Sec. iii] ACROSS THE MOUNTAINS TO TURFAN 565
tain range, and find definite evidence that in Uigur times, also, the direct route connecting Turfân with Pei-t`ing led up the valley above Yoghan-terek and across the Pa-no-pea pass.
It is equally easy to recognize the correctness of what the Tang itinerary tells us of the remaining section of the route between the Chin-sha ling and the town of the Protectorate of Pei-t`ing. It is true that we have no means of definitely locating the ` frontier post of Shih-hui ', though topographical reasons point to its having probably stood somewhere near Ch`üan-tzil-chieh or the mouth of the Pa-no-pea valley. But the 16o li recorded by the itinerary as the distance between ` the Mountain of the Golden Sand ' and Pei-t`ing agrees closely enough with the actual road distance of 45 miles shown by our traverse reckoning, assuming that it is measured from Hsi-yao-tzû and by the direct line connecting the route below Ch`üan-tzti-chieh with the ruined site near Hup`u-tzii.e
It only remains for me to point out that the total of 370 li indicated by the Tang itinerary as the distance from Chiao-ho or Yâr-khoto to Pei-t`ing indirectly helps to make it appear highly probable that the route via Yoghan-terek and Pa-no-pea was already in Han times that regularly followed between Anterior and Posterior Chü-shih. In the Later Han Annals we are told that
` going from the retrenched camp of Kao-ch`ang A-7 fit northward one reaches after 500 li the
town of Chin-man + of the Posterior tribe. These two localities are the gates of the Western
countries.'' Now ` the retrenched camp of Kao-ch`ang' can with certainty be located at the present Kara-kh6ja.8 The distance from this to Yâr-khoto or Chiao-ho is fully twenty-six miles as measured on the map (No. 28. B, C. 3) and by road may safely be put at thirty. Chiao-ho or Yâr-khoto lies quite close to the direct line connecting Kara-khaja with the route leading to Shaftulluk, Yoghan-terek, &c., and from all that we know of Chinese itineraries in the Western countries it appears most probable that the road distance recorded by the Later Han Annals was obtained by first reckoning the distance from Kao-ch`ang to the political capital at Yâr-khoto and then adding to it that from the latter to Chin-man. The position of Chin-man is definitely proved to have been the same as that of Pei-t`ing.8 Hence we are justified in adding the 120 li—which, at the above-mentioned rate of four li to one mile, correspond to 3o miles—to the 370 li reckoned between Chiao-ho and Pei-t`ing. The resulting total of 490 li is practically the same as the rough figure of 500 li which the above-quoted passage of the Later Han Annals names as the distance
between Kao-ch`ang and Chin-man.
6 I mean the nearest route that a traveller now bound for the site would follow. He would descend first via Pano-pea and Ch`üan-tzû-chieh to the Guchen—Urumchi high road and thence move straight towards Hu-p'u-tzü without
7 See Chavannes, T'oung-pao, 1907, p. 169.
8 See below, p. 571.
9 See Chavannes, Turcs occid., p. xx ; above, p. 555.