314 FROM DANDAN-UILIQ TO THE NIYA RIVER [Chap. X
of the sacred hill consists solely of stony detritus overlying reefs of salt. The presence of a real hill so far away from the foot of the mountains, and in a position surrounded by drifting dunes, would, like all striking natural features, suffice to attract local worship, as shown by the numberless ` Svayambhû Tirthas ' of India, ancient and modern 12. But the occurrence of rock salt, so rare elsewhere in this region, must have increased pious regard in this case.
On January 26 we left this curious desert shrine, after having taken along as additional labourers a dozen or so of able-bodied men from the secluded little settlements of Shaikhs,
shepherds, and mendicants dependent on the Mazar. The small watercourses into which the
river splits up before it is finally absorbed in the sand reach only some three miles beyond Imam Jatfar, and as water cannot be got at the site by digging the four iron water-tanks brought by me had to be filled before the start. It would have been very difficult, if not impossible, to keep my camp, counting in the end from forty to fifty people, supplied with the indispensable amount of water so far out in the desert had not the intense cold still prevailing (on January 26 I registered a minimum of 44° Fahr. below freezing-point) permitted of its convenient transport in the form of ice, sacks and nets being improvised for the purpose. At a little farm kept by shepherds of the Mazar, near the terminal marsh of Tiilktich-Köl, our ponies were left behind, and only the camels taken onwards. Some eight miles beyond the Mazar the forest, before invaded by heavy drift-sand, gave way to a wide expanse of sand-cones thickly overgrown with tamarisk scrub. From a high sandhill near the extreme limit of living trees where my camp had been pitched, I could make out a broad scrub-covered belt stretching away
to the NNW. between the great sandy ridges of the true desert. It clearly marked the direction a
of the old extension of the river course.
The march of the next day, January 27, confirmed my surmise that the ancient site would
be reached by following this direction. For the first five miles or so thick patches of dead 6
forest were encountered between the tamarisk-covered hillocks. The time when its trees, mostly b
Toghrak, had flourished could not be very remote, for many of the lifeless trunks still retained
their branches. Winding along the eastern edge of this dead forest, a dry channel, about 4 ft.
deep and about io feet broad, could be traced for some distance. All remains of old forest `
disappeared when, lower down, we entered a zone of steep conical sand-hillocks 15 to 3o ft.
high, rising close to each other, and all covered with tamarisk scrub on their tops. In the f
midst of this belt, about three miles broad from north to south, I came upon a small open area showing broken pottery and remains of an enclosure made of thickly-packed rushes. Inside it the men recognized a few much-withered trunks of planted poplars or Terek. It was manifestly the last trace of some ancient farm, or, perhaps, of some village site completely covered up elsewhere by the closely packed sand-cones 13. Beyond, the ground became more open, with low bare dunes resembling those about Dandan-Uiliq. Above them rose here and there isolated sand-cones, bearing the only living tamarisk bushes that were to be seen. After marching for about five miles patches of ground between the dunes strewn with potsherds, fragments of stone, slag, &c., showed that we had reached the southern edge of the ruined area. The total distance from Imam Jaefar had been about 24 miles, less than the three marches for which my guides Abdullah and Ibrahim had prepared me.
12 It is noteworthy that the small rocky hills, which as the hill of the sacred shrine.'
last isolated remains of a completely decayed mountain range 13 The débris of the Aktaz site near Domoko was found
crop out of the desert near the courses of the Khotan and in exactly similar surroundings ; see below, chap. xIII.
Yarkand rivers (see Hedin, Reisen in Z.-A., pp. 220 sq., 242, sec. ii.
&c.) are everywhere known by the name of Mazâr-tâgh ' the