78 FROM KÄSHGAR TO THE KHOTAN RIVER [Chap. III
for it. One is represented by the gap in the Bel-tagh above described ; the other leads through the defile that separates the southern end of the Bel-tagh from the northern offshoot of the Achaltâgh formed by the low rocky spur of Arach. I had visited this defile on my first rapid reconnaissance from Marâl-bâshi in May, 1908. I ascertained on that occasion that the high road from Tumshuk to Marâl-bâshi had in quite modern times, until after the Chinese reconquest of the Tarim basin in 1877, passed through it, the present line past the Ökur-mazâr being then impracticable owing to great undrained marshes in the area now occupied by the lands of Charbâgh.12 I had been able to trace there the remains of old watch-towers and of other fortifications meant to guard the defile, and their presence left no doubt that an important road must have passed there in ancient times.
Traces of I had no time then to examine the desert ground to the east of the defile, and this was an addi-
old canal. tional reason for now returning to Marâl-bâshi by this passage, appropriately known as Achal, ` the opening '. I found no reason to regret the decision. We passed numerous lines of dead Toghraks with shallow dry channels between them for the first two miles from the Lâl-tagh site, clearly showing by their direction that running water must have reached this part from the south, i. e. from the area west of Tumshuk still liable to inundations from the Kâshgar-daryâ.. Then, after crossing a belt of scanty tamarisk-cones, we emerged upon a bare clay steppe undergoing wind-erosion, and here for more than a mile came across frequent patches of ground covered with ancient potsherds and other ` Tati ' remains. In the midst of them we crossed the unmistakable embankment of an old canal about 12 feet across at its top and owing to erosion of the surrounding land raised some 5 feet above the present ground level.
. Tati ' Low dunes have overrun much of the ground nearer to the defile, and these, together with the
near insltal. approach of darkness, made it impossible to ascertain whether this ` Tati ' and the one traced in
1908 close to the east of the defile belong to one continuous area of old occupation. The small relics picked up en route, including paste beads and a bronze ring, do not permit of any close dating ; but the fragments of pottery, whether plain or glazed, have a distinctly ancient look 13 Obviously the greater extent of the once cultivated area must have lain in the direction of the canal that seemed to run from south-east to north-west. It remains for some future traveller to follow its course farther across the bare steppe and low dunes that stretch away to the present high road south-eastward. I myself felt obliged through practical considerations, including the need of securing water for our animals, to turn back towards Marâl-bâshi by the track which I had followed in 1908 on my first visit to Achal. The marshes that I then saw near the north end of the Achaltâgh had completely dried up, but the extensive beds of reeds and luxuriant tamarisk jungle still flourished. Making our way through them by the guidance of the stars we reached water late that night at the isolated little patch of cultivation marked on the map (No. 8. A. 1) as Talib Hâji's farm. Next morning Marâl-bâshi was gained by the high road busy with market traffic (Fig. 85).
Modern The excursion just described concluded what investigations concerning the old road line to
route from the north of the terminal course of the Kâshgar river I was able to make. I hoped to complete Ak-su to
MarAl-bâshi. them in 1915 on my return journey to Kâshgar by tracing the ancient route in the desert between Chilan (Map No. 7. c. 3) and Chong-tim. But circumstances beyond my control made it impossible for me to spare the time requiredl4 So it remains for me only to sum up such conclusions as, in my opinion, may be drawn from the surveys and archaeological observations actually made. In the first place it deserves to be noted that, as is evident from Maps Nos. 7, 8, the line followed by the present high road from Ak-su towards Kâshgar makes a considerable detour between Chilan and
12 Cf. Serindia, iii. pp. 1310 sqq. 14 See below, Chap. xxiv. sec. ii.
13 For specimens see the List at end of section.