a gently sloping gravel Sai and then between tamarisk-cones and reed-beds, was easy for about seven miles. Then the patches of salt-encrusted ground that we encountered became wider and wider, and soon afterwards the boggy soil proved quite impracticable for the camels, which lagged far behind us who were mounted.
It therefore became necessary to skirt this treacherous ground by steering a more easterly course. After another three miles or so we had to cross a succession of shallow salt-encrusted
channels coming from the south-east ; these evidently represented the delta of the dry flood-bed which our former survey had shown near the abandoned roadside tower and station of K`unghsin-tung (Map No. 38. c. 4). Thinking that we had now passed the belt of this drainage and encouraged by the sight of dunes in the distance to the north-east, I decided to resume the march in our original direction. But we were soon faced by fresh difficulties. The expanse of soft dark-coloured shör we had at first to cross was bad enough for the camels. It felt springy underfoot and hid real bog, as the water oozing out in the track showed only too plainly. But again and again we came upon belts covered with a yellowish salt crust which neither ponies nor men could enter without risk of being embogged and which were quite impassable for camels. Great detours had to be made to avoid these treacherous yellow streaks. Even so, camels repeatedly foundered, and we had to extricate them by removing their loads and spreading felts for them to regain a safe footing.
When we had thus struggled along for another three miles, a continuous ridge of high sands, crescent-shaped, rose at last within near view. It appeared to be scarcely more than a mile and a half distant in a straight line, but the ground still separating us from it proved the worst we had yet encountered and threatened to detain us till nightfall. Only by keeping carefully to winding lanes of dark skör, often only a couple of feet wide, or by dodging the oozy yellow swamp, wherever it was edged by cakes of hard salt, was it possible to draw nearer and nearer to the edge of the dunes and the promise of safety. Hassan Akhûn, my tried old camel factotum, managed, with admirable skill and resource, to bring the animals in his charge along the tortuous track that we reconnoitred ahead. But even that hardy veteran confessed afterwards that he would not care thus to tempt fortune a second time. Darkness was coming on when at last a rapid move across a quivering stretch of moist sand brought us to a safe place for the night's camp.
I have described the experiences of this day's march in some detail because the ground crossed presents some geographical interest. It appears to me quite clear that this salt-bog area, extending probably for some considerable distance to the north-west,2 owes its existence to subterranean drainage from the outermost foot-hills of the Nan-shan range to the south. The flood-beds descending from them towards the trough of the Su-lo-ho were all dry when we passed them between Tun-huang and An-hsi in June, 1907. But as was shown by the observations made by me some weeks later that year in the broad transverse valley of T`a-shih and Ch`iao-tzil,3 there can be little doubt that a good deal of subterranean drainage from the high range in the south finds its way to the foot of the outer hill chain.
This moisture, where it reaches the surface along the southern edge of the Su-lo-ho depression, is probably in the main evaporated by the heat of the summer months. A belt of gravel-covered higher ground extends parallel to the Su-lo-ho between An-hsi and the north-eastern end of the alluvial fan of Tun-huang, and this belt, along which the Limes line between the two
Difficulties of crossing.
Passage across salt-encrusted bog.
Subsoil drainage from Nanshan foothills.
>\loisture accumulating during winter.
2 The entry ` Bare gravel Sai ' in Map No. 38. c. 4 to the north=west of the boggy area described ought to be deleted. It has been wrongly taken over from Serindia, Map No. 81. A. 3, where it refers to the ground immediately
to the north of Ko-ta-ch`üan-tzû (erroneously spelt there as Lo-ta-ching).