is easy to understand if we bear in minci that the effects of wind-erosion in the ancient delta of the Lop Desert, as measured by the depth of the Yârdang trenches, grow distinctly less as we proceed farther south, i. e. farther away from ground lying close to the foot of the Kuruk-tàgh glacis. This difference in the result of the wind's erosive action may itself, I believe, be safely attributed to the fact that the sand which serves as the abrading instrument in that action is a more powerful factor northward, where it consists of coarse grains swept clown from the gravel glacis of the Kuruktâgh, than in the south, where it is made up mainly of locally disintegrated loess dust.
Less than two miles beyond the place where was found the above-mentioned bronze ornament, we came upon a depression winding from WNW. to ESE. and marking an unmistakable ancient river-bed (Fig. 143). Among the rows of dead Toghraks which were to be traced on either side of it, some trunks still stood upright, as seen in the background of the photograph. The width of the bed was about 150 yards, and the divergence of its direction from that which the Yârdang trenches invariably follow made it easily recognizable. The line of high tamarisk-cones that we passed in 1906, just before finding the first relics of the Han period,14 lay exactly in the continuation of this dry river-bed to the WNW.1» Not far from the north bank of it we picked up a well-preserved iron awl, C. xciii. 011.16
Beyond this the ground was cut up into a maze of short Yârdang trenches, reaching a depth of 10 feet or so, as shown by the photographs (Figs. 141, 142) taken at,a point about eight and a half miles marching distance from Camp xciii. Here three Chinese coins of the Han type were picked up in quick succession. One is an inscribed Wu-clzu piece ; another a much-clipped specimen of the same ; while the third, also clipped, still shows a trace of the legend Huo-ch`üan, introduced by the usurper Wang Mang about the time of Christ. Half a mile farther on, after passing an ancient river-bed, much effaced, with rows of dead trees all lying on its banks,17 we entered ground that had been much eroded and was almost bare of ancient vegetation. This for about three miles yielded finds of worked stones, miscellaneous fragments of metal objects, glass beads and potsherds in abundance. To the north the horizon was bounded by a line of high tamarisk-cones. But when after about twelve miles' march I climbed an isolated cone about 3o feet high at some little distance from this line, the familiar Stûpa of the Lou-lan station L.A., together with the neighbouring ruins, could be seen quite clearly to the NNE. The glow of the setting sun was reflected by them just as it was by the towers of the Tun-huang Limes when I searched for them years before on my second journey. It was a great relief to be assured once again by that landmark that we had steered a true course across this dead and desolate desert.
The line of close-set dead tamarisk-cones, 20-30 feet in height, proved on close approach to mark an unmistakable river-bed (Fig. 144), with a fringe of dead Toghraks, many of them still upright, on either bank. The bed, running approximately from west by south to east, was 16 to 18 feet deep and measured 146 yards in width where we crossed it. Low dunes covered its bottom in places, and here and there small tamarisk-cones, all dead, rose within it. They had obviously
14 See Serindia, i. p. 36r.
15 In Map No. 29. D. 4 the row of dead tree symbols and the broken line marking the old bed should have been drawn with a more westerly bearing. At the corresponding point of the 1906 route the ` ancient site ' symbol ought to have been placed above, instead of below, the map entry ` Ancient potsherds and bronze fragments '.
16 The appearance of this littlç implement makes it difficult to feel sure about its not having been dropped perhaps by one of Dr, Hedin's or Mr, Tachibana's men who
may have crossed this ground.
17 This bed with the symbols of dead Toghraks ought to have been shown on the map just to the north of the entry ` I-Ian coins '. Its position and bearing, NW.—SE., corresponds exactly with the ancient channel, marked by a line of dead Toghraks, which our survey of 1906 recorded about a mile and a half south of Camp 123 (see Map No. 29. D. 4 and Serindia, i. p. 361), and which has its probable continuation in the belt of dead trees shown by the map close to Camp 126 a on the route to the Tarim ; cf. Serindia, v,, Map No. 60, c. 3...