miles. The open view then obtained from a much broken hillock of slate which rose above it to a relative height of about two hundred feet allowed us to make quite sure that the barrier interposed by the plateau between the basin on the north and the Su-lo-ho bed on the south was complete. The course of the latter was, in fact, entirely masked by this barrier. The hillock had evidently served at one time as a landmark ; for we found it crowned by a cairn, of uncertain date and origin.12
Having thus cleared up a point of distinct topographical interest, we descended again into the basin below us in order to reach ground where fuel and some grazing might be available and touch be resumed with Surveyor Muhammad Yaqûb and the body of our caravan. Moving to the northeast amidst Mesas of no great height we came, after about three miles' going, to a small plain with scrub and tamarisk-covered hillocks and there camped. The caravan track was found to run about half a mile farther north, through a well-marked depression beyond the area of erosion terraces. The elevation of this point of the basin is shown by the barometrical readings of Lai Singh, who on his way from Tun-huang had also camped here, to be about 90 feet above Béshtoghrak.
Next morning we were joined by Muhammad Yàqùb and the main caravan, who had halted at the usual camping-place, known as Achchik-kuduk, about two and a half miles nearer to Béshtoghrak.13 I was thus able to make doubly sure of the mapping of the Su-lo-ho course right down to, and inclusive of, its terminal lake-bed by letting the Surveyor set out on this task from the point where our onward march of that day brought us to the plateau immediately overlooking the easternmost part of the basin. Having regard to the condition of our animals, all hard tried by the long series of preceding desert marches, and also to the difficulty of obtaining water, which we could expect to find only in, or close to, the actual bed of the Su-to-ho, I felt obliged to march myself straight by the caravan track to the western end of the ancient Chinese Limes. I was the more anxious to reach this without delay that it was important to save time for supplementary explorations along the Limes, before proceeding to Tun-huang for the rest which men and animals alike were now in urgent need of.
Thus what I saw, in the early part of our march on March i6th, of the eastern extremity of the basin was the same as has already been recorded in the account of my previous passage,l4 and the briefest description will suffice here. A short distance from Camp cxiii all wind-eroded terraces were left behind, and the ground now assumed the character of a shallow but unmistakable valley, bordered on north and south by steep cliffs of shale and consolidated gravel. Its width,
as the map shows, steadily narrowed eastwards, while its bottom became like a Sai, covered with coarse gravel but supporting here and there patches of hardy scrub. The appearance of the valley
was unmistakably that of a Wadi carved out by the intermittent floods of a river, and the upward slope of its bottom eastwards was perceptible to the eye. The volume of the floods that were once
12 It has occurred to me since that this hillock might have served to mark the point where an earlier track practicable for carts coming from the side of Toghrak-bulak and the Limes could descend without difficulty into the basin. For to the east of this hillock the edge of the plateau seemed to fall off everywhere in steep cliffs like the cliffs of shale past which the present caravan track, practicable for camels but not for carts, descends into the deep Wadi forming the easternmost extension of the basin.
To make this present drop into the basin fit for cart traffic would require engineering, whereas by following the plateau edge about three miles farther, to the hillock
above named, an easy slope for the descent and ascent of carts could be gained without any serious detour, on the line from Toghrak-bulak to Bésh-toghrak. It is also quite possible that the ancient route between these points, where it led across the ` Three Ridges Sands ', followed a line lying a few miles farther south than the present one.
13 In Map No. 35. c. 4 the position of Achchik-kuduk, where our Camp 1S3 of 1907 stood, has by an oversight not been separately marked. The insertion of this ` camp number ' against our Camp cxiii of 1914 is an error.
14 Cf. Desert Cathay, i. p. 538.