Ridges Sands ', San-lung sha =ma W.? The reference in the Wei lio makes it clear that these designations were applied to the northernmost offshoots of a belt of high sand dunes which the caravan track coming from Tun-huang crosses, just as did the ancient Lou-lan route, a short distance to the north-east of Bésh-toghrak (Map No. 35. B, C. 4). Now our surveys of 1914 have established the fact that the big south-eastern bay of the ancient sea extended continuously within the ` Bésh-toghrak valley ' as far as the vicinity of Yantak-kuduk (Map No. 35. A. 4). Thence a succession of shôr-covered areas or actual salt bogs extends in the same depression close to the west of the wells of Bésh-toghrak (Map No. 35. A. 4; B. 3, 4). The accuracy of the statement in our passage as to the limits of the region with which we are concerned is thus vindicated by plain topographical facts.
It is not possible for us to test in the same conclusive fashion the statement made at the beginning of the passage : ` This region has an extent of a thousand li ; ' for we are not definitely told the direction to which this measurement applies. But it is certainly noteworthy that the recorded estimate' agrees remarkably well with the distances measured along either of the two possible routes by which a traveller starting from the northernmost point of the ` Three Sands ', east of Bésh-toghrak, could attain the nearest habitable ground beyond the western extremity of the sea-bed. As long as the Lou-lan area received water from the Kuruk-daryâ. and was capable in places of human occupation, the western edge of the great salt waste which the Chinese knew by the name of P`u-ch`ang or Yen-tsê, ` the Marsh of Salt ', might be reached at two different points, either by proceeding to Lou-lan by the ` route of the centre ' or else by following the still practicable caravan track to Mirân along the southern shores of the sea-bed.
Now the aggregate of my marches, as measured by cyclometer from L.J., the last outpost of Lou-lan, to the northernmost of the ` Three Sands ' east of Bésh-toghrak, along what I believe to have been the approximate line followed by the ` route of the centre ', amounted to 199 miles.8 If the traveller were to start from the same point east of Bésh-toghrak and to take the southern route still followed by caravans between Tun-huang and Mirân, a total marching distance of practically identical length, as measured by us with the cyclometer in February, 1907, viz. 198 miles, would bring him to Camp 143 of our second journey. Near this we came upon the extreme south-western edge of the hard salt crust area of the ancient sea-bed. From there to the west towards Mirân, desert vegetation in gradually increasing proportion is found along the track, and the region of bare salt, clay, or gravel is definitely left behind.9 Experience gathered on my explorations in the Tarim basin and adjacent regions has abundantly proved that one mile there may ordinarily be reckoned as equivalent to 5 li in Chinese records of distances over level ground.'° Thus from whichever of the two routes the measurement of a thousand li might have been derived by Li Tao-yüan's authority, this estimate of distance would appear surprisingly correct.
For two reasons it seems to me more probable that the recorded estimate of extent was taken from the Lou-lan route. In the first place the whole of the topographical data mentioned in Li Tao-yiian's notice relates to the ground traversed on this route.1' Secondly, we have in the conclud-
Extent of ` Salt Marsh ' region.
Li Taoyüan's distance estimate.
Li Taoyüan on ` N. limit of lake '.
See Serindia, i. pp. 418, 425 ; ii• pp. 555 sq.
8 In this reckoning due allowance has been made for the detour implied by our having proceeded from Camp cvii first to the route on the southern side of the Bésh-toghrak valley and having followed this as far as Yantak-kuduk (Map No. 35. A. 4) instead of moving direct to Bésh-toghrak ; see below, pp. 317 sq., 321 sq.
The distance of 230 miles indicated in Serindia, i. p. 425, note 38, referred to the marches from Camp c to Camp cxii,
farther to the east of Bésh-toghrak, and for the detour between Camps cvii and c g See Map No. 3o. B, C. 2.
10 Cf. Serindia, ii. p. 735, with the in note 28 a.
11 The reference to travellers crossing with hard salt is certainly significant. route it is only skirted.