Sec. 1] BY THE DESERT EDGE TO KHOTAN 97
inquiries, discover no trace ; and as a fairly wide belt of desert in that direction shows jungle growth and is regularly visited by hunters, we may attach some value to this negative evidence a.
These general topographical considerations, combined with the evidence of the archaeological remains to be noticed, convince me that the present route between Karghalik and Khotan practically coincides with what throughout historical times must have been the direct and most frequented line of communication between Khotan and the regions westwards. The earliest historical reference to it we possess is furnished by the Ch`ien Han shu, which in its ` Notes on the Western Regions ' places So-ch`ê (Sha-keu) or Yarkand at a total distance of 770 li from Khotan, the direction being first west for 380 li to P`i-shan (Pe-shan) and thence northwest for 390 li to So-ch`ê e. The total distance here indicated agrees remarkably well with the eight marches reckoned at present for the journey between Yarkand and Khotan, a distance of about 192 miles by the map'. The bearings equally agree, the route from Khotan to the oasis of Gama, which plainly corresponds to P`i-shan, bearing generally west by north or west-north-west, while the bearing of Yarkand to Gama is almost exactly north-west.
.Fa-hsien and Sung Yün may both be assumed to have travelled by this route westwards; but the records of the journeys of neither furnish any details s. The earlier pilgrim and his party are said to have taken fully twenty-five days before they reached Tzû-ho (Tseu-ho), in the neighbourhood of the present Karghalik 9-a superabundant allowance of time, about an explanation of which it would be useless to speculate. During the Tang period this route from Khotan westwards must have seen many a traveller like Wu-kung, of whose passage through Khotan Chinese records have preserved some notice. But the only accounts of the route itself which I can trace—and they are brief—are to be found in an itinerary of the Tang Annals and in Hsüan-tsang's narrative.
The former, extracted by M. Chavannes, acquaints us with a number of localities and certain distances between Yü-t`ien (Khotan) and Kâshgar 10 ; but unfortunately none of the local names appear at present to be traceable elsewhere, and consequently their identification cannot be certain. Restricting ourselves to that portion of the itinerary which must correspond to the route under discussion, we note that it takes us from Khotan first to the Wei passage (Z gg) situated 50 li to the west. Thence after a further journey of 620 li, first westwards and
then to the north-west, during which Po-hai and the river I-kuan ggr are passed,
the town of Chik-chik-manntAttit is reached, also called the district of Chi-nan M 3+1
Antiquity of present route.
The route described in Tang Annals.
i The fact that the profession of ' treasure-seeker ', so common about Khotan, is not represented in the oases west of it, points significantly to the same conclusion. Whatever coins and small antiques were brought to me had avowedly been found at the old sites examined by me close to the present oases.
e Comp. Wylie, J. Anthrop. Inst., x. pp. 3o sq., where, however, the distance from P`i-shan to So-ch'ê (Sha-keu) is given as 38o li. The Hsin-chiang-chih-lio, as translated by Julien, J. as., x846, viii. p. 245, in discussing the passage of the Annals, states the distance from P`i-shan to So-ch'ê (So-kiu) as 390 li, which seems more accurate. Regarding the identity of P'z-shan (P'i-chan) with Güma, see below,
7 These marches are not of uniform length, nor are their stages fixed by necessity, account having to be taken of the
provision of water, the facility of supplies, and similar considerations important in a journey mainly over barren ground. The most convenient stages at present for a traveller starting from Yarkand would generally be Jilkucha (Yakshamba Bazar), south of Posgâm, Akim-Langar, Chôlak-Langar, Guma, Zanguya, Pialma, Zawa, city of Khotan. It must be borne in mind that the old capital of Khotan lay at Yôtkan, some six miles further to the west than the present city.
8 See Travels of Fd-hien, p. 21; Chavannes, Voyage de Song Yun, p. 19.
9 See above, p. 92.
10 Comp. Turcs occid., p. 123, note 1. The same notice, in a less accurate form, had been translated by Julien from the Hsin-chiang-chih-lio, together with certain comments of the author of this compilation, which, however, are of little value ; see J. as., 1846, viii. p. 245.