from China towards the Su-lo-ho and the Tarim basin. Information kindly given by Père Verberne, in charge of the Belgian Mission station at Kao-t`ai, concerning a ruined town known as Lo-to-
j ch`êng at a day's march distance westwards, induced me to let Lai Singh proceed for a preliminary inspection of the site on June 23rd. To save time, I myself set out on the same day for Kan-chou, where I was to make arrangements for our proposed explorations in the Nan-shan.
Lal Singh, after having visited Lo-t`o-ch`êng and thence made his way back to Kao-t`ai, rejoined me at Kan-chou by the more devious but previously unmapped route along the right bank of the Kan-chou river. The report he was able to furnish of Lo-t`o-ch`êng was brief, but sufficed to show that its remains were in part comparatively recent and in no case of great archaeological interest. He had found the ruined town situated on the right bank of the wide deep-cut bed, then practically dry, in which the waters of the Pei-lang-ho and Hsi-ta-ho streams descend towards the Kan-chou river, when not absorbed by irrigation higher up (Map No. 43. D. 2). The sketch-plan prepared by the Surveyor showed a rectangular circumvallation measuring a little over a mile from east to west and about 1,430 yards from north to south. Its walls were of stamped clay and io feet thick. A cross-wall built at a distance of about 33o yards from the west face and parallel to it divided the interior into two unequal portions communicating by a gate in the middle. The outer west wall, built immediately above the steep right bank of the flood-bed, had for the most part fallen. At the corners and along the northern and southern faces rectangular bastions projected. Gates protected by small outworks led through the eastern and northern walls.
The interior was completely devoid of structural remains, except in the south-eastern corner, where walls of less strength partitioned off a small enclosure about 210 yards square. Within this Lai Singh found a well 8o feet deep and some half-ruined structures, perhaps intended to shelter wayfarers, who apparently make this a halting-place between the high road leading to Su-chou and the string of small oases stretching along the foot of the Nan-shan in the south. It was probably within or near this small enclosed area that Lal Singh picked up the fourscore odd fragments of Chinese coins which he brought me. All of them have proved to be modern, the Nien-haos as far as legible ranging from A.D. 1644-62 to A. D. 1851-62. The pottery specimens brought back by Lai Singh also had a modern appearance.
I regret nevertheless that the accident which I suffered three weeks later in the mountains prevented my visiting the site in person on my return journey towards Mao-mei, as I had originally intended ; for the local tradition communicated to me at Kao-Cai ascribed both this site and the remains of another walled town called Sou-san-wan, which Lal Singh sighted among dunes about
~I four miles off on the opposite side of the river-bed, to a ` Mongol ruler' of Tang times. Whether
archaeological evidence could be traced on the spot in support of this traditional dating must remain doubtful. But it would certainly be interesting to investigate how the neighbouring agricultural settlement, the existence of which Lal Singh found attested by numerous ruined farms to the south-west and east of the town, had received its irrigation. Judging from what he saw in June and again when passing the place on his way down from the village of Nan-ch`üan in August, the river-bed adjoining the town site would not now carry water sufficient for the maintenance of regular irrigation on ground far down on the gravel glacis of the hills and fully nine miles from the limit of present cultivation. The change which must have taken place in the conditions here prevailing is evidently one suggesting `desiccation', whatever its physical cause.
The two long marches which brought me on June 23rd and 24th from Kao-Cai to the city of Kan-chou led along the great high road from Su-chou and took me over ground, mostly cultivated, which I had already seen in 1907.2 The old site of Hei-shui-kuo (Map No. 46. B. 2), which was