国立情報学研究所 - ディジタル・シルクロード・プロジェクト
『東洋文庫所蔵』貴重書デジタルアーカイブ

.10

Advanced watch-posts on Lou-lan route.

Commanding position of watch- • station

T. Iv. b.

Westernmost watchtower,

T. iv. a.

Limes wall turns S. of T. Iv. a.

634   THE END OF THE CHINESE LIMES   [Chap. XVII

At Toghrak-bulak is struck the present caravan route to Lop. That the ancient Lou-lan route must thence have followed practically the same line north-westward is made quite certain by the series of ancient watch-towers, T. 1, T. ii, forming an advanced line of outposts for the Limes which is to be described further on. It is at Toghrak-bulak that the Tun-huang—Lop route, which the Limes was primarily intended to protect, finally leaves behind the surface drainage of the Su-lo Ho Basin, and at first sight it may seem as if this point might have offered an equally, or perhaps even more, favourable position for the termination of the Limes. But against this view it will suffice to point out that the very narrow and deep Nullah through which the Su-lo Ho has cut its way here westwards affords but very scanty space for vegetation by the side of the river-bed, and that the gravel-covered plateaus on either side are absolutely barren. Such a position would have been most disadvantageous for what was at the same time the terminal station of the Limes proper and a bridge-head, as it were, for the route crossing the desert to Lou-lan.

I will now describe what I learnt from the exploration of the three watch-stations T. iv. a—c, situated close together at the westernmost end of the fortified border line, and the survey of the ground guarded by them. As I looked out from the isolated broad clay terrace near the edge of the great marsh-covered depression, where on April 3o I had placed my camp, C. 171. a, for the sake of being near to water and yet not without protection from the pest of mosquitoes, etc., that never leave it at that season,4 the most conspicuous of those towers was T. iv. b. It rises on the brink of steep clay cliffs, some 1 zo feet in height, where the narrow, tongue-like plateau already mentioned that runs from T. 111 falls off into the depression westwards. The view I obtained. from this commanding position, both across the great basin and over the scrub-covered inlet and the gravel ` Sai ' northward, was extensive. No better look-out place could have been selected for this outlying, and hence exposed, portion of the border line. But the wall had been carried, all the same, about a mile and a half further west to an isolated clay terrace rising from the scrub-covered north-eastern bay of the basin to a height slightly lower than the end of the narrow plateau ; from this the terrace had, no doubt, been detached through erosion.

The top of this outlying terrace, about loo yards long from east to west,5 was occupied at its western end by a much-decayed tower, T. iv. a, about sixteen feet square at its base and built with layers of stamped clay. Its remains were so poorly preserved that they were somewhat difficult to distinguish from the natural clay of a small knob about seven feet high which served as a base. Close to it were found half a dozen Chinese slips' of wood, mostly fragments (Chavannes, Documents, Nos. 438, 442-4), besides a bossed button in bronze and some miscellaneous iron fragments, T. iv. a. ooi, 002. But far more interesting than these scanty finds at the watch-tower which marked the westernmost point attained by the Limes wall were the clear observations of the change made there in its direction. Considering the interest attaching to the position, I was glad to be able to verify them by another visit in 1914.

From the top of the terrace, which completely overlooked the low-lying ground on all sides, I could quite clearly see the line of the Limes running almost due south in the direction of the

A curious parallel is presented by the terminal course of the Helmand, though on a much bigger scale. Below Rûdbâr the Helmand approaches to within about eight miles of the north shore of the Gaud-i-Zirrah, which occupies the final depression reached by its waters at periodic intervals. Yet at present the Helmand River travels close on ioo miles further north before it empties itself into the marshes of the Haman in Seistan, and another 130 miles or so have to be covered by its waters when, in years of great flood, they penetrate south

again to the west shore of the Gaud-i-Zirrah. A comparative study of the terminal courses of the Helmand, Tarim, Su-lo Ho, and Etsin-gol, with all of which I have had occasion to familiarize myself more or less, would bring out various interesting points of contact ; but this is not the place to examine them.

4 See Desert Cathay, ii. pp. 134, 157 sq.

5 Fig. 17o shows the terrace in the distant background, as seen from near T. iv. c.