ro92 THE MARCHES "OF OLD KUA-CHOU [Chap. XXVI
river which any practicable canal system could command is limited to a triangle roughly 16 miles long and about ro miles wide at its base, it is also certain that the facilities for irrigation furnished here by the Su-lo Ho are greatly inferior to those which are enjoyed by Tun-huang owing to the volume of the Tang Ho and the favourable position of the oasis on a large and fertile alluvial fan.
Great as the drainage area of the Su-lo Ho is, and imposing as are the glacier sources which feed it and which we were able partially to survey in August, 1907, yet much of its volume is lost by evaporation and otherwise on the wide glacis of gravel below Chang-ma and on its long course through arid wastes between Yü-mên-hsien and the canal heads above An-hsi. At An-hsi itself I found, on June 19, the river reduced to the appearance of an insignificant sluggish watercourse, about 20 feet wide and less than 2 feet deep in the middle. At the same time I could see from the width of the dry bed, some 200 yards across, in which this watercourse was meandering, and from its steeply cut banks, 15-2o feet high, how great the floods are which the Su-lo Ho carries down early in the spring after the first snow melts in the mountains, and again in the late summer when the big glaciers of the Stress Range discharge their full quota. It is clear that such conditions must often interfere with the maintenance of canal heads and the provision of an adequate water-supply at the critical seasons, and difficulties on this score were acknowledged by the district officials.
From the information they were able or willing to give me, it appeared that the total population of the An-hsi district was then reckoned at about 90o households. But even if this figure was not exaggerated, it must be remembered that in it were also included several small oases higher up the river, such as Hsiao-wan and Shuang-t`a-pao, as well as a few relatively flourishing villages in the lower hills, two of which I was subsequently able to visit. From all this it seems safe to conclude that Kua-chou even in ancient times must have ranked considerably below Tun-huang in economic resources and importance.
This conclusion is supported by all early references to Kua-chou that are accessible to me, as they show it in close political connexion with, or dependence on, the territory of Tun-huang. Not being able to consult the special notices that the Chinese historical sources are likely to contain concerning Kua-chou, I must be content with pointing out that in Han times the command of Tun-huang must obviously have included it, and that the same may be also assumed regarding the arrondisse-
ment of Sha-chou , i. e. Tun-huang, which was organized in A. D. 345 by Chang Chun,
a local ruler of western Kan-su." With the interesting part which Kua-chou played in the story of Hsüan-tsang's start on his great journey, and to which we owe the earliest mention of the place I can trace among texts accessible to me in translation, I shall have occasion to deal presently. The reference made in the Chien-fo-tung inscription of A. D. 894 to a prefect of Kua-chou who was the grandson of Chang I-ch`ao, the local chief of the Tun-huang region in A. D. 850, and whose elder brother held the prefecture of Sha-chou, proves that both tracts at that period continued to be governed by the same local family.12 Later still the Chinese envoy Kao Chü-hui on his mission to Khotan, A. D. 938-42, found both Kua-chou and Sha-chou occupied mainly by Chinese, and both under a local chief of the Ts`ao family.13
In discussing above the historical records concerning the extension of the ancient Chinese Limes beyond Ton-huang, I have already emphasized the importance attaching to all oases, big or small,
" Cf. M. Chavannes' note on extracts translated by him from the Chin Annals, in his Appendix to Ancient Kholan, i. p. 543, note 4.
The Han Annals, when describing the organization of the frontier territories conquered by the Emperor Wu-ti into the commands of Chiu-ch`üan, Wu-wei, Chang-yi, and Tun-huang,
do not specify Kua-chou, it being manifestly a part of Tunhuang; see Chavannes, Documents, p. v; above, pp. 724 sq. 1z Cf. Chavannes, Dix inscriptions, p. 93; also pp. ro,
13 Cf. Rémusat, Ville de Kholan, p. 77 ; Chavannes, Dix inscriptions, pp. 12 sq.