Sec. v] FROM KARA-TEGIN TO BOKHARA 893
a height over the far-stretched twisting gorge through which the river breaks before it emerges, as the Wakhsh-db, far down towards the plains of the Oxus. Then ascending through a narrow defile we reached the wide upland basin of Ab-i-garm (about 4,200 feet elevation).
Here we have arrived at the westernmost limit of Kara-tegin and also at the end of the alpine portion of that ancient trade route which we first had occasion to trace down the Alai valley. We have already noticed the location of Ptolemy's ` valley of the Kômédai ' in Kara-tegin.3 There only remains for us now briefly to review the references to this territory in Chinese records which conclusively prove this identification to be true. The earliest of them is furnished by the mention which Hsüan-tsang makes of the territory of Chii-mi-t`o. It is contained in his account of the petty States which were comprised in Tu-huo-lo or Tokhàristàn at the time of his outward journey, about A. D. 63o.4 The pilgrim did not himself visit Chü-mi-t`o $pj p é pt, but describes it as
a country situated to the east of K`o-to-lo JpJ p$ . r It is about 2,000 li from east to west, and
200 li from south to north. It is in the middle of the Ts`ung-ling mountains. Its capital is about 20 li in circuit. On the south-west it is near the river Oxus, and on the south it adjoined the Shihch`ih-ni country.' The bearings here indicated of neighbouring territories, together with the great length ascribed to Chü-mi-t`o, leave no doubt that Kara-tegin is meant, as has long since been recognized.5 For K`o-to-lo is certainly identical with the Khottal of early Arab geographers comprising the territory to the east of the lower Surkh-db (Wakhsh-db), i. e. Baljuwân, Kulâb, &c.,6 and Shih-ch`ih-ni is Shughndn, correctly placed to the south.'
Some useful additions to the information recorded by Hsüan-tsang are to be gathered from the Tang Annals. A notice of the Tang-shu inserted between those on Shughnan and Wakhân
tells us that ` [the country of] Chii-mi 3 has its administrative centre in the midst of the
mountains ; it lies to the north-east of Tu-huo-lo ; in the south it is near the Black River ; the king is of the race of the Yen-t`o Turks '. Chü-mi is said to lie 500 li to the north-west of Shih-ch`ih or Shih-ch`ih-ni. Embassies to the Imperial court are recorded in the years A. D. 642, 719, and 742-55.6 Elsewhere we learn from the account given of the administrative organization that the Chinese Government designed after the final victories of A. D. 658-9 over the Western Turks, that
the district of Chih pa tA was established in the town of Ch`u-sê 1E- x in the kingdom
of Chü-mi 3. No indication is furnished as to the position of this town.9
The notices of Hsüan-tsang and of the 'rang Annals supplement each other in a very satisfactory way. In the former the great length of the territory extending along the Surkh-db is quite correctly brought out, while the latter indicates the true bearing of Kara-tegin relatively to
3 See above, p. ii. 849.
4 Cf. Julien, Mémoires, i. p. 27 ; Watters, Yuan Chwang, i. p. Io6.
5 See above, ii. p. 849 ; with reference to Yule, J.R.A.S., 1873, pp. 97 sq., it may be noted that the substitution of Darwaz for Kara-tegin is due merely to the imperfect knowledge then available of the geography of these territories ; also that Kara-tegin was ruled at the time by the chiefs of Darwaz.
The location of Chü-mi-t`o in Kara-tegin appears to have been first definitely indicated by M. Severtzov ; cf. Chavannes, Turcs occid., p. 164, note 1. But see also Tomaschek, Sogdian, 1877, PP. 47 sqq.
6 Cf. Marquart, Ériin-s"ahr, pp. 232 sqq., and in particular the translation there given of the important passage of Ibn Rusta, which correctly describes the course of the Wakhsh-ab, i. e. Surkh-ab, from the land of Kharluk Turks
(Kashgar, &c.) through the Pamir region, the territories of Rasht and Kumédh, and the gorge spanned by the Pul-isangin, to Khottal.
Rasht, which other early Arab geographers (see Marquart, loc. cit., p. 236) mention as situated in a narrow valley through which Turks used to make raids into the easternmost marches of Khorasan, must be looked for in the uppermost portion of Kara-tegin where the Surkh-âb passes a succession of defiles below Kara-muk.
7 The bearing relative to the Oxus is uncertain ; Julien making it south-west, Watters south-east. Owing to the bend made by the Oxus and the great extension of Kara-tegin from east to west either bearing could be accounted for, as the map shows.
8 See Chavannes, Turcs occid., p. 164.
9 Cf. ibid., p. 71, note.
Chii-mi-t`o of Hsüan- tsang.
Kara-tegin in Tang Annals.
Relations between Kara-tegin and Darwaz.