Sec. ii] SHAN-SHAN BETWEEN TANG AND HAN TIMES 327
town site to have been deserted by the fourth or fifth century at the latest, i.e. in the period which elapsed between the redaction of the Former Han Annals and that of Li Tao-yilan's commentary. On the other hand, the location at Charkhlik of I-hsün or the new capital of Shan-shan, as a Chinese commentator of the Cla`ien Han shu has rightly called it by inference,14 is strongly supported by the evidence that Charkhlik has been the chief place of the Lop district from Hsüan-tsang's time onwards. It seems reasonable to suppose that the local conditions were not essentially different in Li Tao-yuan's time, only little more than a century earlier.
[Since the above was written an important paper by M. Pelliot (1916) has discussed two texts of Tang times which seem to prove that at that period Chinese local belief identified I-hsün with Mirân.'4a One is a passage of the itinerary preserved in the Hsin Tang shu, from which we have already quoted the notice concerning ` the garrison of the Stone Town' or Charkhlik. Immediately before this we are told : ' From the southern shore of the P`u-ch`ang sea (Lop-nor) one
passes the town of the seven military colonies which is the town of I-hsiu ffr of the Han.
Then to the west [of this town] after 8o li one arrives at the strong place of the Stone Town (Shih-cla`êng-chên), which is same as the kingdom of Lou-lan of the Han and is also called Shan-span. It is 300 li to the south of the sea P`u-ch`ang (Lop-nor).' M. Pelliot convincingly accounts for the graphic confusion between the forms I--hsün and I-hsiu and quotes two further passages in which the same locality is mentioned under the latter form of the name. They are found in a geographical text dated A. D. 885 and contained in the MS. Ch. 917 which I recovered from the walled-up temple library of the ' Thousand Buddhas ' near Tun-huang. There the ' Town of the military colony ' is twice referred to as identical with the town I-hsizt and as situated i 8o li to the east of the town of Shan-shan, i.e. the town which after A.D. 675 was called the ' strong place of the Stone Town ' (Shih-ch`êng-chen).
M. Pelliot has already pointed out that the name ' Town of the military colony ' is derived from the Chinese military colony which, as the record in the Former Han Annals discussed below shows, was placed at I-hsün (or I-hsiu) in 77 B. c.14v Taking into account the eastern bearing from the ` Stone Town ', i.e. Charkhlik, and the distance of i 8o li indicated (the 8o li of the Tang itinerary is clearly only a graphic error), I am led to conclude that the locality intended by the two texts must be the site of Miran. It is clear that this location of I-hsün (I-hsiu) cannot be reconciled with the one which is inferred above from Li Tao-yuan's text. If we adopt it, Yü-ni, the old Lou-lan capital, would have to be placed at Charkhlik,'4c and this is directly contradicted by the bearing which Li Tao-yuan records for it. The fact that his text goes back to the beginning of the sixth century A.D., if not earlier, while the two Tang texts belong to the ninth century, may explain the discrepancy, but it does not settle the question which of the two locations (Miran or Charkhlik) was the right one. The Lop region had passed out of Chinese control for a long period before the Tang reconquest of the Tarim Basin, and this may well have brought about a confusion of the historical nomenclature. It must also be remembered that at the time when the texts just quoted were written, in the ninth century A.D., Chinese rule over the Lop region had long yielded to Tibetan invasion.]
Li Tao-yuan's notice, as far as it concerns us here, concludes with a statement about the lake which receives the waters of the Tarim. It has its interest for the much-debated question as to the ancient
14 Cf. Chavannes in T'oung-pao, 1905, p. 537, note 2, quoting Hsü Song.
"s See Pelliot, Le ' Cha Tcheou leu lou fou fou king' et la colonie sogdienne de la région du Lob Nor, in J. Asial., 1916, janvier-février, pp. r z6 sq. Cf. also below, M. Chavannes'
Appendix A, H.
'b See below, p. 342.
10 Cf. for this view, first suggested by M. Grenard, Herrmann, Se denslrassen, p. 100.