J38 HISTORICAL NOTICES OF LOP, SHAN-SHAN, AND LOU-LAN [Chap. IX
gate for Hun inroads into the Tarim Basin.15 Hun parties sent thence to intercept Chinese missions could best effect their object on that part of the road which led westwards through Lou-lan, assuming that by this term is meant, as by the mediaeval and modern Lop, the territory comprising the whole of the Tarim delta around Lop-nor. We have probably a definite instance of such a move in the record of the attempt which the Huns made in 104 B.C. to cut off the return of a Chinese expedition to Ta-yuan by a force of cavalry posted in Lou-lan."a
But this route must have been equally exposed to Hun raids from the side of Turfan. If reference is made to the map illustrating the explorations effected by Colonel Kozloff between Turfan and Lop-nor in the course of his and Captain Roborovsky's expedition in 1893-5, or to our own Surveys of 1914-15, it is seen that there are at least three still practicable tracks which cross the wastes of the Kuruk-tagh due south of Turfan. All meet at right angles the ancient route which here approached the southern foot of the range. It is true that probably several of the scanty springs upon which the use of these tracks depends are now salt, like those of Altmish-bulak to the north-east of the Lou-lan site,16 and consequently can only be utilized when their water becomes drinkable through freezing. But these difficulties about water must have been less serious in ancient times, before desiccation had progressed as far as it has now, and it appears to me probable on several grounds that this central portion of the Kuruk-tagh could then be crossed by small parties." In this way the ancient route leading from Tun-huang to the north of Lop-nor is likely to have been exposed to flanking attacks by Hun raiders here also.
In this connexion convenient reference may be made to a passage of Chang Ch'ien's report, as contained in Chapter CXXIII of Ssû-ma Ch'ien's history, which also connects Ku-shih, i. e. Turfan, with Lop-nor. In a summary geographical description of the Tarim Basin we are told that ` adjoining
the salt marsh (i.e. Lop-nor) were the states of Lou-lan and Ku-shih ~, the plains
outside the cities of which reached to the water of the lake ' (Kingsmill).18 M. Chavannes translates
les royaumes de Leon-lava et Kou-che ont des villes munies de remparts intérieurs et extérieurs et sont voisins du marais salé', thus stating that both territories extended to the vicinity of the lake. But in his explanatory remarks he specially emphasizes the fact that the wording of the Chinese text does not imply that the capitals of the two kingdoms were necessarily situated close to the lake.19
In view of obvious geographical facts and of what has been stated above as to the identity of Ku-shih with Chü-shih or Turfan, we must, as far as the latter territory is concerned, accept M. Chavannes' explanation that by the vicinity of Lop-nor the text merely means easy access to it by routes. And that this actually existed I have, I hope, made quite clear by my discussion of the topography. As regards Lou-lan a much closer vicinity to the ` salt marsh ' must be assumed ; but the passage of Chang Chien's report does not help us to define the extent of the territory and the position of its chief place more exactly.
'b It is significant that according to a notice in the Former Han Annals (cf. Wylie, J. Anthrop. Inst., x. p. a r) the Slaves' Protector-General ' appointed by a Hun prince, before Chinese supremacy had asserted itself, ' to rule the Western regions ... always dwelt in the dangerous part of Yen-ch'i (Kara-shahr). He had to levy the taxes on the cultivated land, and received of the wealth of these kingdoms.'
See below, p. 339•
15 Cf. for the ` Sixty ' springs of Altmish-bulak. Hedin, Central Asia, i. p. 366 ; ii. pp. 97 sq., 108.
17 For a similar inference as regards a portion of the eastern Kuruk-tagh, see below, chap. xix. sec. vi.
's Cf. Intercourse of China with Eastern Turkestan, J.R.A.S., 1882, p. 6.
19 See Chavannes, Toung pao, 1905, p. 533, note. That M. Chavannes claimed the benefit of this interpretation also for the erroneous location of Lou-lan at Pichan or Lapchuk (cf. above, p. 336, note 13) does not invalidate the intrinsic correctness of his explanation.