V. A. Obruchev.

On the Question of the Origin of Loess.

(In defense of the aeolian hypothesis).

With 1 table.

Tomsk.

Typo-lithography of the Siberian Printing Society, corner of Dvoryanskaya St. and Yamsk. Lane, own building.

1911.

On the Question of the Origin of Loess

(In defense of the aeolian hypothesis.)

V. A. Obruchev.

(With 1 table.)

The old question of the origin of loess, which aroused so much controversy in the second half of the past century, ten years ago seemed already completely settled in a sense favorable to the aeolian hypothesis. The few adherents of the oldest hypotheses—the marine and the glacial-alluvial—had almost completely fallen silent; after the devastating criticism of P. A. Tutkovsky¹ it seemed that the defenders of the pluvial hypothesis, too, had only to follow the example of the former, and that all these hypotheses now possessed only historical interest. But in recent years authoritative voices have again been heard, not only assigning the most essential role in the process of formation of loess strata precisely to streams of rain water, but even denying the aeolian hypothesis its very right to exist, denying its significance not only for explaining the loess of Russia, but even—horribile dictu—the loess of Asia, i.e. the country that is, to a considerable degree, the homeland of this hypothesis.

In Russian literature, professors P. Ya. Armashevsky and A. P. Pavlov have appeared as especially ardent defenders of the pluvial hypothesis.

The first of them, in a major work devoted to the geology of Southern Russia², describing loess, loess loams, and sandy loams, calls these rocks subaerial-alluvial, i.e. washed up³ under the influence, as distinct from subaqueous-alluvial deposits, deposited under water. Criticizing the aeolian hypothesis in the form in which Richthofen created it and set it forth in the first volume of “China,” P. Ya. Armashevsky correctly notes some of its major shortcomings, arbitrary and unproven assumptions, which, in his opinion, best reveal the weak sides of the hypothesis and make it completely un-

self-sufficient even for explaining the loess of China¹; then the author sets forth the fluvial hypothesis, which he had defended as early as 1881, and finds that it explains quite satisfactorily not only all the features of loess, but also its distribution.

A. P. Pavlov in 1903 published an article, “On Turkestan and European Loess”², in which he says (on pp. 5–6): “Without denying the important geological role that wind plays in this scantily irrigated country, strongly heated by the summer sun, I must note that, with respect to the types of Turkestan loess I have seen, wind plays only a secondary role. Not to the wind, but to water does this loess owe its mighty development, which in no way prevents it from being a subaerial formation and from not having the characteristics of a sediment deposited on the bottom of some basin. Water, over more or less long intervals, flooded the surface of the country, sorted and distributed the mineral particles, washed away, transported, and buried, in silty streams and alluvial accumulations, the shells of land mollusks and the bones of animals that had fallen in the desert (? V. O.), and left a sediment with an even and smooth surface, on which here and there grew tufts of sparse vegetation, capable of surviving prolonged droughts. Such a surface, such vegetation, do not favor the retention of atmospheric dust, except for those patches that long remain moist and become covered with efflorescences of hygroscopic salts (takyrs). But these patches, entirely devoid of vegetation, do not favor the formation of a light, porous rock from decayed rootlets. It will more likely be a dense, layered rock, and not aeolian loess.

“The process of formation of Turkestan loess, in its mechanical conditions, is very close to the process of formation of deluvial deposits of European Russia, but there are also essential differences. Here it is not merely little streams flowing down a gentle slope that operate, but also considerable silty torrents, at times issuing from valleys, generated by downpours. The geological deposits accumulating by the distribution over plains of mineral material carried out by streams that temporarily pour out of mountain valleys and spread over the plains deserve to be separated into a special genetic type, for which it would be convenient to use the term ‘proluvial deposits, proluvium’.”

Thus, according to A. P. Pavlov, Turkestan loess is proluvium.

Six years later, in a report read at the XII Congress of Russian Naturalists and Physicians at the end of 1909 in Moscow, A. P. Pavlov

points still more definitely to how the Turkestan, South Russian, and Western European loess was formed¹): “The loess of the localities of Russia known to me has found its natural place among the formations of deluvial origin, and the picture of its formation appeared before me in an illumination quite different from that in which it presented itself to the majority of European geologists then imbued with Richthofen’s ideas. A journey to the loess regions of Western Europe, and especially acquaintance with the Danubian loess near Vienna and Krems, strengthened in me the conviction that my explanation is applicable also to the Western European loesses. Once thought had entered upon this path, it could not but go farther... the aeolian origin of the Asiatic loess began to arouse doubts in me.”

And further: “A journey in 1901 to Turkestan helped to outline yet another genetic type of continental deposits—proluvium, to which were assigned deposits accumulating by means of the repeated spreading, over desert plains, of mineral material carried out from mountain valleys by temporary streams spreading in different directions. It was noted that enormous masses of Turkestan loess must have been formed in this way.”

The author then gives an interesting example of the disintegration of the granite of the Ak-tau mountains and its transformation into loess, which we shall consider below; the detailed quotations have been given here in order to convey more fully A. P. Pavlov’s view of Turkestan loess as proluvium of desert piedmont plains, and to answer at once that on desert piedmont plains proluvium is indeed formed, but this proluvium has nothing in common with loess.

In foreign literature in recent years the dispute has likewise been renewed between the supporters of different views on the formation of loess; there have even appeared new defenders of the hypotheses of Kingsmill and Pumpelly, long since refuted by Richthofen, who suppose that the Chinese loess is a marine deposit or the deposit of extensive lakes.

In the present article I do not intend either to review the entire vast literature on loess, or to set forth and refute the marine, lacustrine, and glacio-alluvial hypotheses; after the above-mentioned work of P. A. Tutkovsky this seems to me entirely superfluous. I shall confine myself to considering certain aspects of the question to which little attention has hitherto been paid; I shall set forth in general outline the essence of the aeolian hypothesis in its modern form and the history of its development beginning with Richthofen, and, using a number of examples from various regions of loess distribution in Asia, for the most part studied by me

¹ On the Most Ancient Deserts on Earth, pp. 8 and 11; Moscow, 1910.

personally, I shall try to show the untenability of the stream hypothesis and the complete validity of the aeolian one.

It seems to me that many misunderstandings in the newly flaring dispute would be eliminated if the opponents of the aeolian hypothesis were well acquainted with this hypothesis in its modern form, formulated, incidentally, by me as early as 1894, and if they strictly distinguished two questions: 1) the origin of loess dust and 2) the place and modes of its deposition and transformation into layers of loess. On these questions, many years of observations in various parts of Asia have led me to the following conviction: 1) Loess dust, in the overwhelming majority of cases and in its immense mass, is the product of processes of weathering and winnowing in deserts, whether hot or cold, stony, clayey, or sandy, large or small, utterly barren and bare or, like many dry steppes, covered with comparatively scanty and sparse vegetation that does not sufficiently protect the soil from the action of the wind; only the quantity of loess dust produced by a desert will depend on the size and character of the desert, while the composition of this dust will depend on the petrographic composition of its soil. 2) The place and modes of deposition of this dust may be varied, depending on climate, relief, hydrography, vegetation cover, etc. Alongside loess, in the deposition of which the wind plays the primary role, one may also encounter loess whose particles were brought by the wind but deposited with a more or less substantial participation of water.

But the stream, or diluvial, hypothesis, in the form in which it is understood by P. Ya. Armashevsky and A. P. Pavlov, cannot by itself explain the formation of loess and is only an auxiliary to the aeolian hypothesis, its role in comparison with the latter being, in the majority of cases, very limited.

