Sec. i] FROM TURFAN TO SINGER 721
either of these points and the springs just named, and this makes the use of both routes very difficult at all seasons except when ice is available at Achchik-bulak and at the similarly named salt springs on the Deghar route (Map No. 28. c. 4). On this account the route now ordinarily followed from the Lop portion of the Tarim to Turfan passes from Singer north-westwards to Üjme-dong on the Kara-shahr road, and thus reaches the Turfan depression at Toksun.5 Along this devious route springs of drinkable water are to be found at a number of places (P`o-ch`êng-tzû, Kan-so-ho, Shôr-bulak), and this explains its choice as a line of communication by the Chinese administration. After the reconquest of Hsin-chiang, rest-houses, now lying in ruins, were maintained along it for a number of years. For the same reason it also appears probable that even in ancient times the route connecting Lou-lan directly with Kao-chiang and leading north of Singer could have served for traffic only to a very limited extent.
When crossing the salt-encrusted bed of the dried-up marsh on our start from Arpishmebulak on February 21st, I had occasion to observe features which were of interest by reason of their close resemblance to those noticed a year before on our exploration of the ancient Lou-lan route where it crossed the dried-up Lop sea-beds At first the route led over soft clayey shôr ; this farther on changed into a hard crust of salt cakes. Across this troublesome belt, fully a mile wide, traffic, comparatively slight as it is at all times, had worn a winding track, 5 to 6 feet wide in most places. The big cakes of hard salt had been reduced here to a less rugged surface, and hence the track offered better going. Within this belt it was curious to notice narrow salt-coated ridges 10 to 15 feet high, very similar in type to those `White Dragon mounds ' which fringe the bed of the ancient Lop Sea, but on a scale so much larger. Beyond this the crumpled-up salt cakes gave place to a surface of salt-impregnated lumps of clay with characteristic patches of soft clayey shôr between, just such as I remembered seeing when we crossed the bay of the dried-up sea north of Kum-kuduk.7 Along the opposite shore of the marsh there stretched a narrow but far extending belt of live tamarisk-cones and reed-beds. This was fringed again to the south by isolated cones with only withered roots of tamarisks that had died long ago.
There was no trace of living vegetation on the bare gently rising Sai, first of coarse sand, then of gravel, over which the ascent lay to the third range bordering the basin on the south. Near two shallow flood-beds, however, I noticed remains of completely decayed dead tamarisks, suggesting that at an earlier period moisture must have been fairly constant in those beds. After we had ascended to the top of a broad side spur of the range where red sandstone rocks were cropping out from the heavy cover of detritus, there lay before us a wide peneplain streaked with low rocky ledges. A broad Nullah crossing from south to north showed again living tamarisk bushes, and ascending it we passed the salt spring which Lal Singh had halted at and marked on his plane-table. As it lay among low rocky hillocks to the west of the route, we did not see it.8
Snow lying in sheltered spots allowed us to halt for the night within five miles of the broad saddle known as Yagach-ile-dawan. Over this saddle the almost imperceptible watershed of the third range was crossed next morning at an elevation of about 4,600 feet. From it a wide and striking view opened across a broad basin to the south, appropriately known as Kizil-sai from the reddish clay and sand at its bottom, and towards the boldly serrated hill range of the Kizil-tagh, west of Singer. This, as our surveys show, with its continuations west and east, marks the true backbone of the western Kuruk-tagh (Map No. 29. A, B. 2) and contains its greatest elevations.
6 See Maps Nos. 29. A. I ; 28. A. 4.
6 See above, i. pp. 300 sqq.
7 Cf. above, i. pp. 357 sq. The indication of the salt-encrusted surface in Map No. 29. B. I, taken from Lill Singh's
plane-table record, ought to have been correctéd in accordance with the above observations.
s A subsequent statement of Muhammad Bagir, the guide secured at Singer, confirmed the name Parsa-khöja-bulak as recorded.