Gobi

Ghashiun-Gobi and Kuruk-tagh.—The Yulduz valley or valley of the Khaïdyk-gol (83°-86° E., 43° N.) is enclosed by two prominent members of the Tian-shan system, namely the Chol-tagh and the Kuruk-tagh, running parallel and close to one another. As they proceed eastward they diverge, sweeping back on N. and S. respectively so as to leave room for the Baghrash-kol. These two ranges mark the northern and the southern edges respectively of a great swelling, which extends eastward for nearly twenty degrees of longitude. On its northern side the Chol-tagh descends steeply, and its foot is fringed by a string of deep depressions, ranging from Lukchun (425 ft. below the level of the sea) to Hami (2800 ft. above sea-level). To the south of the Kuruk-tagh lie the desert of Lop, the desert of Kum-tagh, and the valley of the Bulunzir-gol. To this great swelling, which arches up between the two border-ranges of the Chol-tagh and Kuruk-tagh, the Mongols give the name of Ghashiun-Gobi or Salt Desert. It is some 80 to 100 m. across from N. to S., and is traversed by a number of minor parallel ranges, ridges and chains of hills, and down its middle runs a broad stony valley, 25 to 50 m. wide, at an elevation of 3000 to 4500 ft. The Chol-tagh, which reaches an average altitude of 6000 ft., is absolutely sterile, and its northern foot rests upon a narrow belt of barren sand, which leads down to the depressions mentioned above.

The Kuruk-tagh is the greatly disintegrated, denuded and wasted relic of a mountain range which formerly was of incomparably greater magnitude. In the west, between Baghrash-kol and the Tarim, it consists of two, possibly of three, principal ranges, which, although broken in continuity, run generally parallel to one another, and embrace between them numerous minor chains of heights. These minor ranges, together with the principal ranges, divide the region into a series of long, narrow valleys, mostly parallel to one another and to the enclosing mountain chains, which descend like terraced steps, on the one side towards the depression of Lukchun and on the other towards the desert of Lop. In many cases these latitudinal valleys are barred transversely by ridges or spurs, generally elevations en masse of the bottom of the valley. Where such elevations exist, there is generally found, on the E. side of the transverse ridge, a cauldron-shaped depression, which some time or other has been the bottom of a former lake, but is now nearly a dry salt-basin. The surface configuration is in fact markedly similar to that which occurs in the inter-mont latitudinal valleys of the Kuen-lun. The hydrography of the Ghashiun-Gobi and the Kuruk-tagh is determined by these chequered arrangements of the latitudinal valleys. Most of the principal streams, instead of flowing straight down these valleys, cross them diagonally and only turn west after they have cut their way through one or more of the transverse barrier ranges.1 To the highest range on the great swelling Grum-Grzhimailo gives the name of Tuge-tau, its altitude being 9000 ft. above the level of the sea and some 4000 ft. above the crown of the swelling itself. This range he considers to belong to the Chol-tagh system, whereas Sven Hedin would assign it to the Kuruk-tagh. This last, which is pretty certainly identical with the range of Khara-teken-ula (also known as the Kyzyl-sanghir, Sinir, and Singher Mountains), that overlooks the southern shore of the Baghrash-kol, though parted from it by the drift-sand desert of Ak-bel-kum (White Pass Sands), has at first a W.N.W. to E.S.E. strike, but it gradually curves round like a scimitar towards the E.N.E. and at the same time gradually decreases in elevation. In 91° E., while the principal range of the Kuruk-tagh system wheels to the E.N.E., four of its subsidiary ranges terminate, or rather die away somewhat suddenly, on the brink of a long narrow depression (in which Sven Hedin sees a N.E. bay of the former great Central Asian lake of Lop-nor), having over against them the écheloned terminals of similar subordinate ranges of the Pe-shan (Bey-san) system (see below). The Kuruk-tagh is throughout a relatively low, but almost completely barren range, being entirely destitute of animal life, save for hares, antelopes and wild camels, which frequent its few small, widely scattered oases. The vegetation, which is confined to these same relatively favoured spots, is of the scantiest and is mainly confined to bushes of saxaul (Anabasis Ammodendron), reeds (kamish), tamarisks, poplars, Kalidium and Ephedra.