The basic idea of the stream hypothesis was first expressed by Folger, who explained by it the formation of the Rhine loess¹; the basic idea of the aeolian hypothesis was indicated by Villers d’Aoust in 1857, who explained by it the Mexican loess layers 60–100 m thick²; but both hypotheses were developed in greater detail by Richthofen in the first volume of his China, which appeared in 1877, although almost everything essential had been set forth by him earlier in letters about his journey, published in Shanghai in 1870–72³.

Richthofen recognized a very substantial role for streams of rainwater in the process of creating loess strata, as is evident from

“…of his following considerations: ‘It remains still to find those agents which supplied the material for the gradual raising of the soil. Even a superficial examination leads to three different kinds of such agents. The first is rainwater, which flowed from the higher parts to the lower ones and washed away the solid constituent particles released by the destruction of the rocks of the neighboring mountains. The second is the wind, etc.’¹

He further supplements what has been said: “At the time when, with deposits of loess dust, one part of it falls on soil overgrown with vegetation, is retained by the vegetation and raises the surface, another part is deposited in such places where it is washed away by the next rain and either distributed over the steppe or carried by streams into salt lakes.”²

Finally, on p. 100 we find the most complete picture of the work of rainwater: “I have already tried to outline the principal features of the activity of water and pointed out that, alongside the activity of water flowing in channels, the activity of sheet-washing water (spülendes) in basins deprived of runoff plays an especially important role³. Rain removes from the cliffs what the wind has left behind and fills in small irregularities with it, so that inclined surfaces form everywhere in their place, along which the material is then shifted downward (hinabgeschoben wird). Usually the falling water is insufficient to saturate the soil; therefore only a small part of the dust is carried into streams, and its (the water’s) influence is essentially purely local. But however slowly the material moves downward, these small quantities, over a long period of time, yield large results. The structure of the loess indicates that periodically stronger atmospheric precipitation occurs, or that a whole interval of time is characterized by such precipitation. Only by this can it be explained that heaps of mountain debris, gradually accumulating at the foot of slopes, are spread out along the slope of the steppe, in places along the entire margin of the mountains and often at a distance of several thousand paces from it, and still farther, toward the center; on the surface there are created peculiar conditions which in later times result in the growth of marly nodules along these surfaces, as we saw in the loess basins. Each such period of intensified movement is followed again by regular growth of the loess, indicating a return of a drier climate.”

It seems to me that these quotations are sufficient to show that the essence of the fluvial hypothesis had already been given by Richthofen; the modern defenders of this hypothesis repeat part of Richthofen’s reasoning, denying only the role of the wind.

Richthofen examines in detail the process by which the steppe basins, devoid of runoff, were filled with loess strata on pp. 85–104, vol. III, China, and illustrates it with fig. 11 on p. 104, fig. 4 on p. 65, and fig. 14 on p. 121. Summarizing what is scattered over a number of pages, the following picture of loess accumulation according to Richthofen may be drawn.

Owing to the dry climate, with the predominance of physical weathering and the intensive activity of the wind, the products of destruction of the rocks composing the ranges and chains of hills are carried down from the heights and gradually accumulate in the valleys and basins in the form of fluvial and lacustrine bedded loess on the floor, and unstratified typical loess on the slopes. Owing to the absence of water runoff in these closed depressions, the accumulation of aeolian dust over a prolonged period creates considerable thicknesses; the floor of the depressions gradually rises; the height of the ranges and hills decreases both absolutely, as a result of the destruction of the bedrock and the removal of material, and relatively, as a result of the raising of the floor of the depressions being filled in. In the end this process leads to the typical landscape of the central regions—broad basins with gentle slopes, gradually rising toward low, smoothed watersheds, on which only occasionally may destroyed outcrops of bedrock be encountered.

Richthofen attempted to determine the thickness of the strata filling the depression in the basin of Lake Anguli-nor in southeastern Mongolia, on the basis of the present difference in heights and slope, and obtained a figure of more than 1000 feet (fig. 11 and p. 104); he believes that for other elongated basins of the same locality the values would be still greater, and for the extensive depressions of western Tibet, Kuku-nor, and Han-hai—still greater.

On the distribution of loess in Central Asia Richthofen expresses himself as follows (p. 85): “The loess basins of northern China have shown us the internal structure of the fillings of the steppe depressions of Central Asia... The analogy with the modern steppes, devoid of runoff, is so far confidently extended only to the margins of the regions adjacent to loess countries. But, on the one hand, the similarity of the final result in all parts of Central Asia justifies the conclusion as to the similarity of the processes that created it; on the other hand, the exposition of the following chapter will show that formations corresponding to the loess of China are distributed around the borders of Central Asia. Therefore we may suppose that, with the exception of the marine deposits of Han-hai, beneath all steppe depressions lie accumulations of loess-like formations, which—”

mixed with coarse gravel, both at the bottom, where they cover the bedrock, and on the slopes of the depressions filled by them. If they were absent, a mountainous country would be exposed, with the most varied elevations and forms, occupying all of Central Asia, in exactly the same way as if northern China were deprived of its loess strata¹.

In addition to these words, on the map (table 2, p. 150) almost all of Central Asia is colored light yellow as “areas of deposits of the central saline steppes,” with the exception of several narrow strips in eastern Mongolia and a broader area in the lower reaches of the Tarim, corresponding to the deepest depressions, in which, according to Richthofen, remnants of the sea should have been preserved longest after the retreat of Han-hai. The boundaries of the latter at its maximum development on this map encompass considerable spaces of the “steppe region,” indicating that, in Richthofen’s opinion, the bottom of this sea, once freed from water—except for the deepest depressions noted above—also turned into loess basins; points of sandy desert and gravelly steppe are placed in places over the yellow color of the loess. In the text of Chapter IV one may find not a few indications that Richthofen did in fact imagine almost all of Central Asia to be covered with loess, and on pp. 110–111 it is said that the processes of filling depressions with loess strata began on the margins of Han-hai and within its limits considerably later and therefore attained insignificant development, spreading from the edges toward the middle, where in places the character of the old sea floor was still preserved.

The essence of Richthofen’s aeolian hypothesis may be summarized as follows: loess is a product of weathering processes in the central regions of continents, which lack runoff of waters into the ocean and have a dry climate; loess dust is blown by winds and washed down by rains from all heights, gradually accumulating and filling depressions; modern Central Asia is almost entirely occupied by such depressions, more or less filled with loess; the loess of northern China was formed in a similar manner at the time when this part of Asia was attached to the central region; upon reunification with the peripheral region, the erosion of the loess strata began, and the structure of the soil of the steppe depressions became apparent.

After my very first acquaintance with eastern Mongolia and northern China at the beginning of the journey of 1892–94, I already expressed doubt as to the correctness of Richthofen’s views on the second aspect of the question of loess—namely, on the place and mode of its deposition². The incorrectness

was explained by the fact that Richthofen traveled almost exclusively through China, where he studied all the peculiarities of loess; but of all Central Asia he saw only a small area on the southern margin of Eastern Mongolia, an area not at all typical of Central Asia; the observations made in this locality he made the basis of his whole hypothesis and extended his conclusions to all of Central Asia.

During my own journey I traversed Central Asia in various directions and became convinced that in this country there are no basins filled with loess in accordance with Richthofen’s hypothesis, and that such basins occur only here and there on the outskirts of this country; moreover, in most cases the thicknesses of the loess have far less magnitude than was supposed by the author of “China.” But the filling of basins with loess to the extent proposed by Richthofen, and by the processes which he assumed, is impossible even theoretically, as the following reasoning shows.

