Sec. iv] EARLIEST RECORDS OF LOU-LAN UNDER THE FORMER HAN 337
With this point accepted, a careful comparison of the evidence furnished by the above record with what we know of the actual topography of this region leads to several highly probable inferences. In the first place, it is clear that Lou-lan, if it was necessary to subjugate it before Ku-shih or Turfân, must be located around Lop-nor, for only here could physical conditions during historical times have permitted of the existence of settlements which might form a `kingdom and serve as a base for an advance upon Turfân. The latter was accessible for a Chinese force operating from the Tun-huang Marches only by the route leading past Lop-nor. For of another, crossing the desert range of the Pei-shan much further to the east, we know, from a definite statement of the Former Han Annals, that it was not opened until the period A. D. r-5.14 Hence it follows that the Chinese expedition proceeded via Lop-nor, and that the Lou-lan of that period must correspond roughly to the Lop tract, as we have also shown of Shan-shan.
In the second place, since the ' high road ' referred to in connexion with the events preceding the Chinese expedition of io8 B. C. was liable to obstruction both from the side of Lou-lan and of Ku-shih (Turfân), the route meant must have lain to the north of Lop-nor. The only line here available for the early Chinese missions was the route which once connected the site of ` Lou-lan oil the east, with the end of the Great Wall and, on the west, with the northern string of oases in the Târim Basin.14a Reference to the map shows that this line of communication must have been exposed to Hun raids both from the side of Kara-shahr and from that of Turfân due north, and it is exactly this condition of things which the record concerning the expedition of io8 B.C. indicates.
Kara-shahr offered easy access from the great grazing tracts north of the Tien-shan and in the Yulduz Valley which were held by the Huns, and must always have been a particularly convenient
Lou-lan as base against
Route obstructed from Loulan and Turfân.
Hun raids from Karashahr.
also by M. Chavannes, which would place Shan-shan at the modern Pichan, east of Turfân, or else at a locality, called Na-chih gAge, corresponding to the present Lapchuk northwest of Hâmi, is quite untenable. The geographical arguments against such a notion are so strong and have been so clearly stated by Dr. Herrmann that it does not seem necessary to discuss in detail the Chinese statements, mostly modern, upon which this conjectural location was based, and which M. Chavannes' note on the Wei tio, Toung-pao, 1905,
pp. 531 sq., reproduces.
With reference to Na-chih, however, there is a point of interest, which deserves to be noted here, because it connects the place with Lop and explains how the erroneous identification probably arose. The Tang shu, indeed, says that Nachih was established in A.D. 63o at the site of the ancient town of Shan-shan. But in view of what the same Annals state elsewhere about the position of Lou-lan or Shan-shan being marked by Shih-ch'êng, the ` Stone Town ', to the south of Lop-nor (cf. above, p. 320), it is obvious that, as Dr. Herrmann has rightly recognized, the record given by an earlier text, the Yuan ho chili: hsien t`u chih, is more deserving of attention. According to this text, which was published between A.D. 8o6 and 814 (see Chavannes, toc. ca.), the sub-prefecture of Na-chih was 120 li to the south-west of I-chu (Hâmi). This town was built by people from Shan-shan. Since the barbarians give to Shan-shan the name of Na-chih, this same name was applied also to the sùb-prefecture.'
The true explanation of this passage has been found by M. Pelliot. In 1910 he was kind enough to point out to me
that, as the locality meant: by Na-chih is undoubtedly the present oasis of Lapchuk, some thirty-three miles to the west-north-west of Hâmi (see Map No. 69 ; for its old remains, cf. below, chap. xxvili. sec. iii), the statement recorded by Li Chi-fu, the author of the Yuan ho chitn hsien !'u chih, obviously refers to a connexion between the name Na-chih and the old indigenous designation of the Lop region which we find reproduced already in Hsïtan-tsang's Nafu po. We have fresh evidence here for the antiquity of the name Lop, which we have previously traced in Hsuan-tsang's Na-/u-pc' and in the Nob of the Tibetan records from Mirân, and confirmation also for the correspondence between the Chinese transcription na and the initial 1 of the indigenous form of the name, as indicated in our discussion of the name (see above, pp. 321 sq.). [M. Pelliot has now fully explained this view and supported it by an interesting notice of Na-chih drawn from our Ch'ien-fo-tung MS. Ch. 917, in/ As., 5916, pp. 11I sqq., note.]
" Cf. Wylie, Notes on the Western Regions, J. Anthrop. Inst., xi. p. 509 ; also Chavannes, Toungpao, 5905, p. 533, note s. The exact line of this ` new road ' will have to be determined elsewhere ; cf. below, chap. xix. sec. vi. In the Later Han Annals we find the still later route from Tunhuang via I-wu (Hâmi) to Kao-ch'ang or Turfân, which was opened after A.D. 73, quite correctly described as a main road to the ' Western Countries ' alternative to the route via Shanshan ; see Chavannes, Heorc Han chou, T'oungpao, 5907, p. 569.
"a See below, chap. xiv. sec. ii.