Sec. if] OLD REMAINS AND ROUTES BEYOND MARAL-BASHI 79
Marâl-bâshi. From the limits of the cultivated zone of Ak-su as far as Chilan (a small roadside hamlet receiving its scanty water from the debouchure of the Kelpin streams) the high road keeps close to the foot of the gravel glacis below the outermost Trien-span hill range and along a line that runs practically straight in the direction of Marâl-bâshi. But beyond Chilan it turns off to the south. The reason for this, considering the total absence of any cultivated area until the vicinity of Tumshuk is reached, can be no other than the necessity of finding halting-places with water and grazing. The successive stages of Yaide, Yaka-kuduk and Châdir-köl are points where water is obtainable either from wells fed by underground drainage from the dying Kâshgar river or else from terminal beds periodically filled by its summer floods."
During the time of these floods traffic is liable to be much impeded by inundations on the line followed by the present road through the riverine jungle-belt between Yaka-kuduk and Ak-tam. This trouble, as well as a great detour, would be completely avoided were it now possible to take the straight line from Chilan through the desert to the sites of Chong-tim and Lal-tâgh. It is solely the absence of water along this line that prevents it from being used as a caravan route at the present day, and that this obstacle did not exist down to 'rang times is conclusively proved by the remains of ancient occupation at those two sites. To this may be added the evidence of the high ruined tower of Soksuk-shahri sighted by me in 1908 to the SW. from the outlet of the Kelpin river ; and also the reliable information then received from my ` treasure-seeking ' guides as to the survival of ancient mounds at two points between Chong-tim and Soksuk-shahri.16 I need scarcely point out that the low dunes of drift-sand which now cover much of the clay steppe on either side of Chong-tim can only have overrun this ground since it ceased to be cultivated by means of the canals whose lines may still be traced." The abundance of living tamarisk-cones and scrub between these dunes proves that subsoil water must find its way there even now, though perhaps at a considerable depth.
Whatever the immediate cause of abandonment may have been in the case of the area formerly occupied about Chong-tiro and Lal-tagh, it is certain that the volume of water available nowadays in the terminal course of the Kashgar-daryâ beyond Marâl-bâshi would not permit of irrigation being carried so far as the first of these sites. We have here clear proof of ` desiccation ' during the last thousand years or so, as far as that river is concerned, and whatever its character or explanation may be,17a we are fully justified in taking the fact into account when we consider the remaining portion of the Ak-su—Marâl-bâshi road, that which lies to the west of the old sites just mentioned.
Ancient route once reached by water.
Reduced volume of terminal Kàshgar R.
15 The brackish wells of Yaide evidently derive their water from the old river-bed known as Ghöra-akin, passing some three miles farther south. This, according to local information supported by topographical indications, connects with the dried-up river-course, known by the same name, that I traced in 1908 past ruined sites in the desert north of Tumshuk ; cf. Serindia, iii. p. 1309.
Yaka-kuduk and Châdir-köl are somewhat less desolate stations situated in the belt of luxuriant Toghrak jungle that accompanies the terminal flood-bed of the Kâshgar river called Kara-köl-jilga (see Map 7. c. 4). This bed, which is periodically filled, seems to have its continuation in the well-marked bed, also known as Kara-köl, that we passed south-west of Ghöra-chöl, the southernmost village tract of the Ak-su district (see Map 7. D. 4). This was said to receive water from the Kashgar-daryâ during summer floods.
The hydrography of the whole Kashgar river delta from the vicinity of Maràl-bâshi, where it may be said to have its
present head, down to its apparent extreme end south-west of Ghöra-chöl remains still for the most part unsurveyed. This whole deltaic system is obviously liable to great periodical changes, and the close approach to, or junction with, it of the flood-beds of the Yârkand river still further complicates its examination, whether as regards the present or past periods.
16 Cf. Serindia, iii. p. 1308.
17 The rapidity with which drift-sand spreads itself from the adjoining desert areas over ground that was once irrigated is well illustrated by my observations at the village sites north of Nan-hu, abandoned only since about 1840 and in places only since the Tungan rebellion ; cf. Serindia, ii. pp. 625 sqq.
173 Regarding a likely cause of desiccation', as manifested by a diminution during the historical period of the volume of water carried by the rivers of the Tarim basin, see now my paper on ` Innermost Asia ', G.J., 1925, Ixv.
PP. 487 sqq•