Let us imagine in cross-section (Fig. 1) a depression—for example, a valley deprived of runoff—between two mountain ranges. Let us even allow, in accordance with Richthofen’s views, that all the products of the destruction of both ranges are not carried away by winds from the central region (which does in fact occur, as we shall see below), but remain, merely moving from the heights into the valley. It is easy to see, by looking at the drawing and on the basis of the volume of both ranges and of the intervening valley, that the process of accumulation of material can proceed with any rapidity only at the very beginning; the more the ranges are lowered as a result of the removal of material from them, and the more the floor of the basin is raised as a result of the deposition of this material, the less favorable becomes the relation between the volumes—the one supplying material and the one receiving it—and the more the process of loess accumulation must slow down. In the end, the small remaining outcrops of bedrock on the crests of the heights (Fig. 2) may furnish such an insignificant quantity of loess dust that the growth of the loess in the valley, which has expanded at the expense of the slopes of the ranges, must slow down to infinity; and we inevitably arrive at the conclusion that, even under the most favorable conditions, mountain ranges cannot be completely hidden beneath thicknesses of loess formed from the products of their destruction.

The same reasoning is applicable to the stream hypothesis as well: assuming that wind plays an insignificant role and that the products of weathering are carried from the ranges into the basin chiefly by streams of rainwater, we shall arrive at the same conclusion—that mountain ranges cannot be completely hidden beneath thicknesses of loess.

Consequently, the ideal landscape which Richthofen pictured to himself for vast expanses of Central Asia—broad basins filled with loess to a thickness of 1–2000 fathoms, and intermediate-

...smoothed hills, also covered with loess (see China, vol. I, p. 103 above, the left half of fig. 14 on p. 121 and fig. 4 on p. 65), among which only here and there are ruined ridges or outcrops of bedrock visible, cannot be explained either by the fluvial hypothesis or by the aeolian one in the form in which its author set it forth.

In extending his steppe depressions (Steppen-becken), filled with loess, over almost all of Central Asia, Richthofen, of course, could not but know that the innumerable ridges of this country a priori cannot have a uniform composition; but he did not go into the details of the process of physical weathering, evidently assuming that loess dust is the principal product of the destruction of all possible rocks, besides which there is produced also a certain quantity of grains of sand accumulating at the bottom of the depression. On p. 99 (China, vol. I) Richthofen says: “Apparently a tract of sandy desert is found in each of the more considerable steppe depressions and usually occupies an eccentric position. The small loess steppe depressions are free of it, as is for the most part also the gravel steppe. The greatest distribution of sands is found in certain very extensive depressions into which the floor of Han-hai is divided; here it is derived for the most part from Cretaceous sandstones, the products of whose destruction the wind has swept into one place and freed from the considerable admixture of clay which they contain according to Stoliczka’s observations. The gravel desert is limited, so far as is known, to the Sha-mo depression, where there was material for its formation.”

From these words it is evident how unclear Richthofen’s conception was of the processes of weathering in deserts, about which at that time we in fact still had very insufficient information. Today we could not imagine a depression filled with loess as the chief product of the destruction of neighboring ridges composed, for example, of quartzites, sandstones, granites, gneisses, and similar very widespread rocks, which contain much quartz. Of course, from these rocks too a known quantity of loess dust should be obtained, but where is the sand to disappear that constitutes the main part of the products of their destruction? In general, the process that Richthofen assumed—the transformation of various rocks into layers of loess within the limited confines of a comparatively small steppe depression—is, given the present state of our knowledge, entirely implausible; the same conclusion must be drawn also with regard to the fluvial hypothesis, which likewise presupposes the possibility of forming a thick loess from the products of destruction of neighboring mountains or hills, carried down by rainwater. We shall return to this question once more when considering the views of A. P. Pavlov.

The aeolian hypothesis of Richthofen is likewise inapplicable to explaining the accumulation of loess in many localities of China, that is, precisely the country for which it was devised.

The loess regions of Northern China have, as it turns out, a very varied pre-loess relief, which may be reduced to three types: a) mountainous country; b) a country of table heights; and c) a plain.

In the northern parts of the provinces of Chih-li and Shan-si, adjoining Eastern Mongolia, we find more or less high and parallel mountain ranges, the slopes of which and the intervening valleys are covered with loess of varying thickness, though in general not great. Restoring the thicknesses of loess already removed by erosion, according to Richthofen’s hypothesis, we obtain a picture (Fig. 3) analogous to that discussed above; this is the first type of loess regions in China.

In the southern and western parts of the same province of Shan-si and in the adjacent western part of the province of Shen-si we find not mountain ridges, but more or less broad plateaus in the form of steps of a tableland, cut by faults; prolonged erosion, even before the beginning of the loess period, dissected the scarps of these steps, composed chiefly of more or less horizontal Carboniferous and Mesozoic deposits, among which masses of older or igneous rocks rise in places; more or less considerable thicknesses of loess lie not only on the surface and on the scarps of the steps, but also on the higher parts of the plateau (Fig. 4). Restoring, according to Richthofen’s hypothesis, the eroded parts of the loess cover, we obtain a country with broad valleys and basins separated from one another by still broader and flatter heights. This is the second type.

Finally, in the western part of the province of Shen-si and in the eastern part of the province of Kan-su, bordering on the sands of the Ordos from the south, we find a slightly dissected plain, composed of almost horizontal Mesozoic deposits, covered by a layer of loess of the greatest thickness—up to 400 m (Fig. 5). This loess plateau is cut by valleys, gorges, and ravines into flat mountains and ridges of table-like or rounded form, of approximately equal height, on whose slopes the loess is exposed and only at the very bottom the underlying Mesozoic rocks. Toward the south the terrain lowers somewhat and the thickness of the loess diminishes; here the former plateau has been better preserved and is dissected into separate table masses of various length and width. Still farther south, separate mountain ranges appear and the terrain becomes more dissected; the thickness of the loess diminishes further. This is the third type, approaching the first toward the south. If, according to Richthofen’s hypothesis, we restore the eroded masses of loess, we shall obtain an inclined plateau, falling steeply to the north, gently descending to the south, where it will pass into a flat-hilly country. Let us even suppose that, under especially favorable conditions, the filling of a depression between two ranges could have occurred.

products of the weathering of the latter in the form of a thick stratum of loess, as Richthofen supposed, i.e., a loess country of the first type could have formed; but can the loess countries of the second and third types be explained by means of this hypothesis, where the highest points are even now still covered with thick loess, where for considerable distances—tens and even hundreds of versts—there are no dominant heights composed of bedrock? Whence, then, came the material for the formation of this loess?

The hypothesis of the filling of steppe depressions with products of the destruction of neighboring heights, created by Richthofen, cannot explain the origin of loess countries of the second and third types and only with great stretches, as we saw above, can explain the origin of a loess country of the first type. But the same must also be said of the fluvial hypothesis—its applicability is likewise undermined by the absence of more considerable heights from which rainwater could carry down loess material. For one cannot resort to the supposition that all the more considerable heights, composed of bedrock, which existed at the time of the formation of the loess and supplied material for the work of rain streams, were destroyed by erosion at the very last moment. P. Ya. Armashevsky¹ and Lapparent resort to such a supposition, and nothing else can really be devised; but it is not enough to make such a supposition, it must be proved, which for the loess countries of China of the second and third types is utterly impossible. P. A. Tutkovsky² has already pointed out this Achilles’ heel of the fluvial hypothesis; the entire hypothesis in its pure form is built on this unproved supposition and falls together with it, for if it cannot explain the formation of loess in the countries of the indicated types, then is it needed at all as something independent?

One theoretical consideration of the hypotheses of Richthofen and the fluvial hypothesis, based on acquaintance with the modern relief and structure of the most typical loess country, has led us to this result.

