sandstone, intermixed with lumps of rock salt.' The neighbouring spurs of the Kum-tâgh seemed to be of the same formation, and the foot of the hill-side was everywhere seen to be studded with shallow pits, the result of salt-digging operations. The point is known as Tuzluk (` the salt place') and marked by a couple of reed huts. Beyond this we continued across the bare inundation belt by the side of the lake, which, towards its southern extremity, shrunk considerably in width and split up into a series of lagoons between extensive reed-beds (Map No. 8. B. 2).
The statement of Aziz ` Pâwân', that on a hunting trip the year before he had found there drinkable water, induced us to pitch camp amidst tamarisk-cones not far from the southernmost of these lagoons. But the water proved decidedly brackish, probably because the lake reservoir had not been sufficiently refilled by the last summer flood, and next day we had to move back with the camels some six miles before fresh water was found by the lake shore wherewith to fill tanks and skins for the journey. As the water was very shallow and the men had to wade far out, it took practically a whole day's work to fill, secure, and load up our six galvanized iron tanks and the forty-two goat-skins brought from the ist Sappers and Miners' Workshops at Roorkee.
On October 29th we set out to the south-east after a start delayed by the packing and distribution of the heavy loads of water. For three miles we crossed bare salty soil, marking a former extension of the lake end, now dried up. Then the first dunes were reached, stretching in irregular lines across flat ground of disintegrated loess dust, covered on its surface with abundant remains of quasi-petrified reeds and scrub. On this ground, which extended for about another three miles and obviously marked a former riverine belt, living vegetation was entirely absent. Farther on the dunes increased in height to thirty feet or more, and lay much closer together. Yet scanty living tamarisk scrub appeared again between them, and in a small depression we found even a group of young Toghraks rising above the sand. Their presence encouraged the hope that subsoil water was near, and on pitching camp here, a well dug to a depth of about four feet sufficed in fact to reach it. When the march was resumed next morning we found the dunes growing rapidly higher and very closely packed. The crests of individual dunes here faced with their concave sides regularly to the south-south-west. But the huge ridges or ` Dawâns ' into which they lay heaped up, and which we soon encountered rising in endless succession to heights of 200-300 feet, were found to run invariably with an approximate bearing ENE.-WSW. 'Their lines thus lay at a fairly wide angle diagonally across our intended direction, and consequently involved constant ascents and descents, both very trying to the camels.
It soon became evident to me that this uniform bearing of the ` Dawâns' was a result of a Iaw that I had previously observed in the desert west of Lou-lan, in the Keriya river delta, and elsewhere, under the action of which these big accumulations of drift-sand in the Taklamakân are aligned parallel to the main direction of the nearest large river.8 But there were important differences to be noted here and there distinctly unfavourable to us. In the areas of high sands previously crossed by me these ` Dawâns ' had not presented themselves as unbroken chains such as faced us here, but as separate ridges varying in length, yet always showing at their ends low saddles or shoulders over which movement with laden camels was comparatively easy. The steering round those ` Dawâns ' had indeed compelled us to make considerable detours ; but the broad and flat valley bottoms found between them afforded easy ground, and the march across it served to give relief to the camels.
The tract now encountered offered obstacles far more formidable in all respects. To get round the ` Dawâns ' was impossible, for nowhere did they show distinct saddles or breaks. To move
7 For the specimens taken here and beyond, see Prof. 8 Cf. Serindia, i. pp. 241, note 2, 4$1 sq. ; iii. p. 1239;
Sollas's App. 0. also Hedin, Central Asia, i. p. 363: