THE NORTH-WEST FRONTIER OF KAN-SU [Chap. XXVII
High road from Su-chou to Yümên-hsien.
Han Limes line further north.
Reconnais- sance to Shih-êrh- 1nn.
Bifurcation of Su-lo Ho drainage.
'SECTION V.—THE HAN LIMES FROM Yt1-MÊN-HSIEN TO AN-?-ISI
The high road by which I moved on to Yü-mên-hsien after regaining Su-chou undoubtedly follows what must have always been the main line of communication for the Chinese towards Tunhuang and the Western regions. But there are no ancient remains above ground to mark it. Nor can the small areas of cultivation which are to be found, as Maps Nos. 85, 86 show, near some of the roadside stations between Chia-yi.i kuan and Yi.i-mên-hsien ever have been of any importance considering the scanty surface drainage from the mountains which traverses this plateau belt. The rugged and barren range, which we have already had occasion to notice to the north-west of Chiayü kuan, continues with the same bearing along the great route and must have served as an effective flank protection for it on the north ; for along the greatest part of its length it is quite impracticable except on foot, and the two narrow gorges in which the Po-yang Ho and the stream of Ch`ih-chin break through it are easily guarded. It was the strength of this natural barrier which first led me erroneously to conjecture that it had been utilized for the line of the Han Limes. In reality this line had been drawn through- the chain of depressions which lie to the north of the range. Want of time and of suitable transport did not allow me to visit them until 1914, and thus it was only then that I was able to trace the Han Limes right through from the Su-lo Ho to below the junction of the Su-chou and Kan-chou Rivers.
Nevertheless it was on this return journey to An-hsi that I succeeded in correctly locating the starting-point of the section of the Han Limes just referred to. On moving to Yii-mên-hsien on September 20, across the absolutely bare gravel plateau which forms the watershed between the Su-lo Ho and the drainage of Ch`ih-chin, I noticed far away to the north a string of hamlets at the foot of the low Pei-shan hills.' As the route map of M. Obrucheff's journey of 1894 seemed to indicate ruined watch-towers in the neighbourhood of that ground, and as the very names of the 'hamlets Shih-tun, ` Tower X ', and Shili-Irk-tun, ` Tower XII', were obviously derived from those towers, I made on the following day a long reconnaissance from Yii-mên-hsien northward in order to visit them. It brought me after a ride of more than i6 miles across a reed-covered basin, periodically inundated by canals and overflow waters from the Su-lo Ho, to the hamlet of Shih-êrh-tun.
I can refer only in passing to the geographically very curious fact that the small stream which accounts for the existence of this hamlet and those lying along its line further on is undoubtedly fed from the Su-lo Ho ; but its easterly course is directly opposite to that followed by the Su-lo Ho itself from the adjacent point of its great bend, as can be seen from the map. We have here a clear instance of bifurcation far away from the terminal delta of the river with which we have become familiar more than 200 miles further west.2 Regarding the interesting archaeological discovery, too, which rewarded this reconnaissance, it must suffice here to state the main facts ; for
' See Map No. 85. A, B. 2.
2 My explorations of 1914 enabled me to follow this stream, locally known as Hsi-wan Ho, on its course to the east where it falls into the drainage area of the Ch`ih-chin River. This itself apparently links up with the marshy basin beyond the oasis of Hu-hai-tzu or Ying-p`an. Thus water derived from the Su-lo Ho finds its way eastwards into a terminal basin hundreds of miles removed from that where the river now ends on its course along the depression leading towards the ancient Lop Lake basin. The essential facts of this remarkable bifurcation. will be illustrated in the new atlas comprising the surveys of all my three expeditions. Exact details about the hydrography of this interesting desert de
pression could, of course, be expected only from a more elaborate survey including careful levelling operations.
The existence of the Hsi-wan Ho, with the small marshes along it and the inundation areas near its head, may be hehl to explain the indication of a large lake or marsh near the Su-lo Ho bend which is to be found in Chinese maps and has been borrowed from them by European cartographers also. Cf. regarding this somewhat problematic feature, Futterer, Geographische Skizze der ü ïisle Gobi, in Pelcrmann's MillheilrnJgen, Ergänzungsheft No. i39, p. 24 ; also Sheet A. 1 of the, maps illustrating Count Széchenyi's expedition (I877-8o) ; Tibel map of R. Geographical Society (r 900), etc.