Thus, Richthofen’s aeolian hypothesis, built on the assumption that steppe depressions are filled with loess dust of local origin, with endemic dust, requires a substantial correction. This correction consists in referring the source of the dust beyond the limits of the given depression, to a more or less considerable distance, i.e., in introducing into the whole process dust of distant origin, exotic dust. For the aeolian hypothesis such a correction is entirely possible, since light dust can be carried through the air over enormous distances, across mountain ranges and table elevations—

...regions, hills, and plains—up to the limit of the spread of the winds of the given direction that carry this dust. For the jet hypothesis such an amendment is impossible: rain jets are always tied to the given slope of the given depression and cannot even fall into one neighboring to it; therefore they carry and deposit local products of weathering.

The idea of the transport of fine products of weathering over great distances is just as old as the loess hypothesis itself—Villier d’Aoust expressed it in 1857; many others expressed it as well, but Richthofen did not use it to construct his hypothesis on other grounds. This is all the more strange because in the same volume of “China” he says that fine products of weathering can be carried far away (p. 96), speaks of the dusty atmosphere of Central Asia (p. 97), of the possibility of carrying dust from the central regions into the peripheral ones (p. 98). He quotes Johnson’s words about the fertility of the country of Khotan, conditioned by the fact that fine dust is brought by winds from the desert and deposited on the plains (p. 97), and reports Stoliczka’s phrase about Yarkand: “here, in a land of deserts, where clouds of fertile dust replace clouds of beneficent aqueous vapors, where the atmosphere is rarely free of dust, and at times is saturated with it, the explanation almost forces itself upon one, against one’s will, that loess is a subaerial deposit.” But, quoting these words (in a note on p. 124), Richthofen adds: “if Stoliczka had been able to continue his observations, he would probably have come to the conviction that, although the surface of the loess is still constantly being built up by falling dust, exposures of it exceeding 80 ft. prove that here either loess depressions, originally without drainage, entered the drainage area toward Lake Tarim, or that the marginal region of the Tarim basin, consisting of loess, was cut through as a result of subsidence of the locality at the foot of the Tian Shan, which he (i.e., Stoliczka) acknowledges to have caused the increase of the slope.”

Thus Richthofen did not attach importance to the dust carried beyond the limits of Central Asia and, with regard to the plains of the Tarim basin, tried to apply the same hypothesis of steppe depressions filled with loess. Stoliczka, in essence, was much closer to a correct and comprehensive solution of the question of the origin of loess than Richthofen, despite the fact that the latter studied in detail the typical country of loess in China.

I. Walther, who reported so much that was new in his book on denudation in deserts and who examined the processes of weathering from various sides, devoted almost no attention to the fine products of these processes; he only indicated that they are carried away by winds and deposited—

On the question of the origin of loess.

are found far from the place of their origin, on the edge of a desert¹. He did not find loess strata in the deserts of Africa and Sinai that he investigated, but paid absolutely no attention to this contradiction between his observations and the hypothesis of his teacher Richthofen.

In his “Guide for research travelers”², published 9 years after the first volume of China, Richthofen, however, had already expressed himself differently about the processes of weathering in the desert and about the laws governing the sorting of loose masses by the wind; he says that the coarsest materials remain in place, forming a stony and pebbly desert, while all mobile particles are distributed into two belts—of sand and of dust (p. 441). Dust in Central Asia is deposited chiefly in steppes with grassy vegetation, situated quite close to the bare deserts, gradually smoothing out the irregularities of the relief (pp. 446–447). Richthofen describes the process of the transport and deposition of dust in considerable detail, but alongside the new views he also retains the former ones concerning the filling of steppe depressions without runoff as a result of the weathering of the surrounding ridges (pp. 447 and 479), and concerning the idea that loess represents steppe soil exposed by erosion in the marginal regions of steppe countries deprived of runoff.

Thus, by the end of the 1880s the question of the origin of loess as eolian dust had been clarified, and the question of the regularity of its distribution was already approaching the correct solution; but it had not yet reached the latter, as is shown by the first edition of the Physical Geology (vol. II, 1888) of I. V. Mushketov, where 30 pages are devoted to sands, while the regularity of the distribution of continental sands is not considered, whereas the genesis and distribution of loess are set forth in accordance with the new views of Richthofen expressed in his “Guide.”

Somewhat later J. Walther, in his description of the North American deserts³, only mentions the removal of fine products of weathering from the desert, but does not at all touch upon the question of the places and modes of their deposition, nor the question of the distribution of sands.

The idea of the removal of fine products of weathering from desert regions was also expressed by several Russian travelers in Asia, my predecessors. Thus, for example, G. N. Potanin, in describing the sands of the Ordos, speaks of the sorting of material lifted by air currents, and of two streams—the sandy and the dusty—and tries to explain the regularity of the distribution of sands over the surface

Central Asia¹). But he does not arrive at the seemingly inevitable conclusion that the dust stream forms loess strata, and he regards the conditions governing the distribution of sands as unclear to him. Both this traveler and G. E. Grum-Grzhimailo, in his letters from his travels, make no mention of the presence of loess in the parts of Central Asia they visited; in these letters one may also find indications of an abundance of exposures of bedrock in this country not only on the slopes and crests of ranges, but also on hills and on the floors of valleys. Yet neither the one nor the other tried to clarify the regularity of this phenomenon, nor did they compare their observations with Richthofen’s assumptions about the loess steppe basins prevailing in Central Asia.

Consequently, by the time of my journey through Central Asia there already existed data on the absence of loess in certain parts of this country, and various persons, including Richthofen himself, had expressed thoughts about the removal of the finer products of weathering from deserts and their deposition in neighboring steppe lands. It remained to verify and supplement the observations proving the absence of loess in Central Asia, to explain the causes of this contradiction with the aeolian hypothesis, and to elucidate the regularity in the distribution of the products of weathering formed in the deserts.

This task fell to me, and a two-year journey through various parts of Central Asia and the loess lands of China gave me sufficient material for its solution. While wholly subscribing to Richthofen’s aeolian hypothesis in its basic idea—that loess is an accumulation of aeolian dust—I was obliged to oppose Richthofen’s view of the manner in which loess strata accumulate. Observations showed me that in Central Asia typical and thick loess in general does not exist and cannot exist, since the finest products of weathering are carried beyond the limits of that country and deposited in localities with a different climatic regime; although by no means all of Central Asia is desert, and considerable expanses of it are occupied by steppes on which, according to Richthofen, loess dust ought to have been deposited, the soil of these steppes, too, does not consist of typical loess; the latter appears only along the margins, and attains great thickness outside the limits of Central Asia, in countries already having drainage to the ocean; only the barriers of very high mountains, such as the Kuen-lun, the Pamir, and the Tian Shan, detain a more or less considerable part of the dust and compel it to be deposited still within the limits of the area deprived of drainage; but Richthofen’s typical present-day steppe basins, filled with thick loess that is still continuing to grow, proved not to exist. Not only dust, but also coarser products of weathering, such as grains of sand, are carried in considerable masses out of

central parts of the country, from the desert, accumulating in barkhan sands along its margins. In the distribution of sands and loess, in the thickness of their strata and in the size of the spaces occupied by them, a complete regularity and mutual dependence was found; there also became clear the beneficial role, on the one hand, of the desert, which gives the surrounding countries fertile loess, and its destructive role, on the other, owing to the sands advancing upon these countries.

