty miles away from the nearest traceable bed of the Yarkand river. That the latter could ever have followed in prehistoric times a course so far south appears to me distinctly improbable considering the general configuration of this portion of the Tarim basin, which implies a steady if gentle rise towards the glacis of the K`un-lun.ro But in view of what I shall have occasion to point out below as regards the relation between the surviving fragments of the ancient desert range and the present river courses, the possibility of
9 Cf. Mr. R. A. Smith's paper in Man, xi (191r). pp. 81 sqq. ; also Serindia, i. p. 357 ; below, Chap. vi. sec. i, v.
9a •[The materials, according to Mr. R. A. Smith, are chert, jasper and quartzite.]
to For reasons explained elsewhere, it would be wrong to attach too great importance to comparatively small differences of elevation records derived solely from aneroid and hypsometer readings. Yet it deserves to be noted that our height records, which show fair consistency along the
neighbouring portion of the Yarkand river course (see Map No. 8. A—c. r) and particularly around the Chok-tagh (Camps xxiv, xxv, xxx), seem to indicate a distinct rise of the ground for our three desert marches, the computed heights for Camps xxvi, xxvtr, and xxviii being 3,610, 3,710, and 3,890 feet respectively. Of these camps only the last stood on a sand ridge and therefore above the level of the underlying clay,