Desert of Lop.—This section of the Gobi extends south-eastward from the foot of the Kuruk-tagh as far as the present terminal basin of the Tarim, namely Kara-koshun (Przhevalsky’s Lop-nor), and is an almost perfectly horizontal expanse, for, while the Baghrash-kol in the N. lies at an altitude of 2940 ft., the Kara-koshun, over 200 m. to the S., is only 300 ft. lower. The characteristic features of this almost dead level or but slightly undulating region are: (i.) broad, unbroken expanses of clay intermingled with sand, the clay (shor) being indurated and saliferous and often arranged in terraces; (ii.) hard, level, clay expanses, more or less thickly sprinkled with fine gravel (say), the clay being mostly of a yellow or yellow-grey colour; (iii.) benches, flattened ridges and tabular masses of consolidated clay (jardangs), arranged in distinctly defined laminae, three stories being sometimes superimposed one upon the other, and their vertical faces being abraded, and often undercut, by the wind, while the formations themselves are separated by parallel gullies or wind-furrows, 6 to 20 ft. deep, all sculptured in the direction of the prevailing wind, that is, from N.E. to S.W.; and (iv.) the absence of drift-sand and sand-dunes, except in the south, towards the outlying foothills of the Astin-tagh. Perhaps the most striking characteristic, after the jardangs or clay terraces, is the fact that the whole of this region is not only swept bare of sand by the terrific sandstorms (burans) of the spring months, the particles of sand with which the wind is laden acting like a sand-blast, but the actual substantive materials of the desert itself are abraded, filed, eroded and carried bodily away into the network of lakes in which the Tarim loses itself, or are even blown across the lower, constantly shifting watercourses of that river and deposited on or among the gigantic dunes which choke the eastern end of the desert of Takla-makan. Numerous indications, such as salt-stained depressions of a lacustrine appearance, traces of former lacustrine shore-lines, more or less parallel and concentric, the presence in places of vast quantities of fresh-water mollusc shells (species of Limnaea and Planorbis), the existence of belts of dead poplars, patches of dead tamarisks and extensive beds of withered reeds, all these always on top of the jardangs, never in the wind-etched furrows, together with a few scrubby poplars and Elaeagnus, still struggling hard not to die, the presence of ripple marks of aqueous origin on the leeward sides of the clay terraces and in other wind-sheltered situations, all testify to the former existence in this region of more or less extensive freshwater lakes, now of course completely desiccated. During the prevalence of the spring storms the atmosphere that overhangs the immediate surface of the desert is so heavily charged with dust as to be a veritable pall of desolation. Except for the wild camel which frequents the reed oases on the N. edge of the desert, animal life is even less abundant than in the Ghashiun-Gobi, and the same is true as regards the vegetation.

