国立情報学研究所 - ディジタル・シルクロード・プロジェクト
『東洋文庫所蔵』貴重書デジタルアーカイブ

714   230. FEMELES (ISLAND OF WOMEN)

in western Kan-su, and says it is Bokhara. It is true that Bokhara was known as the kingdom of

An c. 600 A. D., and then erroneously thought to be the kingdom of    An-hsi (*•Ân-sink) of
Han times, which was in fact the Arsacid or Parthian empire, and the name of which rendered Arsak (I do not believe in the derivation of An-hsi from «Antiochia [Margiana] », Merv., recently advocated by W. W. TARN, The Greeks in Bactria and India, 281) ; and in the administrative nomenclature established in 658 through a misapplication of ancient names, Bokhara became the district of Anhsi (*•Ân-sjak; cf. CHAVANNES, Doc. sur les Tou-kiue, 136-137). On the other hand, in the Chinese translation made by Fa-ch'êng of the Prophecy of the Li country (cf. PELLIOT and HANEDA'S edition, in Manuscrits de Touen-houang, Kyoto, 1926, No. 2), the 'An-se of the Tibetan text is rendered by the An-hsi (*•Ân-skak) of the Arsacid empire. Yet, I have little doubt that Fa-ch'êng made a mistake. If he adopted An-hsi (*•Ân-siak), it must be on account of a literary reminiscence, because, in much older « prophecies », the Parthian king was one of the three wicked monarchs who were to endanger and almost ruin the Buddhist law. In the Prophecy of the Li country, the conditions, and the names are on the contrary those of middle T'ang times, when the « An-hsi district » of Bokhara was a pseudoadministrative designation which never came into actual use. Moreover, occlusive final consonants were still heard in the 9th cent., and, just as Shu-lo (*Sj\vo-lak) gave in Tibetan Su-lig, An-hsi (*•Ânsjak) ought to be rendered not 'An-se, but *'An-sig. As was said by RocKHILL, 'An-se actually is

An-si (*. kn-siei), but not the An-hsi of western Kan-su, an administrative name which dates only from the Man-chu dynasty; what ROCKHILL alluded to, though misplacing its seat, is the « Protectorate General of An-hsi », i. e. Kuel in Chinese Turkestan, which certainly is the 'An-tse, or better 'An-se, of the two prophecies; Russian Turkestan is out of the question. The same may be said of Gus-tig, for which LAUFER had thought of Kucâ, but which THOMAS states to be « Hon-tö » (read « Hou-tê ») of TP, 1905, 559, N. E. of Samarkand. But Hou-tê was a name of the Han period, no more known in the 9th cent., and here again the scope of the prophecy remains within the limits of Chinese Turkestan : Gus-tig certainly is the place situated between « Aksu » (Aq-su) and Maralbasi (almost at Maral-basi) the name of which is written 411 * 1 Chü-shih-tê (*Kiwo-§i-tak) in Chia

Tan's itineraries (Hsin T'ang shu, 43 B, 14 b),   T F Chü-shê-tê (*Kiwo-§u t-tak) in Wu-k'ung's

itinerary (cf. CHAVANNES and LEvl, in JA, 1895, II, 363), and, erroneously, N   fit Wo-shê-tê

(*•Âk-svt-tak) in Kao Hsien-chih's biography (cf.CHAVANNES, Doc. sur les Tou-kiue, 152). Nor is Parmkhan « Ferghana » (Faryâna), as stated by THOMAS (pp. 47, 61). It renders fa A Po-huan (*PuâtYuân; to CHAVANNES' texts, add Ts'ê fu yüan-kuei, 986, 23 a, and Hsin T'ang shu, 216 A, 3a), also

written   7 Po-huan (*Puât-Yuân; cf. CHAVANNES, Doc. sur les Tou-kiue, 8, 353) and g 4,. Po-
huan (*Puât-Yuân; cf. Ts'ê fu yüan-kuei, 992, 7 a), which I have shown many years ago to render the name by which what is now Aq-su (« Aksu » of our maps) was anciently known to Mussulman writers (cf. TP, 1907, 553-556; 1923, 129-130; 1936, 362). I had started from Idrisi's ~lyl, Babwân; since an Iranian form Parvân is now attested (cf. Mi, 482; BAILEY, in BSOS, ix, 567-568; I have now discarded the connection with Mo-lu [see « Cotton », p. 494-495]), BAILEY has proposed to correct Idrisi's form, always uncertain, to ~Il *Bârwân; but the Tibetan Par-mkhan also has a -h- (-kh-), so that I am tempted to restore in Idrisi JII *Barhwân (_ *Parhwân), as a doublet of Parvân. On the other hand, MINORSKY (Mi, 295) has suggested that Idrisi's Bàhwân may have to be corrected into 314 Bârmân, which seems to be another name of Aq-su in Kâsyari (BROCKELMANN, 241). The