This regular combination of regions of deflation and deposition is shown schematically in Fig. 6; the region of deflation occupies the central part of the country and in general represents a depression, because of the constant removal of the products of weathering to its margins and beyond its limits by winds blowing centrifugally; toward the southern side more frequent and stronger winds are directed, carrying away the main mass of dust and all the sand; therefore on the south the region of deflation is bordered by a belt of sands, followed by a broad belt of thick loess; the winds blowing to the north are much weaker and rarer; they are capable of carrying away only silt; therefore on the north the region of deflation is bordered only by a narrow belt of thin loess. In application to the various countries of the globe, this scheme must of course undergo more or less considerable changes depending on the local peculiarities of relief and climate.

Referring to my articles published in 1894—96¹), I must note that the views expressed by me met with no objections in the scientific literature, but at the same time remained unknown to some geologists interested in loess, although they were included almost in their entirety in the new edition of I. V. Mushketov’s “Physical Geology.” I shall point out, for example, that P. Ya. Armashevsky, in the above-mentioned work, where he discusses the genesis of loess, does not mention my works in a single word, confining himself to criticism of Richthofen. In the two above-mentioned articles on loess by A. P. Pavlov, that new form of the aeolian hypothesis which I defend is likewise ignored. A. A. Inostrantsev is not acquainted with it either; in the 3rd edition of his course “Geology” (vol. I, 1899) he devotes only 15 lines to loess dust. Even J. Walther, in his new book on the laws of desert formation²), referring to one observation of mine in the Trans-Caspian Region, is completely silent about the more important observations directly concerning the question under discussion—

¹) a) Orography of Central Asia and its southeastern outskirts. Izv. I. R. G. O. XXXI.
b) On the processes of weathering and deflation in Central Asia. Zap. Miner. Obshch. 1895, part XXXIII, issue 1.
c) Several travel reports in Izv. I. R. G. O. XXIX, XXX and XXXI.
d) Central Asia, Northern China and Nan-shan, 2 vols., 1900–01 (contain all the factual material).
²) Das Gesetz der Wüstenbildung, Berlin 1900.

same themes of my articles on Central Asia, although one of them was published in German¹. In this book the chapter on dust and loess is the least developed, and in essence the author remained at the point of view set forth by Richthofen in the “Guide” 15 years earlier.

I venture to think that, had the authors mentioned reckoned with the new form of the aeolian hypothesis, their conclusions would have had to undergo a considerable modification.

If it is admitted that loess is deposited outside the place where aeolian dust is formed, then all three types of loess countries in China are easily explained; with an exotic origin of the dust, the latter may be deposited on terrain of any relief, gradually smoothing out all irregularities, filling both the valleys between ranges and the plateaus, as well as the ranges themselves, plateaus, and lowlands; the thickness of the loess will depend, on the one hand, on the amount of dust brought in and on the duration of the corresponding climatic conditions, and on the other—on the character of the local hydrographic network. The deposition of loess can occur and undoubtedly does occur under the present climatic conditions of Central Asia and of the countries surrounding it; these countries are not distinguished by an abundance of atmospheric precipitation, and the accumulation of loess in them struggles against its washing away; where the relief is less favorable to accumulation and more favorable to erosion, as in a loess country of the first type, the thickness of the loess cannot attain any considerable magnitude; under opposite conditions—loess is very thick, as in countries of the second and third types. But even in different parts of a country of one and the same type the growth of loess is, of course, not uniform: on the even surfaces of plateaus, in lowlands, and on gentle slopes the growth proceeds more rapidly; on steeper slopes rainwater washes part of the dust down into the valleys, ravines are formed and expand, and vertical cliffs are formed.

But how can one prove that the deposition of loess in the countries of Northern China also occurred in the past under climatic conditions very close to the present ones? This circumstance is very important, since Richthofen assumed that all the loess countries of China had formerly belonged to the central region of Asia, i.e., had been deprived of drainage of their waters into the ocean, and only thanks to this could strata of loess have accumulated in them, burying steppe depressions (and also plains and plateaus, we add); according to Richthofen (China, Bd II, p. 756), the “steppe” period came in China owing to the retreat of the sea as a result of the uplift of mountain ranges to the southeast of the Great Chinese Lowland; these ranges barred access for moist sea winds into the interior of Northern China,

on which, therefore, the steppe climate of Central Asia spread; the country was deprived of the runoff of waters developed in the preceding humid, “erosional” period; the hydrographic network disappeared, the dried-up valleys were gradually filled with thicknesses of loess. A new advance of the sea and the subsidence of the southeastern mountain barriers brought about the onset of the “loess” period with the present climatic conditions; Northern China was reunited with the peripheral region of Asia; the restored rivers gradually conquered their former channels, cutting through and exposing the loess of the steppe basins; this period of humid climate continues at the present time as well.

This necessity for considerable uplifts and abrupt climatic changes constitutes the weak side of Richthofen’s hypothesis, as P. Ya. Armashevskii rightly points out ¹), although the former adduces much evidence in favor of his view. Without entering into a critical examination of them, which it is better to make elsewhere ²), I shall for the present confine myself to pointing out that there is no particular need to resort to these assumptions. The climate of the loess countries of Northern China can in no way be called humid; it differs little from the climate of Central Asia in the amount of atmospheric precipitation, as is proved by the steppe character of the vegetation of the loess slopes and plateaus; and the farther we move from the great Chinese plain to the north, northwest, and west, the smaller will be the difference between the climate of the Mongolian steppes and that of the loess countries. Such is the traveler’s immediate impression, which is confirmed by scattered information and scanty meteorological data.

The second proof is provided to us by the hydrographic network; Richthofen assumes that it was developed even before the “steppe” period, during which it was destroyed by thicknesses of loess, and then succeeded in renewing itself along its former directions. The latter is very improbable; with those enormous thicknesses of loess which filled all the valleys and leveled almost the entire country, it is difficult to imagine that the renewing hydrographic network restored all its former lines; we should have had to see epigenetic valleys in great development, and the entire network should have produced the impression of youth. Meanwhile this is not the case—the river valleys are deeply incised, the river systems very ramified; many small rivers and even large rivers, such as the Yellow River, are incised to a more or less considerable depth into the thickness of varied bedrock underlying the loess, and by no means everywhere have the character of gorges. Let us take, for example, the stretch of the Yellow River below its bend to the south; a whole series of its right tributaries takes its beginning on the plains of Eastern Ordos and then cuts into the loess plateau, higher than these plains. If the valleys of these rivers—

¹ L. c., pp. 231–234.
² In volume III of my report on the journey, which is being prepared for publication.

began to develop after the deposition of loess, then we ought to see two systems of river valleys: one running eastward from the watershed to the Yellow River, the other westward toward the Ordos, where the streams would have had to flow into lakes or be lost in the sands. But if we suppose that the headwaters of the streams flowing into the Yellow River had already managed to cut through the watershed and capture the streams of the Ordos slope, radically altering the original hydrographic network, then this proves the antiquity, not the youth, of the latter.

I believe that loess in Northern China was deposited earlier as well under conditions close to the present ones, and that the loess regions of China even then belonged not to the central, but to the peripheral region of Asia. Perhaps the climate of these regions was still somewhat drier than at present, atmospheric precipitation fell in still smaller quantities, and the streams carried still less water than now. To account for a slight change of climate in these regions toward aridity, there is no need to resort to major orographic changes—a considerable retreat of the sea, or the uplift of mountain ranges.