Desert of Kum-tagh.—This section lies E.S.E. of the desert of Lop, on the other side of the Kara-koshun and its more or less temporary continuations, and reaches north-eastwards as far as the vicinity of the town of Sa-chow and the lake of Kara-nor or Kala-chi. Its southern rim is marked by a labyrinth of hills, dotted in groups and irregular clusters, but evidently survivals of two parallel ranges which are now worn down as it were to mere fragments of their former skeletal structure. Between these and the Astin-tagh intervenes a broad latitudinal valley, seamed with watercourses which come down from the foothills of the Astin-tagh and beside which scrubby desert plants of the usual character maintain a precarious existence, water reaching them in some instances at intervals of years only. This part of the desert has a general slope N.W. towards the relative depression of the Kara-koshun. A noticeable feature of the Kum-tagh is the presence of large accumulations of drift-sand, especially along the foot of the crumbling desert ranges, where it rises into dunes sometimes as much as 250 ft. in height and climbs half-way up the flanks of ranges themselves. The prevailing winds in this region would appear to blow from the W. and N.W. during the summer, winter and autumn, though in spring, when they certainly are more violent, they no doubt come from the N.E., as in the desert of Lop. Anyway, the arrangement of the sand here “agrees perfectly with the law laid down by Potanin, that in the basins of Central Asia the sand is heaped up in greater mass on the south, all along the bordering mountain ranges where the floor of the depressions lies at the highest level.”2 The country to the north of the desert ranges is thus summarily described by Sven Hedin:3 “The first zone of drift-sand is succeeded by a region which exhibits proofs of wind-modelling on an extraordinarily energetic and well developed scale, the results corresponding to the jardangs and the wind-eroded gullies of the desert of Lop. Both sets of phenomena lie parallel to one another; from this we may infer that the winds which prevail in the two deserts are the same. Next comes, sharply demarcated from the zone just described, a more or less thin kamish steppe growing on level ground; and this in turn is followed by another very narrow belt of sand, immediately south of Achik-kuduk.... Finally in the extreme north we have the characteristic and sharply defined belt of kamish steppe, stretching from E.N.E. to W.S.W. and bounded on N. and S. by high, sharp-cut clay terraces.... At the points where we measured them the northern terrace was 113 ft. high and the southern 85¼ ft.... Both terraces belong to the same level, and would appear to correspond to the shore lines of a big bay of the last surviving remnant of the Central Asian Mediterranean. At the point where I crossed it the depression was 6 to 7 m. wide, and thus resembled a flat valley or immense river-bed.”

Desert of Hami and the Pe-shan Mountains.—This section occupies the space between the Tian-shan system on the N. and the Nan-shan Mountains on the S., and is connected on the W. with the desert of Lop. The classic account is that of Przhevalsky, who crossed the desert from Hami (or Khami) to Su-chow (not Sa-chow) in the summer of 1879. In the middle this desert rises into a vast swelling, 80 m. across, which reaches an average elevation of 5000 ft. and a maximum elevation of 5500 ft. On its northern and southern borders it is overtopped by two divisions of the Bey-san (= Pe-shan) Mountains, neither of which attains any great relative altitude. Between the northern division and the Karlyk-tagh range or E. Tian-shan intervenes a somewhat undulating barren plain, 3900 ft. in altitude and 40 m. from N. to S., sloping downwards from both N. and S. towards the middle, where lies the oasis of Hami (2800 ft.). Similarly from the southern division of the Bey-san a second plain slopes down for 1000 ft. to the valley of the river Bulunzir or Su-lai-ho, which comes out of China, from the south side of the Great Wall, and finally empties itself into the lake of Kala-chi or Kara-nor. From the Bulunzir the same plain continues southwards at a level of 3700 ft. to the foot of the Nan-shan Mountains. The total breadth of the desert from N. to S. is here 200 m. Its general character is that of an undulating plain, dotted over with occasional elevations of clay, which present the appearance of walls, table-topped mounds and broken towers (jardangs), the surface of the plain being strewn with gravel and absolutely destitute of vegetation. Generally speaking, the Bey-san ranges consist of isolated hills or groups of hills, of low relative elevation (100 to 300 ft.), scattered without any regard to order over the arch of the swelling. They nowhere rise into well-defined peaks. Their axis runs from W.S.W. to E.N.E. But whereas Przhevalsky and Sven Hedin consider them to be a continuation of the Kuruk-tagh, though the latter regards them as separated from the Kuruk-tagh by a well-marked bay of the former Central Asian Mediterranean (Lop-nor), Futterer declares they are a continuation of the Chol-tagh. The swelling or undulating plain between these two ranges of the Bey-san measures about 70 m. across and is traversed by several stretches of high ground having generally an east-west direction.4 Futterer, who crossed the same desert twenty years after Przhevalsky, agrees generally in his description of it, but supplements the account of the latter explorer with several particulars. He observes that the ranges in this part of the Gobi are much worn down and wasted, like the Kuruk-tagh farther west and the tablelands of S.E. Mongolia farther east, through the effects of century-long insolation, wind erosion, great and sudden changes of temperature, chemical action and occasional water erosion. Vast areas towards the N. consist of expanses of gently sloping (at a mean slope of 3°) clay, intermingled with gravel. He points out also that the greatest accumulations of sand and other products of aerial denudation do not occur in the deepest parts of the depressions but at the outlets of the valleys and glens, and along the foot of the ranges which flank the depressions on the S. Wherever water has been, desert scrub is found, such as tamarisks, Dodartia orientalis, Agriophyllum gobicum, Calligonium sinnex, and Lycium ruthenicum, but all with their roots elevated on little mounds in the same way as the tamarisks grow in the Takla-makan and desert of Lop.

