As a sea the Gobi or Han-hai must have been comparable in extent to the Mediterranean, and the ancient coast-line can be pretty clearly recognized. In its present state it may be divided into two distinct basins, the western taking its name from the river Tarim or Tarym, and the eastern from the Chinese Shamo or “Sand Desert.” The Dzungarian valley stretches westwards like a gulf. The Tarim basin is bounded on the S. by the range of mountains which, under various names applicable to different portions, such as the Kwen-lun and the Altyntag, forms the northward rim of the great plateau of Thibet; on the west it comes up to the spurs of the Pamir plateau, and on the north it lies along the foot of the Thian Shan. If we measure from the source of any of its principal tributaries, the Tarim must have a course of more than 1000 miles. The head-waters rise in the mountains just named, and the more important of them in the south and west. The Khot,an river and its confluent the Kara-Kash both descend from the Karakorum mountains, and flow in a generally northward direction; the Zarafshan or Yarkand River, rising in the same range, winds about in the first part of its course so as to enter the Gobi almost from the west; and the Kizil Su or Kashgar River has its numerous head streams in the Kizil Yart mountains belonging to the Pamir plateau. The Aksai River and the Shah Yar are the most important contributions from the Thian Shan. The course of all of these rivers after they enter the Gobi is largely matter of conjecture, and all that can be asserted with confidence is that they unite to form the Tarim, and find their final goal in an inland lake. They have probably all reached a common channel about 82° E. long.; but as the stream presses eastward it again breaks up into numerous branches, the arrangement of which, except along the route followed by Przhevalski, is still unknown. As it passes east the stream gradually loses in volume by absorption, evaporation, and the demands of riparian populations. In the neighbourhood of the Ugendarya, the breadth is about 300 or 360 feet, and the depth about 20. The course of the Tarim lies much nearer the northern side of the Gobi than the southern, but it gradually trends south east, and at length passing through Lake Karaburan, loses itself in Lake Chon Kul (7c., great lake) or Kara-Kurchin. This last lake is identified with the famous Lob-nor, the position of which has been one of the outstanding problems of comparative geography. Against the identification a number of objections have been urged by Richthofen {cf “ Bemerkungen zu den Ergebnissen von Öber-lieut. Prejewalski’s Reise”in Verh.d. Ges.für Erdk., Berlin, 1878), the most important of which are the prevailing tradition that the Lob-nor was a salt lake while the Chon·Kul is fresh, and the fact that the Chinese maps place the Lob nor to the north of the position assigned to the Chon Kul, which according to Przhevalski lies about 39° 30' N. lat., immediately to the N. of the Altyntag range (13,000 to 11,000 feet high). The country through which the Lower Tarim flows is dreary and monotonous. “In general,” reports the traveller, “the Lob-nor desert is the wildest and most unfertile of all that I have yet seen in Asia; it is sadder than the desert of Ala-Shan.” A meagre vegetation of tamarisks and reeds lines the course of the river. Away towards the southwest there stretch, if we may trust to native reports, those vast fields of drifting and treacherous sands which have given so much of its terror to the legendary account of the desert of Gobi. That the reports are in the main true, and that the legends are founded on fact, appears to be rendered probable by the statements of Sir T. Douglas Forsyth, who has contributed an interesting paper on the subject to the Proceedings of the Royal Geographical Society (187 6). The population of the Tarim basin is scanty and poverty-stricken. On the Lower Tarim there are nine villages with a total of 1200 souls. Cattle-rearing is more general than agriculture, which indeed is of the most recent introduction, and confines itself to barley and wheat. Mahometanism is the universal religion, and the language appears to be identical with the Taranchi and the Sart.

The Shamo or eastern basin is cpιite different in its character. Here we have no large river like the Tarim, and, instead of its boundaries being marked by lofty ranges of mountains from 13,000 to 20,000 feet high, the ground gradually rises in a series of scarcely marked terraces. The central point, at Ozon Khoshu, is the lowest discovered in Central Asia, being only 607 metres (1948 feet) above the level of the sea. “The aspect of the country,” says Ney Elias, “who crossed in a north-westerly direction from China, is that of low hills or downs, with valleysandplains intervening, the whole of a rocky or stony nature rather than sandy, though patches of sand do occur every here and there. What little vegetation exists is chiefly composed of weeds, ‘ scrub,’ and heath, there being scarcely any grass, and only a dwarfed and stunted tree here and there, in the gorges or passes of those low rocky ranges that at uncertain intervals cross the desert in almost parallel lines from east to west.” Of the western portion of the basin we have no modern account.

Marco Polo was the first European who gave a distinct description of the desert of Gobi. He tells ns how on quitting Charchan (the modern Chachan, according to Yule) “you ride some five days through the sands finding none but bad and bitter water; and then you come to a city called Lop at the edge of the desert .... The length of the desert is so great that it would take a year and more to ride from one end of it to the other. It is all composed of hills and valleys of sand.” And then he goes on to speak of spirits that haunt the waste, and syllable men’s names, and of strange noises like the tramp and hum of a great cavalcade, of the sound of drums, and a variety of musical instruments. Polo appears to have proceeded east from Khotan to Lob, and then further east to Etsina on the southern edge of the desert, and afterwards to have spent forty days in crossing the desert northwards to Karakorum.

Later notices of the Gobi, especially of its eastern portions, are given by Gerbillon, 1688-98 (in Duhalde’s appendix), by the Dutchman Evert Ysbrand Ides (1692-94), and by Lorenz Lange, who was sent in 1727-28 and in 1736 by Peter the Great to Peking. But it was not till the present century that accurate information began to accumulate about the eastern portions, and the traveller who has lifted the veil from the western portions is still engaged in his explorations. In 1830-31 Fuss and Bunge crossed the eastern Gobi from Urga to Kalgan; and Dr Fritsche executed a series of journeys in the same district between 1869 and 1873. The missions of the Russian officials Andre Gustavitch Prinz (1863) and Shishmaroff (1868) added little to the knowledge of the region; but in 1870 Pavlinoff, consul at Chuguchak, being accompanied by a Government topographer Matusovski, made valuable observations on the route from Suok to Kobdo, and from Kobdo to Uliassutai. Of still greater moment were the travels of Ney Elias in 1872-73, and of Przhevalski between 1870 and 1877. In his eaιlier journey (1870-72) Przhevalski travelled across the Gobi in a line almost due south from· ürga, and in 1877 he struck south-east from the Yulduz range, one of the outrunners of the Thian Shan.

Besides the works referred to in the text see especially Richthofen’s masterly account of the Gobi in his China, vol i., Berlin, 1877, anil Prejevalsky, Mongolia, the Tangut Country, &c., London, 1876, and From Kulja across the Tian-Shan to Lob Nor, London, 1879. An account of Elias’s journey will be found in Journal of the Royal Geographical Society, 1873, and in Guido Cora’s Cosmos 1874.