Sec. ii] TURFAN FROM LATER HAN TO TANG TIMES 579
To complete this summary of the notices in the Tang Annals concerning Turfan, the description of Kao-ch`ang given in Chapter CCXXI of the Tang shu may be reproduced here from M. Chavannes' translation : 43 ` Kao-ch`ang is over four thousand li, as the crow flies, to the west of the capital ; it measures eight hundred li from east to west and five hundred li from north to south ; it comprises twenty-one towns ; the king has his capital in the town of Chiao-ho, which is the same as the Court of the Anterior king of Chü-shih at the epoch of the Han ; the town of Tien-ti was the administrative seat of the Wu-chi-hsiao-wei.44 The soil is fertile ; wheat and cereals produce there two harvests every year ; there is to be found there a plant called po-tieh 01_; ; one picks its flower which one can spin in order to make cloth of it.45 It is the [inhabitants'] custom to tie their hair into a plait which hangs behind the head.' 46 The dimensions indicated for the territory probably represent, as is usual in such cases, rough estimates of the distances to be covered on the high roads leading through the territory from east to west and from north to south. Taken in this sense, the measurements are reasonably correct ; for from Chik-tam where the high road from the east first reaches inhabited ground in the Turfan basin, to Manan-chosedawan where it leaves it in the south-west, eight daily marches would be the present customary estimate. Similarly five marches would be reckoned in proceeding from Hsi-yao-tzû, where the most frequented road across the Tien-shan descends from the watershed, to the outermost Kuruktagh range, which constitutes the southern rim of the basin.
The administrative absorption of Turfan into the Chinese Empire has deprived us of the account which Hsüan-tsang's Hsi yû chi would otherwise have furnished of the territory. The great pilgrim had reached it in 63o on his way westwards from Hami, and had been received there with much honour by Ch`ü Wên-t`ai.47 The king had in fact wished to detain him altogether, and in the end consented to release him only on Hsüan-tsang's promise that on his way back he would stay at Kao-ch`ang for three years. But when Hsüan-tsang was returning towards China in 644-5, Kao-ch`ang had ceased to exist as a kingdom, and he was free to travel by way of Khotan and Lop. If Kao-ch`ang thus dropped out of his ` Memoirs of the Western regions ', Hsüan-tsang at least bears witness to the close connexion existing at the time between its ruler and the Western Turks ; for he tells us that a sister of Ch`ü Wên-t`ai was married to the eldest son of Tung Shê-hu, the supreme Kagan of the Western Turks, and that the Kao-ch`ang chief's recommendation to the Kagan had secured for him powerful support on his journey all through the latter's vast dominion.48
The transfer of the An-hsi Protectorate to Kucha in A. D. 658 marked the establishment of a new base for Chinese political activity in Eastern Turkestan. This helps to explain why the abundance of exact and reliable data which M. Chavannes's masterly researches have gathered from the Tang Annals for this period of Chinese expansion in Central Asia, supplies but little information as to the events and conditions particularly affecting Turfan. Between the years 64o and 67o the district may be assumed to have remained in undisturbed Chinese occupation. But whether this continued during the two following decades appears very. doubtful. We know that after A. D. 67o the ` Four Garrisons ' controlled by the Protectorate of An-hsi (Kucha, Khotan, Kashgar, Tokmak) were overrun by the Tibetans, who had in that year won a signal victory over
Turf An described in T'ang shu.
Hsüantsang's passage through TurfAn.
TurfAn between A. D. 65892.
43 Cf. ibid., pp. lox sq.
44 Regarding the identification of Tien-ti with the present Lukchun, cf. Chavannes, Turcs occid., Errata et Corr., p. 310.
43 Obviously the cotton plant is meant, as explained by Chavannes, ibid., p. 102, note z. Its cultivation is still one of the chief agricultural resources of Turfan and its
product a main article of export.
46 See also the imperial decree concerning Ch'ü Po-ya, quoted by the Pei-shih ; ibid., p. 103, note ; also Franke, Inschrift aus Idikutsahri, p. 28.
47 Cf. Julien, Vie, pp. 32 sqq. ; Chavannes, Turcs occid.,
pp. 193 sq.
48 Cf. Julien, Vie, pp. 61 sq.