Farther east, towards central Mongolia, the relations, says Futterer, are the same as along the Hami-Su-chow route, except that the ranges have lower and broader crests, and the detached hills are more denuded and more disintegrated. Between the ranges occur broad, flat, cauldron-shaped valleys and basins, almost destitute of life except for a few hares and a few birds, such as the crow and the pheasant, and with scanty vegetation, but no great accumulations of drift-sand. The rocks are severely weathered on the surface, a thick layer of the coarser products of denudation covers the flat parts and climbs a good way up the flanks of the mountain ranges, but all the finer material, sand and clay has been blown away partly S.E. into Ordos, partly into the Chinese provinces of Shen-si and Shan-si, where it is deposited as loess, and partly W., where it chokes all the southern parts of the basin of the Tarim. In these central parts of the Gobi, as indeed in all other parts except the desert of Lop and Ordos, the prevailing winds blow from the W. and N.W. These winds are warm in summer, and it is they which in the desert of Hami bring the fierce sandstorms or burans. The wind does blow also from the N.E., but it is then cold and often brings snow, though it speedily clears the air of the everlasting dust haze. In summer great heat is encountered here on the relatively low (3000-4600 ft.), gravelly expanses (say) on the N. and on those of the S. (4000-5000 ft.); but on the higher swelling between, which in the Pe-shan ranges ascends to 7550 ft., there is great cold even in summer, and a wide daily range of temperature. Above the broad and deep accumulations of the products of denudation which have been brought down by the rivers from the Tian-shan ranges (e.g. the Karlyk-tagh) on the N. and from the Nan-shan on the S., and have filled up the cauldron-shaped valleys, there rises a broad swelling, built up of granitic rocks, crystalline schists and metamorphosed sedimentary rocks of both Archaic and Palaeozoic age, all greatly folded and tilted up, and shot through with numerous irruptions of volcanic rocks, predominantly porphyritic and dioritic. On this swelling rise four more or less parallel mountain ranges of the Pe-shan system, together with a fifth chain of hills farther S., all having a strike from W.N.W. to E.N.E. The range farthest N. rises to 1000 ft. above the desert and 7550 ft. above sea-level, the next two ranges reach 1300 ft. above the general level of the desert, and the range farthest south 1475 ft. or an absolute altitude of 7200 ft., while the fifth chain of hills does not exceed 650 ft. in relative elevation. All these ranges decrease in altitude from W. to E. In the depressions which border the Pe-shan swelling on N. and S. are found the sedimentary deposits of the Tertiary sea of the Han-hai; but no traces of those deposits have been found on the swelling itself at altitudes of 5600 to 5700 ft. Hence, Futterer infers, in recent geological times no large sea has occupied the central part of the Gobi. Beyond an occasional visit from a band of nomad Mongols, this region of the Pe-shan swelling is entirely uninhabited.5 And yet it was from this very region, avers G. E. Grum-Grzhimailo, that the Yue-chi, a nomad race akin to the Tibetans, proceeded when, towards the middle of the 2nd century B.C., they moved westwards and settled near Lake Issyk-kul; and from here proceeded also the Shanshani, or people who some two thousand years ago founded the state of Shanshan or Loû-lan, ruins of the chief town of which Sven Hedin discovered in the desert of Lop in 1901. Here, says the Russian explorer, the Huns gathered strength, as also did the Tukiu (Turks) in the 6th century, and the Uighur tribes and the rulers of the Tangut kingdom. But after Jenghiz Khan in the 12th century drew away the peoples of this region, and no others came to take their place, the country went out of cultivation and eventually became the barren desert it now is.6

