Sec. iii] WATCH-STATIONS ALONG THE ANCIENT ROAD TO KORLA 771
it was clear from the distinguished explorer's detailed account of the natural features of the ground that we should be able to follow more or less closely the foot of the gravel Sai, which in spite of the prevailing dust haze would enable us to keep a good look-out for the towers. The same account allows me to restrict my own description of the ground which we traversed in our search for the ruined watch-towers to the minimum needed to show its general character.
The march to the nearest of the towers reported by Dr. Hedin brought us gradually closer and closer to the terrace-like edge of the Sai which we had last seen at the Ying-p`an site. Where we crossed flood-beds descending from valleys in the hills, the line of these terraces, up to 3o feet or so in height, was interrupted for some distance. But elsewhere it was very pronounced. In places it receded into small bays like the cliff-lined coast of a sea or was scooped out by local drainage into finger-like projections, just as I had seen them along the northern edge of the Kuruk-darya belt and on a larger scale by the side of the marsh basins of the terminal Su-to-ho.s It was difficult to escape the feeling that we were still moving along the shore-line of that ancient sea which once covered the whole of the Lop desert and stretched its easternmost arm far away into the valley of Besh-toghrak. On the left, to the west, a belt of vegetation comprising scrub, high tamarisk-cones, and occasional clumps of wild poplars kept close within view, clear evidence that the present bed of the Konche-darya could not be very far away. A dark line showing in the dim distance, in fact, probably marked the jungle growing on its banks.
The tower that Dr. Hedin refers to by the name of `Ayag-tora' (i. e. Ayak-tura, ` the lower tower ') was duly sighted after we had covered sixteen miles. The distance separating it from Y. I (Kurghan) is far greater than that between the watch-towers farther on. My suspicion that an intermediate post, whether owing to the deceptive nature of the ground, which is covered with tamarisk-cones, or in consequence of far-advanced decay, had here escaped attention, was subsequently confirmed by a statement made by Ibrahim, a hunter from Shinalga, whom we picked up in the jungle south-east of Sai-cheke (Map No. 25. B. 2). The remains of the watch-station Y. II were likewise almost hidden by close-set tamarisk-cones and would not have been detected by us had we not been moving high up on the Sai. Patches of reeds close by and a neighbouring belt of Toghraks to the south clearly indicated the vicinity of subterranean drainage probably brought down by the small flood-beds that we crossed here. Obviously water must have been obtainable here when the ancient watch-station was built. Indeed we found, about 3o yards to the south of it, a shallow hole apparently marking an old well. Wet soil was reached when we had dug here to a depth of about four feet ; but as the soil was permeated with sher, no drinkable water was hoped for and the intention of camping here was abandoned.
Close examination of Y. II (Fig. 345) revealed features of interest, forcibly recalling various watch-stations that I had explored on the ancient Chinese Limes. The remains comprised a watchtower of the size and construction usual on the Tun-huang Limes and poorly preserved walls of quarters situated to the west of it, as seen in the sketch-plan (PI. 38). Both tower and quarters had been built on a small plateau formed by cutting down and then artificially enlarging what evidently had been the top of a fairly big tamarisk-cone. This now rises in the centre about 12 feet above the scrub-covered ground surrounding it. The watch-tower built on the eastern side of this platform still stands to a height of about 20 feet and at the base appears to have originally measured about 20 feet square. But its foot has been strengthened on the western and southern faces by the addition of slanting masonry about 7 feet wide at its lowest point. This as well as the outer casing of the original tower is built of bricks, 15" x 7-8" x 3" in size, just like those used at Y. I and in most of the watch-towers of the Tun-huang Limes. Reed layers at intervals of 16 inches divide the
6 See above, i. pp. 345 sqq. ; Serindia, ii. pp. 576, 589, 642 ; Desert Cathay, ii. pp. 139 sqq.