Sec. i] TO THE EASTERNMOST OUTPOST OF LOU-LAN 283
We left the shelter of the little oasis on the morning of February 25th under a sky more than usually hazy, and after a night during which the temperature fell only to 28° Fahr. While Lai Singh under Abdurrahim's guidance set out eastwards to Yetim-bulak, I myself steered approximately south-south-east towards a point near which, judging from our previous mapping, the ancient post L.F. was to be looked for. Having reached this, I should look out for indications as to the line beyond it that the ancient Chinese route might have followed. We first skirted the western edge of the wide shallow valley that holds the flood-beds coming from Altmish-bulak, and descended for about four and a half miles across almost completely decomposed rocky ledges. Then after proceeding for some two miles down a well-marked watercourse, dry but with some scrub in it, we reached a big bed cut into the gently sloping glacis of gravel and fully half a mile across. The tamarisk drift-wood, of which quantities were found here, had evidently been brought down by occasional spates from the Kuruk-tagh. We next struck a smaller and more easterly flood-bed, and after having covered close on ten miles from camp reached the foot of the glacis after descending about 35o feet by aneroid.
Immediately beyond the foot of the glacis we crossed a narrow lagoon-like depression. It showed a surface of salt-encrusted clay cracked into big cakes, and in one place still held a small pool of briny water. There could be no doubt that this depression was flooded from the torrent-beds we had left behind, on the rare occasions when rain falls on the outermost ranges of the Kuruktagh.4 Afraz-gul's survey of 1914 subsequently showed that it extends farther to the south-west and there widens. Living scrub had ceased after we left the easterly flood-bed. But when we had crossed this narrow depression and passed through a line of steep clay Mesas, fantastically eroded and up to sixty feet in height, we came upon a second, somewhat wider, depression encrusted with clayey shôr and having a thin line of live ` Kaurük ' scrub along its edge. This evidently marked the farthest limit to which moisture had ever been carried in recent times by floods from the Kuruk-tagh.
Having left this depression behind and covered about eleven and a half miles from Altmishbulak, we entered ground (Map No. 32. A. 3) where a succession of physical difficulties seriously
impeded the camels' progress and soon frustrated my hope of making this first of our marches a long one. First, for about two miles, we had to make our way right across closely packed lines of Yardang ridges, up to fifteen feet in height. Their direction was, as usual, from ENE. to WSW., and this alone would have sufficed to make them an awkward obstacle, as our route lay at right angles to it. But it was a novel and very trying experience to find them coated throughout with a whitish crust of hard salt-permeated clay which severely affected the camels' feet.
These strange Yardangs showed throughout fairly gentle rounded slopes, such as the action of water might produce on clay carved out by wind-erosion. It has therefore occurred to me that these ridges, originally, no doubt, wind-eroded, owed both the modification of their slopes and their heavy salt-incrustation to the vicinity of the depression farther north, which at an earlier period probably held water more or less permanently and to a height sufficient at times to flood the belt of these Yardangs. Their colour and shape naturally brought to my mind the ` White Dragon Mounds ' referred to in the Former Han Annals and the Wei lio's account of the Lou-lan routes Evidently this reference could not apply to the belt just described ; for this was far away from any possible line that the ancient route I was anxious to trace could have followed. But it was no
4 Dr. Hedin, too, on his march in 1901 from tlstin-bulak towards the Lou-lan site, observed along the first portion of his route ` distinct signs of running water, showing that the rain-water does sometimes get down as far as this ' ;
cf. Central Asia, ii. p. 233.
5 Cf. Serindia, i. pp. 340 sq., 418 sq. ; Chavannes, T`oung-pao, 1905, pp. 529 sq.