Ala-shan.—This division of the great desert, known also as the Hsi-tau and the Little Gobi, fills the space between the great N. loop of the Hwang-ho or Yellow river on the E., the Edzin-gol on the W., and the Nan-shan Mountains on the S.W., where it is separated from the Chinese province of Kan-suh by the narrow rocky chain of Lung-shan (Ala-shan), 10,500 to 11,600 ft. in altitude. It belongs to the middle basin of the three great depressions into which Potanin divides the Gobi as a whole. “Topographically,” says Przhevalsky, “it is a perfectly level plain, which in all probability once formed the bed of a huge lake or inland sea.” The data upon which he bases this conclusion are the level area of the region as a whole, the hard saline clay and the sand-strewn surface, and lastly the salt lakes which occupy its lowest parts. For hundreds of miles there is nothing to be seen but bare sands; in some places they continue so far without a break that the Mongols call them Tyngheri (i.e. sky). These vast expanses are absolutely waterless, nor do any oases relieve the unbroken stretches of yellow sand which alternate with equally vast areas of saline clay or, nearer the foot of the mountains, with barren shingle. Although on the whole a level country with a general altitude of 3300 to 5000 ft., this section, like most other parts of the Gobi, is crowned by a chequered network of hills and broken ranges going up 1000 ft. higher. The vegetation is confined to a few varieties of bushes and a dozen kinds of grasses, the most conspicuous being saxaul and Agriophyllum gobicum7 (a grass). The others include prickly convolvulus, field wormwood, acacia, Inula ammophila, Sophora flavescens, Convolvulus Ammani, Peganum and Astragalus, but all dwarfed, deformed and starved. The fauna consists of little else except antelopes, the wolf, fox, hare, hedgehog, marten, numerous lizards and a few birds, e.g. the sand-grouse, lark, stonechat, sparrow, crane, Podoces Hendersoni, Otocorys albigula and Galerita cristata.8 The only human inhabitants of Ala-shan are the Torgod Mongols.

Ordos.—East of the desert of Ala-shan, and only separated from it by the Hwang-ho, is the desert of Ordos or Ho-tau, “a level steppe, partly bordered by low hills. The soil is altogether sandy or a mixture of clay and sand, ill adapted for agriculture. The absolute height of this country is between 3000 and 3500 ft., so that Ordos forms an intermediate step in the descent to China from the Gobi, separated from the latter by the mountain ranges lying on the N. and E. of the Hwang-ho or Yellow river.”9 Towards the south Ordos rises to an altitude of over 5000 ft., and in the W., along the right bank of the Hwang-ho, the Arbus or Arbiso Mountains, which overtop the steppe by some 3000 ft., serve to link the Ala-shan Mountains with the In-shan. The northern part of the great loop of the river is filled with the sands of Kuzupchi, a succession of dunes, 40 to 50 ft. high. Amongst them in scattered patches grow the shrub Hedysarum and the trees Calligonium Tragopyrum and Pugionium cornutum. In some places these sand-dunes approach close to the great river, in others they are parted from it by a belt of sand, intermingled with clay, which terminates in a steep escarpment, 50 ft. and in some localities 100 ft. above the river. This belt is studded with little mounds (7 to 10 ft. high), mostly overgrown with wormwood (Artemisia campestris) and the Siberian pea-tree (Caragana); and here too grows one of the most characteristic plants of Ordos, the liquorice root (Glycyrrhiza uralensis). Eventually the sand-dunes cross over to the left bank of the Hwang-ho, and are threaded by the beds of dry watercourses, while the level spaces amongst them are studded with little mounds (3 to 6 ft. high), on which grow stunted Nitraria Scoberi and Zygophyllum. Ordos, which was anciently known as Ho-nan (“the country south of the river”) and still farther back in time as Ho-tau, was occupied by the Hiong-nu in the 