Loess is being deposited in Northern China at the present time as well; why, then, could it not have been deposited under the same conditions before, accumulating over tens of thousands of years? In a dry climate and with steppe vegetation, the accumulation of loess may proceed faster than its erosion; if for whole weeks or even months there is no rain, if in winter snow falls rarely and in small quantities and does not accumulate, but is blown away by winds and evaporates in the dry air—then there are no conditions for the energetic work of water. The few water-poor streams, poor in tributaries, wind lazily through their broad valleys, carrying in their muddy waters a certain quantity of loess dust; over the broad spaces between riverbeds—plains, plateaus, or mountain ranges—loess dust is deposited almost daily and is held by steppe grasses, beaten down to the ground by occasional rains or snows, gradually building up the soil. An occasional strong downpour may wash away part of the loess and carry it into the streams, may cut new ravines and enlarge old ones; but its work is limited to the steeper slopes, and on plains and gentle slopes it cannot wash away all the loess that has accumulated over the long interval between two such downpours, especially since this permeable dry soil absorbs a mass of moisture before becoming so saturated with it that water will flow in little streams over its surface; moreover, loess, when wetted with water, becomes sticky, and on sticky soil the erosive force of water on gentle slopes is slight.

And so, I believe that even under the present climatic conditions of Northern China there is an accretion of loess from dust brought from Central Asia; this accretion also occurred earlier—from the time when the water expanses of the Inner Asian freshwater

seas (or a complex of lakes), and over the vast areas laid bare a desert regime became established.

This process began, of course, not at once, not when the last remnants of this sea disappeared (they exist somewhere even now!), but considerably earlier—as soon as the sea, in drying up, exposed sufficiently extensive areas of its bed. The desert regime was established gradually and kept intensifying; the sands and clayey expanses of the desiccating bed furnished the wind with abundant material, probably far more abundant than at present, when such material is obtained mainly through the weathering of bedrock, which proceeds slowly. The clear reddish color of the lower horizons of the loess strata of China, especially noticeable in the provinces of Shan-si and eastern Kan-su, as well as their great clayiness, are explained by the fact that they were formed at a time when the red-colored “Gobi” deposits, left by the Tertiary sea (or lakes), still covered considerable expanses of Central Asia and provided the principal material for deflation. Now the remnants of these deposits are encountered comparatively rarely, and the chief material is supplied by much more varied and older bedrocks, the barkhans of extensive sandy areas, and the loess deposits (proluvium) of the pedestals of mountain ranges; the mixture of all this heterogeneous material gives the loess its typical brownish-yellow color.

The change in the climatic conditions of North China toward still greater dryness, connected with the dying out of many scantily watered river valleys, could, of course, have accelerated the accumulation of loess, but I see no need for such a change, since accumulation also takes place under existing conditions, and there has been enough time for the formation of enormous strata up to 400 m thick, especially if our supposition is correct that this process began as early as the end of the Tertiary period. I also cannot refrain from expressing the apprehension that, if North China were joined to the central region of the Asiatic continent, such a climatic regime ought to have become established in that country as favors not the accumulation of aeolian dust but its deflation; for at the present time, within Central Asia, loess is formed in the form of more or less thick strata only in especially favorable places, at the barriers of very high mountains that hold back the winds, as, for example, on the margins of the Tarim basin; the mountain ranges, plateaus, and plains of North China do not present such barriers, and if they were joined to the central region, i.e. to the region of deflation, they themselves would begin to yield loess dust, but could not serve as a place of its accumulation.

Thus the need for a strong change in climatic conditions to explain the deposition of loess in North China, it seems to me, disappears, and at the same time all the objections fall away which, in the eyes of P. Ya. Ar—, constituted weighty arguments against the aeolian hypothesis in general.

...of Mashevsky. But in its modern form this hypothesis does not require such assumptions, since it says that sand and loess are products of deflation in present-day deserts, semi-deserts, and desert-steppes¹, carried by winds and deposited: sand—for the most part on the margins of these areas of deflation, and loess—beyond the limits of the latter, in the surrounding countries, where climatic conditions and vegetation favor the accumulation of dust. The areas of deflation and deposition must be regularly associated with one another; if, for one reason or another, an area of deflation disappeared—for example, became covered by water or vegetation—then the source of the dust and the accumulation of the latter in the surrounding countries also ceased; the desert became “fossil,” and so did the loess strata surrounding it. Therefore the loess of southern Russia and western Europe must be regarded as fossil, since the deserts that created it have disappeared; but the loess of northern China cannot be called fossil, because the deserts of Central Asia exist and continue their activity.

The new form of the aeolian hypothesis, as far as I know, has not yet been tested on all the continents of the globe. I shall now try to examine from the point of view of the regularity of the association of deserts, sands, and loess other countries of Asia, concerning which there is sufficient information.

After China, loess has its greatest distribution in Russian Turkestan, and a good map of this country immediately reveals the dependence between loess, sands, and deserts.

Loess occurs in the form of a more or less broad belt at the foot and on the slopes of the mountain ranges of the Tian Shan and the Pamir-Alai, alongside mudflow deposits (proluvium) composing stony and clayey plains; it reaches its greatest development in the Fergana and Kulja valleys, which are open to the west, closed on the other sides, and represent two gigantic sacks for trapping aeolian dust and even sand, forming accumulations at the bottom of both valleys. The totality of the indicated mountain ranges constitutes that high barrier which detains the aeolian dust brought by northwestern winds.

Moving to the northwest from the area of development of Turkestan loess, we enter the extensive area of the Kyzylkum sands, reaching the shores of the Aral Sea, and to the northwest of the latter we shall see the deserts and desert-steppes of Ustyurt. The latter repre—

...constitute the area of winnowing; but in the area of the sands as well we shall find, besides the sandy barkhans, many sources of dust formation in the form of rocky cliffs on the western shore of the Aral, sandy islands and spits of its other shores, silty deposits of the lower Syr- and Amu-Darya, and elevations, half covered with sands, of Tertiary and older rocks scattered among the Kyzyl-Kum. Yet, despite these local peculiarities of the sandy region, the regular combination here of desert, sands, and loess is entirely clear.

All the distinctive features of Turkestan loess noted by A. P. Pavlov, and appearing in his eyes as proofs of the diluvial formation of the former, such as: the stratification of the lower horizons, caused by interbeds of sand and fine gravel and by the presence of beds of nonuniform structure; the connection of the loess of mountain valleys now with glacial deposits, now with products of the weathering of clays, conglomerates, and granites¹—all of them do not in the least contradict the aeolian hypothesis. Aeolian dust, brought from the desert and the sandy region, was deposited over tens of thousands of years; during this time, on the slopes and at the foot of the mountains, various processes of weathering and washing took place, the products of which were naturally mixed with the aeolian dust or covered the aeolian accumulations. By separating the questions of the origin of aeolian dust and of the place and modes of its deposition, we can without any strain explain all the peculiarities of the loess strata of Turkestan.

In the Transcaspian region neighboring Turkestan we see the same regular combination of a loess area on the slopes of the Kopet-Dagh and Parapamiz, an area of the sands of the Kara-Kum, extending from these ridges to the foot of the Ust-Yurt escarpments, and an area of winnowing on the surface of the latter. The areas of sands and loess are especially well known to me, thanks to personal investigations in 1886–88; here we encounter such peculiarities as are explicable only with the help of the aeolian hypothesis.

In the eastern part of the region, along the Afghan frontier from the Amu-Darya River to the Tedzhen River, there stretches in a broad belt a hilly steppe, the soil of which is composed of thick loess; this locality is very poor in water—numerous valleys are waterless, others contain streams of brackish water, and only in the rivers Tedzhen, Kushk, Murghab, and Amu-Darya, which take their rise outside this belt, is the water fresh; wells are scattered sparsely, are usually very deep, and often likewise contain brackish water. The monotonous broad and flat ridges, “bairs” in the local language, are covered

more or less dense steppe vegetation and constitute excellent pastures, especially on the surface of the higher Kara-bel plateau, which reaches 2500–2600 m absolute elevation (about 2200 feet above the level of Murgab at the mouth of the Kushk River) and extends to the northeast from Murgab along the border. Here and there, in closed depressions among the baïrs, there are saline self-depositing lakes, shors, and takyrs. The baïrs consist of loess, as is proved by outcrops along the banks of the Kushk and Murgab rivers (nearer to the Tejen River appear heights composed of the Tertiary and Cretaceous rocks of Kopet-dag), and this is confirmed by the whole character of the locality and its vegetation, by the rarity and quality of the water; only near the town of Andkhoy does their soil become more sandy, and from there to the Amu Darya River the locality acquires a different character—plains with sandy soil predominate, and shifting sands are encountered. To the north the baïrs gradually become lower and pass into an almost level sandy steppe, still with good vegetation, but already with coarser, sandy soil; still farther north the sandy steppe gradually passes into the hummocky and barkhan sands of the Kara Kum.

