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

Sec. i]   THE TERMINAL STATION ON THE WALL   639

•   The gravel plateau above Toghrak-bulak, where the route now crosses the deep-cut Su-lo Ho bed, would have suggested itself as a very suitable position for a first watch-post on this line thrown out to the north-west. But no tower survives there, and the rapid inspection that I was able to make of the river's left bank did not reveal any traces of ancient occupation. Perhaps closer search by some future traveller may find its reward here. The trench-like Nullah in which the present bed of the Su-lo Ho lies, sunk some fifty to sixty feet below the level of the ` Sai ', is so narrow and steep that it is only visible on close approach. The river by May 2 filled its bed completely and was unfordable at the marshy spring where we had before crossed and camped on March 7. Descending a short distance, I found a place where the becl somewhat widened and was sufficiently shallow for crossing. The volume of water then carried by the river amounted to about 1,800 cubic feet per second. This observation has its interest as showing how great a proportion of the total discharge of the Su-lo Ho is lost by evaporation and by absorption in the marshes around and below the Khara-nôr before it finally empties itself into the terminal basin below Toghrak-bulak. A measurement taken on April 1, 1907, on the Su-lo Ho, at a point well above its junction with the Tang Ho, had indicated a volume over 4,000 cubic feet per second, and four days later I found that the Tang Ho at Tun-huang carried over 2,100 cubic feet in a second, without counting the water taken off in the canals of the oasis. Thus the discharge at Toghrak-bulak represented less than one-third of the total amount of water •carried down into the Khara-nbr. Yet it may be safely assumed that this total volume had considerably increased in the course of a month, owing to the continued melting of snow and ice in the high ranges of the Nan-shan.

The first ruined tower, T. 1, reached after about two and a half miles beyond Toghrakbulak, stands on the very brink of steep cliffs that rise fully seventy feet above an old becl of the Su-lo Ho, now dry, and overlooks, for a considerable distance, the wide depression in which it lies. It is possible, and even likely, that this river-bed, the debouchure of which into the ancient terminal basin crossed on the way from Bash-toghrak I have already mentioned,' still received occasional floods, or at least held drinkable subsoil water, at the time when the Limes was occupied. The dry salt pools noticed in 1907 close below T. I contained fairly large sheets of water when I passed here seven years later. The extensive reed-beds that fill most of the depression here also point to occasional inundations from the actual terminal course of the Su-lo Ho.

Though the ground close by was much cut up by small ravines due to the action of water, the tower, occupying the top of a small and almost completely isolated ridge, difficult of access, still rose to a height of about twenty feet. It measured about sixteen feet square at its base and was built with stamped clay in layers of three to four inches thick. Wind-erosion had cut down the soil near the north-east corner by some three feet or so. A structure of some sort had once stood on the west of the tower, but only a brick-built base remained, measuring about twenty feet by fifteen. Owing to the steepness of the slope below, little débris and refuse survived. Among it, however, I found three fragments of Chinese records on ' slips'. One of these, Doc., No. 673, contains a statement about the manufacture of bricks by soldiers. Large stones which I found lying at the foot of this tower are likely to have been originally placed on the top for use in defence, just as I have seen them in all the modern watch-towers and circumvallations of westernmost Kan-su.17

Su-lo Ho bed at Toghrak- bulak.

Volume of water

carried by Su-lo Ho.

Tower T. r overlooks old riverbed.

Con-

struction of tower T. 1.

1e Cf. above, p. 552.

" I may here note in passing that when I returned to this tower in 1914 I could trace my own footprints of seven years before quite clearly on the gravel surface around, a curious

experience repeated again and again along the revisited portion of this westernmost Limes, and one for which similar observations of far more ancient tracks, which will be recorded later, had fully prepared me. See below, pp. 656 sq.