1st and 2nd centuries A.D., but was almost depopulated during and after the Dungan revolt of 1869. North of the big loop of the Hwang-ho Ordos is separated from the central Gobi by a succession of mountain chains, the Kara-naryn-ula, the Sheiten-ula, and the In-shan Mountains, which link on to the south end of the Great Khingan Mountains. The In-shan Mountains, which stretch from 108° to 112° E., have a wild Alpine character and are distinguished from other mountains in the S.E. of Mongolia by an abundance of both water and vegetation. In one of their constituent ranges, the bold Munni-ula, 70 m. long and nearly 20 m. wide, they attain elevations of 7500 to 8500 ft., and have steep flanks, slashed with rugged gorges and narrow glens. Forests begin on them at 5300 ft. and wild flowers grow in great profusion and variety in summer, though with a striking lack of brilliancy in colouring. In this same border range there is also a much greater abundance and variety of animal life, especially amongst the avifauna.

Eastern Gobi.—Here the surface is extremely diversified, although there are no great differences in vertical elevation. Between Urga (48° N. and 107° E.) and the little lake of Iren-dubasu-nor (111° 50′ E. and 43° 45′ N.) the surface is greatly eroded, and consists of broad flat depressions and basins separated by groups of flat-topped mountains of relatively low elevation (500 to 600 ft.), through which archaic rocks crop out as crags and isolated rugged masses. The floors of the depressions lie mostly between 2900 and 3200 ft. above sea-level. Farther south, between Iren-dubasu-nor and the Hwang-ho comes a region of broad tablelands alternating with flat plains, the latter ranging at altitudes of 3300 to 3600 ft. and the former at 3500 to 4000 ft. The slopes of the plateaus are more or less steep, and are sometimes penetrated by “bays” of the lowlands. As the border-range of the Khingan is approached the country steadily rises up to 4500 ft. and then to 5350 ft. Here small lakes frequently fill the depressions, though the water in them is generally salt or brackish. And both here, and for 200 m. south of Urga, streams are frequent, and grass grows more or less abundantly. There is, however, through all the central parts, until the bordering mountains are reached, an utter absence of trees and shrubs. Clay and sand are the predominant formations, the watercourses, especially in the north, being frequently excavated 6 to 8 ft. deep, and in many places in the flat, dry valleys or depressions farther south beds of loess, 15 to 20 ft. thick, are exposed. West of the route from Urga to Kalgan the country presents approximately the same general features, except that the mountains are not so irregularly scattered in groups but have more strongly defined strikes, mostly E. to W., W.N.W. to E.S.E., and W.S.W. to E.N.E. The altitudes too are higher, those of the lowlands ranging from 3300 to 5600 ft., and those of the ranges from 650 to 1650 ft. higher, though in a few cases they reach altitudes of 8000 ft. above sea-level. The elevations do not, however, as a rule form continuous chains, but make up a congeries of short ridges and groups rising from a common base and intersected by a labyrinth of ravines, gullies, glens and basins. But the tablelands, built up of the horizontal red deposits of the Han-hai (Obruchev’s Gobi formation) which are characteristic of the southern parts of eastern Mongolia, are absent here or occur only in one locality, near the Shara-muren river, and are then greatly intersected by gullies or dry watercourses.10 Here there is, however, a great dearth of water, no streams, no lakes, no wells, and precipitation falls but seldom. The prevailing winds blow from the W. and N.W. and the pall of dust overhangs the country as in the Takla-makan and the desert of Lop. Characteristic of the flora are wild garlic, Kalidium gracile, wormwood, saxaul, Nitraria Scoberi, Caragana, Ephedra, saltwort and dirisun (Lasiagrostis splendens).

