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

Sec. iJ   MARCO POLO'S LOP AND HSÜAN-TSANG'S NA-FU-PO   321

Shan-shan, and was identified with the Lou-lan of the Han epoch ; for this makes it clear that Charkhlik was already considered in Tang times as the chief place of the whole Lop region, as it undoubtedly is at the present day. The questions as to the original application of the names of Shan-shan and Lou-lan will be considered in detail further on. But as regards the Chinese designation of ` Stone Town ', which is not otherwise known, I may at once state my belief that it may well have been connected with the presence of large stones in the ancient mound known as Tora.20 This stone material, evidently brought from a distance and in Tang times, perhaps, found also in other mounds which have since disappeared, was bound to attract special attention on this alluvial flat which elsewhere only showed a uniform surface of loess or, outside the oasis, fine gravel and sand.

It is regrettable that the Tang records dealing with the Tarim Basin, as far as they have been made accessible by M. Chavannes' translation, do not include .a special notice on Shan-shan or the Lop tract, even though the ` kingdom of Shan-shan ' is twice mentioned along with Kucha, Karashahr, Charchan (Chü-mo), etc., among the territories of Eastern Turkestan which, about A.D. 639, acknowledged allegiance to a certain paramount chief of the Western Turks.21 But fortunately Hsüan-tsang's route on his return from Khotan to China in A.D. 644-5 passed through this district, and to this fact we owe the brief but interesting notice which is the very last in the pilgrim's long itinerary.22 From the old territory of Che^-mo-t`o-na or Chu-mo, corresponding to Charchan, ` he continued his journey ', so the Hsi-yii-chi tells us, ` for about a thousand li to the north-east, and

arrived at the old territory of Na-fu-fioatj   ft, which is the same as the old Lou-lan country'.

That the locality thus briefly mentioned corresponds to the present Charkhlik is made clear by the distance and bearing, both of which agree with the present route from Charchan to Charkhlik.23 There was no other practicable route which the pilgrim could have followed on his way from Charchan to Sha-chou 24 To the importance to be attached to Hsiian-tsang's identification of his Nafupo with the ancient Lou-lan I shall recur presently. But what calls for our immediate attention is the name Na fu j'o itself. In no other Chinese record is this designation met with, a circumstance which naturally suggests the assumption that Hsüan-tsang has preserved here, as often elsewhere, the phonetic rendering of a name in actual local use, and not some traditional or literary appellation.

No satisfactory etymology has been given by Hsiian-tsang's commentators for the name which the Chinese transcript Nafupo was intended to render.25 The conjectured original form *Navapa may ` look like Sanskrit ', but certainly has no meaning. Sir Henry Yule was guided by a happier philological instinct when he surmised that a derivative of the indigenous name meant by Hsüantsang's Na-fu-po may be found in the Lop of Marco Polo and modern local nomenclature.26 Striking confirmation for this conjecture has been furnished by the Tibetan records on wood and paper which my excavations brought to light in such abundance from the ruined fort of Miran. The origin, date, and general character of these records will be discussed in a subsequent chapter.27

R0 See above, p. 314.

" Cf. Chavannes, Turcs occid., pp. 30, 57.

n See Julien, Mémoires, ii. p. 247; Watters, Yuan Chuang, ii. p. 304.

23 The identity of Hsüan-tsang's Na-fu-po with Charkhlik appears to have been recognized by M. Grenard (see Mission D. de Rhins, ii. p. 61) and probably by others also; but I cannot find any definite statement of it before my reference in Desert Cathay, i. p. 343.

R4 Sha-chou is distinctly mentioned in Hsüan-tsang's biography as the first town reached by the pilgrim within the

1374   T t

Chinese border; see Julien, Vie, p. 290; Beal, Life, p. 212. The mention there made of the change from camel and pony transport to carts before reaching Sha-chou is significant, and points distinctly to the desert route having been followed.

25 For suggestions, cf. e.g. Julien, Mémoires, ii. p. 247 ; Beal, Si yu-ki, ii. p. 325, where Mr. Kingsmill's bold equation of *Navapa = *Navapura, i.e. Neapolis, is quoted.

70 Cf. Yule, Marco Polo, i. p. 197, note ; Wylie, J. Anthrop. Inst., x. p. 24, note.

27 See below, chap. XII, sec. v.

Hsüantsang's mention of

Nafupo.

Na-fi-po corre- sponds to Charkhlik.

Nafupo

transcribes

Lop.