Tien-shan to Marâl-bâshi (Chap. III. sec. i, ii). I then endeavoured to traverse from that point the great `sand ocean' of the Taklamakân in a straight line to the Mazâr-tagh hill on the Khotan river. The attempt, however, was baffled, after trying marches, by the formidable sand ridges that we encountered, but not before I had been able to secure definite evidence of the geographically important fact that an ancient range, now completely effaced by wind-erosion, had at an earlier period joined those hills to the isolated rock islands around Maràl-bâshi. Travelling along theYârkand river and then up the dry bed of the Khotan river, I reached the Mazâr-tâgh hill, where numerous Tibetan documents were recovered near a ruined fort, and the remains of a Buddhist shrine were traced (Chap. III. sec. iii, iv).
Having regained my old base at Khotan I secured there a considerable collection of small antiques from the ancient capital and other old sites of the oasis (Chap. IV. sec. i, ii). Rapid as was necessarily the journey of close on 700 miles to the Lop Desert, the main goal of that winter's explorations, it allowed me to revisit the areas of ancient settlements abandoned to the desert beyond Domoko and the termination of the Niya river. In the latter area I succeeded in supplementing my former discoveries by observations and finds of distinct antiquarian interest, the latter including a further collection of Kharosthi documents on wood (Chap. IV. sec. iii, iv). Having reached the last inhabited ground towards the Lop Desert by the beginning of 1914, we explored two small sites to the south of Charkhlik, and then, resuming work at Mirân, recovered early frescoes and other remains from Buddhist shrines of the ` old eastern town ' of Shan-shan (Chap. V). There I was rejoined by R. B. Lâl Singh after an absence of four months, during which among other survey work he had carried his triangulation along the main K`un-lun range over five degrees of longitude eastwards.
A short-lived ` revolutionary ' outbreak at Charkhlik having allowed me to escape the obstruction with which I was seriously threatened, I started my long-planned explorations in the waterless desert of Lop. They led to the discovery of two ruined sites of ancient Lou-lan, abandoned since the early centuries of our era, which yielded interesting relics. The crossing of the wind-eroded desert northward revealed a succession of dry river-beds, unmistakably proving a southern extension of the delta in which the ` Dry River ' that once watered the Lou-lan of Han times had emptied itself into the bed of the ancient Lop Sea (Chap. VI).'
Resumption of work at and around the walled Chinese station of Lou-lan led to the discovery of more relics of the traffic that once came here by the earliest Chinese route leading into the Tarim basin. From grave-pits containing burial remains of the first centuries before and after Christ we recovered, besides other relics, a mass of remarkable textiles, including fine specimens of the earliest known figured Chinese silks as well as woollen tapestries showing clear evidence of Hellenistic art influence (Chap. VII. sec. i—vi). Reconnaissances pushed farther into the desert northeastward led to the discovery of an ancient Chinese casirum and an outlying watch-post with a burial-ground containing remarkably well preserved bodies of the ancient indigenous population of Lou-lan (Chap. VII. sec. viii).
With the help of the indications thus secured, we were able subsequently to start on the very difficult task of tracing the route which the Chinese had followed in their earliest trade and military
7 In order adequately to show all the details of geographical or archaeological interest recorded in the course of our surveys across the Lop Desert, both on the second and third expeditions, there has been prepared for insertion in Vol. iv the special Map of the Lop Desert, on the scale of 5 miles to I inch.
This map, which is enlarged with additions and correc
tions from Sheets Nos. 29, 32 of the general Map Series contained in Vol. iv, was reproduced at the Dehra Dun Survey Office after Chapters vI—vIII and XX. sec. iii had been passed through the press. It should be referred to in preference for all topographical and other details mentioned in those chapters.