This great desert country of Gobi is crossed by several trade routes, some of which have been in use for thousands of years. Among the most important are those from Kalgan on the frontier of China to Urga (600 m.), from Su-chow (in Kan-suh) to Hami (420 m.) from Hami to Peking (1300 m.), from Kwei-hwa-cheng (or Kuku-khoto) to Hami and Barkul, and from Lanchow (in Kan-suh) to Hami.

Climate.—The climate of the Gobi is one of great extremes, combined with rapid changes of temperature, not only at all seasons of the year but even within 24 hours (as much as 58° F.). For instance, at Urga (3770 ft.) the annual mean is 27.5° F., the January mean −15.7°, and the July mean 63.5°, the extremes being 100.5° and −44.5°; while at Sivantse (3905 ft.) the annual mean is 37°, the January mean 2.3°, and the July mean 66.3°, the range being from a recorded maximum of 93° to a recorded minimum of −53°. Even in southern Mongolia the thermometer goes down as low as −27°, and in Ala-shan it rises day after day in July as high as 99°. Although the south-east monsoons reach the S.E. parts of the Gobi, the air generally throughout this region is characterized by extreme dryness, especially during the winter. Hence the icy sandstorms and snowstorms of spring and early summer. The rainfall at Urga for the year amounts to only 9.7 in.

Sands of the Gobi Deserts.—With regard to the origin of the masses of sand out of which the dunes and chains of dunes (barkhans) are built up in the several deserts of the Gobi, opinions differ. While some explorers consider them to be the product of marine, or at any rate lacustrine, denudation (the Central Asian Mediterranean), others—and this is not only the more reasonable view, but it is the view which is gaining most ground—consider that they are the products of the aerial denudation of the border ranges (e.g. Nan-shan, Karlyk-tagh, &c.), and more especially of the terribly wasted ranges and chains of hills, which, like the gaunt fragments of montane skeletal remains, lie littered all over the swelling uplands and tablelands of the Gobi, and that they have been transported by the prevailing winds to the localities in which they are now accumulated, the winds obeying similar transportation laws to the rivers and streams which carry down sediment in moister parts of the world. Potanin points out11 that “there is a certain amount of regularity observable in the distribution of the sandy deserts over the vast uplands of central Asia. Two agencies are represented in the distribution of the sands, though what they really are is not quite clear; and of these two agencies one prevails in the north-west, the other in the south-east, so that the whole of Central Asia may be divided into two regions, the dividing line between them being drawn from north-east to south-west, from Urga via the eastern end of the Tian-shan to the city of Kashgar. North-west of this line the sandy masses are broken up into detached and disconnected areas, and are almost without exception heaped up around the lakes, and consequently in the lowest parts of the several districts in which they exist. Moreover, we find also that these sandy tracts always occur on the western or south-western shores of the lakes; this is the case with the lakes of Balkash, Ala-kul, Ebi-nor, Ayar-nor (or Telli-nor), Orku-nor, Zaisan-nor, Ulungur-nor, Ubsa-nor, Durga-nor and Kara-nor lying E. of Kirghiz-nor. South-east of the line the arrangement of the sand is quite different. In that part of Asia we have three gigantic but disconnected basins. The first, lying farthest east, is embraced on the one side by the ramifications of the Kentei and Khangai Mountains and on the other by the In-shan Mountains. The second or middle division is contained between the Altai of the Gobi and the Ala-shan. The third basin, in the west, lies between the Tian-shan and the border ranges of western Tibet.... The deepest parts of each of these three depressions