Sec. iii] THE RUINED FORT AND STOPA OF THE ENDERE SITE 435
There is nothing in the ruined fort of Endere, the survey of which we have now completed, directly to indicate the purpose for which it was originally placed here. But the approximate date we have ascertained for its abandonment, when compared with what Hsiian-tsang tells us of the condition of this region in the middle of the preceding century, supplies reasonably safe ground for surmises. When the pilgrim had started eastwards from Ni fang or Niya, he entered ' a great drifting sand desert' 6 the terror and dangers of which for wayfarers he graphically describes, much in the same fashion as Marco Polo, more than six centuries later, described the great desert between Lop Nor and Sha-chou 7. After having travelled through the moving sands for four hundred li, or four marches, he arrived at ' the old country of the Tu-huo-lo was'. This country had long been uninhabited. All the towns presented the appearance of an unoccupied waste 8.
Six hundred li further to the east he reached the old kingdom of Che^-mo-eo-na ffr )-# ACS,
where the city walls still stood high, but there were no inhabitants. This Chê-mo-t`o-na, for which the older name of Chii-me (Tsiu-mo) M. 7f, is correctly given in Hsiian-tsang's ' Life' 8, can be no other than the oasis of Charchan ; the position of the latter is clearly indicated on the one hand by the distance of a thousand li from Ni fang or Niya, which accurately coincides with the ten marches reckoned at the present day between the two places, and on the other by the bearing and distance to Na-fo-po or Lou-lan, which the pilgrim's narrative places to the north-east and at a thousand li. Na-fo-po or Lou-lan has long ago been identified as the vicinity of Lop-Nor, and a look at the map shows that the oasis of Charchan lies to the south-west of the latter and almost exactly halfway between Lop-Nor and Niya.
The relative distances which Hsiian-tsang has recorded to Ni fang or Niya and Chê-mo-t`o-na or Charchan, respectively, oblige us to look for the deserted settlements of what he calls ' the old country of Tu-huo-lo' 10 in the tract surrounding the actual Endere site ; for the latter lies within twelve miles or so of the direct route from Niya to Charchan and, as Dr. Hedin's map shows, some sixteen miles nearer to Niya than to Charchan 11. Seeing that about 645 A. D. this tract was already a waste abandoned to the desert, the question arises how to account for the existence in it of the ruins excavated by me, which were undoubtedly occupied during the early part of the eighth century. Two explanations appear open. Either the tract, perhaps in consequence of the improved conditions following the establishment of Chinese authority throughout Eastern Turkestan, which as we have seen took place within a little over ten years after Hsüan
12, had again come under cultivation and received a settled population. The
Hsüantsang's ' old Tu-huo-lo country'.
Hsüan- tsang's Chê-mo-t'o- na, or Charchan.
Endere fort built in tract deserted at Hsüantsang's time.
6 See Mémoires, ii. p. 247 ; Beal, ii. p. 324 sq.
7 See Yule, Marco Polo, i. p. x96 sq.
8 See Mémoires, ii. p. 247 ; Beal, ii. 325.
See Vie de H.-Th., p. 290 ; Chavannes, Voyage de Song Yun, p. 13, note r. The Hsiyii-chi erroneously reads
Ni-mo. Comp. also Turcs occid., pp. 3o, 57, 306, and for the Han Annals' account of the same oasis, Wylie, ' Notes on the Western Regions,' J. Anlhrop. Insl., x. p. 28.
10 Hsiian-tsang's mention of this tract as ' the old country of the Tu-huo-lo', i.e. of the tribe of the Tochari which played so important a part in the early history of the Oxus region, and the memory of which long survived in the name of Tokharistan, has been the subject of much learned speculation : comp., e. g., Marquart, Erànlahr, pp. 206 sqq.; Franke, Zur Ken/nzss der Tiirkvölker, pp. 28 sqq. A discussion of the various problems connected with the real character and origin of the Tochari (the Tukhara of Sanskrit
texts) does not come within the scope of my task here. But I may mention that, if the conquerors of the Bactrian Greek dominion really came from this region,Dr. Marquart's argument against their having been nomads at the time would find strong support in the physical. character of their alleged old home.
That the term Taklamakan, by which all desert ground within the central area of the Tarim Basin is popularly designated, can neither on linguistic nor on historical grounds be derived from the name Tu-huo-lo (see Hedin, Through Asia, ii. pp. 784 sq.) scarcely needs to be demonstrated to critical students.
" The route followed by me between the Endere Site and Niya, and measured on my map, covers about i10 miles. Between Endere and Charchan Dr. Hedin's map (see Reisen in Z.-A.) shows about 1 to miles as the road distance without allowance for windings.
12 See above, pp. 59 sq.