Can the accumulation of loess in this locality be explained, according to the fluvial hypothesis, by processes of silting? The heights of Parapamiz and Hindu Kush, which might have served as the source of this silting, lie 50–100 versts to the south and, moreover, are separated from the northern part of the baïr belt by the valleys of the rivers Egrigek, Kushk, Murgab, Kila-vali, and Kaiser (Andkhoy); the baïrs of this northern part rise above the valleys of these rivers by several hundred and up to a thousand feet. Can it be that these valleys and the entire dissected relief of this whole locality were created after the formation of the loess strata? By what agents? At the present time this is a typical country of accumulation with a dry climate, and one would have to resort to a whole series of unproved assumptions about abrupt climatic changes in order to explain first the deposition of the loess stratum from the heights of Parapamiz and Hindu Kush, then the energetic erosion of this stratum, and finally the modern conditions. The aeolian hypothesis explains all the features easily: here we see a broad belt of accumulation at the barrier of mountain ranges; into this bay, projecting southward between the heights of Kopet-dag in the west and the Pamir-Alai in the east, an especially large amount of aeolian dust arrives from Ust-Urt and from the sands of Kara Kum; here, for tens of thousands of years, typical loess has accumulated and continues to accumulate to this day, which to the north, nearer the sands, becomes increasingly sandy and finally passes into sand; here one may also encounter closed steppe depressions of various sizes, with or without a salt lake, among gently sloping heights covered with dense grass and clothed from bottom to top in layers of loess—in a word, the typical landscape that Richthofen described for Central Asia; but here, too, this loess was formed from exotic dust, covering with its mantle the ancient relief of the country, the traces of which...

formed from Tertiary deposits, as is shown by the exposures along the banks of the Kushk River. In the vicinity of the latter one can also find sandy accumulations caused by the reworking of the loess stratum, exposed in cliffs and ravines¹.

In the western part of the Transcaspian Region, from the western end of Kopet-Dagh to Gyaurs station, the northern foothills of this range consist chiefly of desert-steppe, the soil of which is composed of irregularly interbedded loess-like loam, proluvium, and alluvium; northward these steppes gradually pass into bare and smooth takyrs, beyond which sands begin. The vegetation of the desert-steppes consists of bushes of saltwort, wormwood, caragana, etc., and of very low and sparse grass, in places entirely absent; along the streams and brooks flowing from the range there are gardens, arable fields, and clover fields. The soil is predominantly clayey in character, but in places the quantity of pebbles and rubble increases to such an extent that the steppe may be called pebble-rubble steppe.

Upon this steppe, of course, loess dust is deposited, brought by northern winds from the sands and deserts; but owing to the insufficient protection afforded by the sparse and scanty vegetation, this dust can be retained only in places—under bushes, around grass stems and little stones, in depressions and cracks in the soil, while the greater part is blown away by the winds and carried farther, still closer to the range, where it is retained on areas of grassy steppe at the foothills and on the mountain slopes. In general, the increase of soil on the desert-steppe would proceed extremely slowly if it were not assisted by another agent—rain water: downpours, which occur in the mountains several times a year and sometimes reach great force, produce temporary torrents that burst from the gorges of the range onto the desert-steppe of the foothills and often spread over the latter across a considerable area; these torrents carry with them silt, sand, rubble, pebbles, sometimes even boulders, as well as loess

dust washed from the slopes and gradually deposit this material, but as the quantity of water, owing to absorption into the soil and evaporation, diminishes, its carrying power, owing to the division of the flow into rivulets and streams, weakens. These temporary rain-fed flows or networks lay down “proluvium,” to use the term proposed by A. P. Pavlov, but this proluvium is by no means loess: it is a mixture of washed loess dust, silt, sand, and coarser materials, forming a more or less stratified deposit, though often irregularly and indistinctly stratified. The farther from the mountains this proluvium is deposited, the more fine material it contains, since the coarse material, because of its weight, settles earlier; the finest material—almost pure loess mud—is deposited farthest away, in takyrs, which during heavy rains turn into temporary lakes, since further runoff of the water is prevented by a barrier of sands. These lakes exist for several hours or several days, and then dry up, leaving on the surface of the takyr a thin layer of clay; in this way the soil of the takyrs is gradually built up, representing a clearly and finely stratified secondary loess, a mixture of entopic and exotic dust, sometimes with interlayers of small grains of sand brought by an especially heavy downpour or blown in from neighboring sands.

In those areas of the desert-steppe which for a long time were not flooded by temporary streams because of the shifting of the beds of the latter, herbaceous vegetation gradually becomes stronger, which causes greater retention of aeolian dust and the growth of unstratified loess; after a more or less prolonged interval of time such areas are again subjected to flooding, and proluvium will be deposited upon them. This is why the soil of the desert-steppes represents an alternation of thin layers of typical loess, proluvium and, in places (where the water of permanent streams flowed, likewise changing their beds), alluvium.

Only at the very foot of the mountains, in the intervals between ravines, can unstratified loess accumulate without frequent disturbances, and here it must attain the greatest thickness and purity, if one does not count those interbeds of rubble which are washed out onto the steppe directly from the slopes during especially heavy downpours; in such areas the herbaceous vegetation is considerably denser than in the desert-steppe.

Eastward from Gaurs station and as far as the town of Lutfabad, the foot of the Kopet-Dag has a different character: here the northern range of the Zaryn-Kukh ridge is not broken by transverse ravines and stretches as a low continuous rampart for 55 versts. From the northern slope of this rampart only small and weak temporary streams can run down onto the steppe at the foot, not disturbing the process of accumulation of aeolian loess; here

why here, from the ridge to the edge of the sands, there stretches a strip about 20 versts wide of true grassy steppe, sharply differing from the desert-steppe of the rest of the foothill; its soil consists of non-stratified loess. The absence of takyrs at the edge of the sands throughout the entire extent of this steppe also constitutes its distinguishing feature and confirms the absence of abundant temporary streams; only the solonchak occupying the depression between the Babadurman and Artyk stations, into which the water of the Dergöz River drains, disturbs the general picture.

To the east of the city of Lotfabad, in the northern chain of the Kopet-dag, numerous gorges again appear, while at the foot of the ridge the desert-steppe, bordered by takyrs, is renewed.

Thus the northern foothill of the Kopet-dag gives us a clear example of the fact that where temporary rain streams silt up the surface of the steppe, not typical loess is formed, but something different; where there is no such silting, loess accumulates under the protection of grasses, which cannot grow on a steppe that is often silted up. The deposits of temporary streams, having a peculiar character, deserve to be set apart into a special category of alluvial deposits, and the term “proluvium,” proposed for them by A. P. Pavlov, seems to us quite apt; but only proluvium is not like aeolian loess and proluvial processes cannot create typical non-stratified loess.

