Sec. iii] A TANG ITINERARY FROM AK-SU TO KASHGAR 839
for the most part through well-cultivated village tracts, and a short one next morning along the orchard-lined banks of the river, brought me back once more to the hospitable shelter of Chini-bagh, the British Consulate General at Kashgar, and the base of all my Turkestan journeys.
After this brief survey of the line actually followed by the high road from Ak-su to Kashgar,
we may turn to the only early account that I am able to trace, of the route connecting these two places. It is furnished by a Chinese itinerary contained in the Tang Annals to which M. Chavannes makes a passing reference,18 and of which Dr. L. Giles has very kindly provided me with a transla-
tion. This itinerary runs as follows : ` From Po-huan: and Sui-shê you go south-
west and cross the Hun river j 180 li from here is Chi-cho Kuan ;j, the ancient
Ho-Ong P`u %tri 2F . Continuing you pass through the old city of Ta-kan , and after
I 20 li arrive at Yeh-chê Kuan og. 6o li farther on you reach the city of Chü-shih-tê tie
in the territory of Kucha R. One account says that Yü-t`ou Chou 344 is on the
Ku-shih (Orphan Rock) Hill w on the north bank of the Cheih Ho (Red River)
Crossing the Red River and passing Mount Ch`i j, after 34o li you arrive at Chia-lu Kuan
N. Farther on you pass the city of Ta-man to, and after 140 li arrive at Su-lo Chên
j j , which is surrounded by mountains on three sides, south, north, and west.'
The starting and closing points of the itinerary are definitely fixed. There can be no doubt
that Po-huan V. corresponds to the present Ak-su, and Su-lo (Su-18) a Ay] to Kashgar ;
while the mention of Sui-shê 4V, a town which, as preceding passages of the text clearly show, must be looked for in the vicinity of the modern Tokmak, may, according to M. Chavannes, be considered a faulty interpolation. Nor is it possible to doubt that the itinerary describes a route which, in its general direction, agrees with the present high road ; for in the first place the southwesterly bearing of the latter at the start is correctly indicated, and in the second the only alternative route which might come into consideration—that which leads first westwards to Uch-Turfan and thence up by the Taushkan river and across the hills past Kara-jol and Sughun to Kalta-yailak and so on to Kashgar 19—is, in its initial portion, separately described in a preceding passage of
the same Chinese text.20 The Hun river , the crossing of which is mentioned at the outset,
can safely be identified with the united Ak-su river ; for this is clearly indicated by two other passages in the same text of the Tang shu which mention the river by its full name of Ssû-hun
Ho ~p~ and correctly place it to the south of Po-huan or Ak-su and its river, the present
But once beyond this safe start on the route leading towards Maral-bashi, we are confronted
by uncertainties as regards the identification of the localities successively named in the itinerary. None of them are to be found in the other Chinese texts accessible to me in translation, and it is impossible to place much reliance on the distances indicated between them. If judged by their aggregate, 84o li, they are hard to reconcile with the actual distance of about 3o1 miles between Ak-su and Kashgar, as measured on our marches by the present high road, and the ratio of 5 li to the mile which evidence derived from the other Chinese itineraries in and near the Tarim basin indicates as that to be adopted for road measurements of the Tang period.22 At the same time we cannot feel sure whether the enumeration of distances is complete ; for at least two places are named in the itinerary without mention of separate distance measurements. For these reasons the following
18 Cf. Chavannes, Turcs occid., p. ro, note. The itinerary is quoted by Dr. L. Giles from Hsin Tang shu, XLIII B. p. 18.
19 See Maps 7. A. 3 ; B, C. 2 ; 4. B, C. 4, D. 3 ; 5. A, B. I.
20 See Chavannes, Turcs occid., p. 9 ; also Serindia, iii. pp. 1299 sqq.