occur near their northern borders; towards their southern boundaries they are all alike very much higher.... However, the sandy deserts are not found in the low-lying tracts but occur on the higher uplands which foot the southern mountain ranges, the In-shan and the Nan-shan. Our maps show an immense expanse of sand south of the Tarim in the western basin; beginning in the neighbourhood of the city of Yarkent (Yarkand), it extends eastwards past the towns of Khotan, Keriya and Cherchen to Sa-chow. Along this stretch there is only one locality which forms an exception to the rule we have indicated, namely, the region round the lake of Lop-nor. In the middle basin the widest expanse of sand occurs between the Edzin-gol and the range of Ala-shan. On the south it extends nearly as far as a line drawn through the towns of Lian-chow, Kan-chow and Kao-tai at the foot of the Nan-shan; but on the south it does not approach anything like so far as the latitude (42° N.) of the lake of Ghashiun-nor. Still farther east come the sandy deserts of Ordos, extending south-eastward as far as the mountain range which separates Ordos from the (Chinese) provinces of Shan-si and Shen-si. In the eastern basin drift-sand is encountered between the district of Ude in the north (44° 30′ N.) and the foot of the In-shan in the south.” In two regions, if not in three, the sands have overwhelmed large tracts of once cultivated country, and even buried the cities in which men formerly dwelt. These regions are the southern parts of the desert of Takla-makan (where Sven Hedin and M. A. Stein12 have discovered the ruins under the desert sands), along the N. foot of the Nan-shan, and probably in part (other agencies having helped) in the north of the desert of Lop, where Sven Hedin discovered the ruins of Loū-lan and of other towns or villages. For these vast accumulations of sand are constantly in movement; though the movement is slow, it has nevertheless been calculated that in the south of the Takla-makan the sand-dunes travel bodily at the rate of roughly something like 160 ft. in the course of a year. The shape and arrangement of the individual sand-dunes, and of the barkhans, generally indicate from which direction the predominant winds blow. On the windward side of the dune the slope is long and gentle, while the leeward side is steep and in outline concave like a horse-shoe. The dunes vary in height from 30 up to 300 ft., and in some places mount as it were upon one another’s shoulders, and in some localities it is even said that a third tier is sometimes superimposed.

Authorities.—See N. M. Przhevalsky, Mongolia, the Tangut Country, &c. (Eng. trans., ed. by Sir H. Yule, London, 1876), and From Kulja across the Tian Shan to Lob Nor (Eng. trans, by Delmar Morgan, London, 1879); G. N. Potanin, Tangutsko-Tibetskaya Okraina Kitaya i Centralnaya Mongoliya, 1884-1886 (1893, &c.); M. V. Pjevtsov, Sketch of a Journey to Mongolia (in Russian, Omsk, 1883); G. E. Grum-Grzhimailo, Opisanie Puteshestviya v Sapadniy Kitai (1898-1899); V. A. Obruchev, Centralnaya Asiya, Severniy Kitai i Nan-schan, 1892-1894 (1900-1901); V. I. Roborovsky and P. K. Kozlov, Trudy Ekspeditsiy Imp. Russ. Geog. Obshchestva Po Centralnoy Asiy, 1893-1895 (1900, &c.); Roborovsky, Trudy Tibetskoi Ekspeditsiy, 1889-1890; Sven Hedin, Scientific Results of a Journey in Central Asia, 1899-1902 (6 vols., 1905-1907); Futterer, Durch Asien (1901, &c.); K. Bogdanovich, Geologicheskiya Isledovaniya v Vostochnom Turkestane and Trudiy Tibetskoy Ekspeditsiy, 1889-1890; L. von Loczy, Die wissenschaftlichen Ergebnisse der Reise des Grafen Széchenyi in Ostasien, 1877-1880 (1883); Ney Elias, in Journ. Roy. Geog. Soc. (1873); C. W. Campbell’s “Journeys in Mongolia,” in Geographical Journal (Nov. 1903); Pozdnievym, Mongolia and the Mongols (in Russian, St Petersburg, 1897 &c.); Deniker’s summary of Kozlov’s latest journeys in La Géographie (1901, &c.); F. von Richthofen, China (1877).