Sec. in THE LOCATION OF THE ' TOWN OF THE DRAGON ' 295
thought likely to tempt exploration. That popular imagination in the Tarim basin has from early times been always much exercised by stories about ruined towns to be found in the desert, I have repeatedly had occasion to point out."- And the legend current in the Uch-Turfan tract, which places ancient castles full of treasure among the fantastically serrated rocky peaks of Kàka-jade, provides a striking example of the readiness with which this belief attaches itself to natural features i2 Parallels to this process could, no doubt, be found in Western folk-lore also, especially in that of alpine regions.
This identification of the Town of the Dragon derives direct support from what at first might appear a fanciful exaggeration. I mean the huge extent attributed to the town by the mention that it was a long day's journey from the western city gate to the eastern. As a matter of fact, it took us a march of fully eighteen miles from L.I. in a straight line to get past the southern face of the Mesa belt, and this belt, moreover, as Map No. 32. A. 3 shows, extends for about four miles to the south-west of our starting-point.13
Owing to the urgent necessity of assuring rapid progress along what I had recognized as the line of the ancient route, it was to my regret impossible at the time to attempt any close examination of the Mesa-covered area. I am therefore unable to offer any suggestion as to where the canal of which traces were supposed to survive at the scarped foot of the town ' should be placed. Still less can I locate the particular clay ridge from the form of which tradition appears to have derived the name ` Dragon Town ', as recorded in the concluding words of his passage.
But it is interesting to note that in the reference to the wind which ` blowing has gradually produced the form of a dragon ', we have what looks like a correct comprehension of the great part played by wind-erosion in the surface configuration of this ground. Whether the description of ` the form of a dragon, which with its face turned westwards regards the lake ', has any possible connexion with the general NNE. to SSW. bearing of individual Mesas is a question to which even renewed examination of that strange ground might not permit us to give a confident answer. But we shall see presently how fully my own observations confirmed the accuracy of Li Tao-yüan's notice concerning the physical characteristics of the region adjoining the Town of the Dragon, and this must strengthen our belief in the intimate local knowledge possessed by the authority from which he borrowed the statements above discussed, together with the rest of those relating to Lou-lan topography.
SECTION III.—ACROSS THE SALT-ENCRUSTED LOP SEA-BED
The level plain of clay which lay before us at Camp ci offered promise of easy progress ; I was therefore tempted to abandon the assumed bearing of the ancient route which we had followed so far for an almost due easterly course. I was anxious to avoid lengthening unnecessarily the march across the hard salt crust of the ancient sea-bed which I knew lay ahead of us. On the other hand, there was also danger in approaching the glacis of the Kuruk-tagh too closely and thus being led away from the nearest line towards our definite goal, the wells of Kum-kuduk, in the south-east. The morning was cloudy when, on February 28th, we started at daybreak, and the hills of the Kuruk-tagh to the north seemed strangely near. Finally, not without hesitation, I chose as our immediate guiding point the single small prominence rising above the flat plain eastwards, obviously a Mesa.
11 Cf. Ancient Khotan, i. pp. 455, 460 ; Serindia, iii. p. 1234 with note II on earlier notices.
12 See Serindia, iii. pp. 1301 sq.
13 The total length of the Mesa belt from L.E. north
eastwards was wrongly estimated, Serindia, i. p. 423, at ` close on thirty miles ', the correct compilation from the several plane-table sheets not being available at the time of writing.