Let us now move farther north: the complex of ranges of the Dzungarian Alatau is bordered on the north and west by a belt of loess, to the north and west of which lie the sands of Muyun-kum, Sary-ishik-otrau, and others along the southern shore of Lake Balkhash; beyond this lake stretches, over a great expanse, the well-known Hungry Steppe of Bek-pak-dala. Consequently, here too we see a regular combination of desert, sands, and loess.

To the east of the Dzungarian Alatau lies the Barlyk Range, studied by me in considerable detail in 1905 and 1909; here the distribution of loess is very instructive (Fig. 7). This range is a stepped horst, diminishing northward by a series of ledges and steeply truncated from the south; therefore its main watershed is situated near the southern foothill and represents a rather dissected and high ridge, Ker-tau, lowering toward the NE; in this direction the entire range rapidly narrows and tapers out. Adjoining the northern slope of the broadest western part of the range are groups of low mountains and hills, Kiik and Tyulklyu, extending to the left bank of the Emil River. Loess is distributed chiefly along the NW slope of Barlyk, reaching its greatest thickness (up to several sazhens) in the basin of the Tsagan-tokoi River; the lower ledge of the horst to the north and south of the middle course of this little river presents a rather typical loess landscape with a strongly rounded relief of heights, good steppe vegetation, and rare outcrops of bedrock—on

on higher peaks and in the precipices of river gorges and ravines. To the east, on the same ledge, and to the SE, on higher ledges, it gradually becomes thinner and disappears on the higher part of the ridge. On the southern slope of the main watershed it appears here and there and is not thick, with the exception of a small section of the southern foothill, about 20 versts west of Sary-kulsyn Peak, where it attains almost the same development as on the Tsagan-tokhoi ledge; but only thanks to the proximity of the rocky summits of the main watershed does it contain much rubble and stretch in a narrow, inclined strip between the mountains and the valley of the river Kunt.

From the northwest there adjoins Barlyk an area of more or less bare sands, occupying rather large areas along the lower reaches of the Emil River and the northern shore of Lake Ala-kul; and still farther to the NW we find extensive solonchaks around Lake Sasyk-kul and a desert-steppe with waterless channels of a whole series of streams flowing from Tarbagatai and the Dzungarian Alatau. Again, all three regions—the blowing-away and accumulation of coarse and fine products—are situated side by side, as evidence of the validity of the aeolian hypothesis. Between the areas of loess and sand here one can even find a transitional belt—a sandy steppe, situated along the western foothill of the Kiiik and Tyulklyu mountains, with good steppe vegetation and an aeolian soil consisting of clayey fine sand, for the most part unstratified and porous, like loess; here, together with part of the dust, chiefly finer grains of sand accumulated, carried by winds from the barchan sands and from the area of blowing-away. The transition from this loess-like sand, covering the plain, to the purer loess lying on the heights to the east and west of it is very gradual.

The regularity in the distribution of loess on Barlyk seems to us entirely clear: the greater part of the dust brought by NW winds settled on the windward side of the ridge, where it occupied the greatest space; but part of it was carried by the winds farther to the ESE beyond the watershed and settled on the leeward side of the ridge, especially thickly in the aforementioned belt west of Sary-kulsyn Peak, after also being carried across the valley of the Kunt still farther to the SE, where, at the northern foothill of the Dzhair ridge, generally poor in loess, we encounter yet another area of loess steppe; and here again the loess is thick, although more sandy, probably owing to the proximity of the bare riverbeds and the pebble desert of the Kunt valley.

If one looks at a map of the Dzungarian Alatau Range, then in the immediate vicinity of Barlyk, to the west of the Dzungarian Gates, a triangular projection of mountains jutting out to the N catches the eye; within it are situated the town of Lepsinsk and the villages of Konstantinovka, Gerasimovka, Kolpakovka, and Osinovka; this locality is famed for its fertility, which is why rich villages arose here. Without consulting the literature on the composition of the soils of this projection, I assert,

that a thick loess should have developed here, since to the NW there are fairly extensive sands at the eastern end of Lake Balkhash, and beyond them—the desert of the southern margin of the Kirghiz steppe.

Can the stream hypothesis explain this capricious distribution of loess on Barlyk, where it now thickens, now thins and disappears? Rains fall everywhere, temporary streams run from all gullies and ravines, yet loess has accumulated only in certain places.

Let us move still farther north, within Siberia. The extensive steppe bordering the western spurs of the Altai to the NE of the town of Semipalatinsk is famed for its fertility, and here are located the enormous and numerous villages of the Semipalatinsk–Barnaul route; the soil of this steppe is loess, which also rises onto the spurs of the Altai. Of particular interest is the Bel-Agach steppe, situated about 50 versts north of Semipalatinsk and rising above the surrounding country; according to P. N. Krylov, who investigated this steppe in the summer of 1910, its soil is likewise loess, lying in a thickness of several sazhen on ancient bedrock, as drilling showed. The fertility of this steppe has caused an influx of settlers, despite the complete absence of running water; the latter is obtained in places from deep wells, and in places from winter snow, artificially gathered into drifts and covered with straw. How can the stream hypothesis explain the increase of loess on this steppe, which is higher than the surrounding locality?

To the southwest and west of the loess steppes of western Altai, along the right bank of the Irtysh River, stretch extensive sands, for the most part covered with pine forests; and still farther west, beyond the Irtysh, lies an extensive region of winnowing—the southern dry part of the Kirghiz steppe. From there loess dust is brought by the prevailing southwesterly winds¹.

Loess is also developed in the vicinity of the city of Tomsk, on the surface of the high terrace of the right bank of the Tom River, composed of strongly dislocated Carboniferous sandstones and shales, on which trans-

...gress Tertiary pebble beds and sands with plant impressions; the latter are covered by loess reaching 2–3 sazhens in thickness, unstratified and porous; it differs from Chinese loess by its greater hardness and viscosity, owing to a high clay content, and also by a somewhat darker color; at the surface it is colored by sor. It is quite impossible to explain the Tomsk loess by the stream hypothesis; the nearest heights from which this loess could have been washed down are in the Kuznetsk Alatau, about 100 versts to the southeast.

The only explanation here, too, is the deposition of dust brought to this region by the southwesterly winds, prevailing to this day, from the nearest area of deflation—the vast Kirghiz steppe; this also accounts for the great clayiness of this loess: over so considerable a distance only the finest dust was carried. Let us note, incidentally, that in Tomsk on dry summer or spring days, with southwesterly winds, there is sometimes the same dry haze as in Central Asia—the air is filled with the finest dust, the reddish sun shines dimly, and the distance is hidden as if in fog.

The Tomsk loess, apparently, is not an isolated occurrence, but the northeastern corner of a large loess region, which embraces the eastern part of Baraba almost as far as Lake Chany and stretches southward along the banks of the Ob River, joining near the Aley River with the western Altai loess region. To the SW of the line Lake Chany—the middle course of the Kulunda River, the loess gives way to hummocky calcareous sands, forming the soil of the Kulunda steppe and attaining on the watersheds 30 or more meters in thickness. Although these sands are stratified and are considered ancient aqueous Tertiary deposits¹, it is possible that their upper horizon represents an aeolian deposit replacing the loess in regions of more southerly aspect.

The next example of the extensive development of loess is the Minusinsk Basin, enclosed on all sides by mountains and intersected from south to north by the Yenisei River; on the left bank of the latter, within the basin, there are rather extensive dry steppes—the Kachinsk, Abakan, and Koibal steppes—with numerous salt flats and salt lakes, with hills and ridges composed of Devonian deposits. This is an area of deflation; the prevailing southwesterly winds, when passing over the Kuznetsk Alatau, give up to the mountains and dense forests of the latter all their moisture and, in the Minusinsk Basin, become dry; in the indicated steppes precipitation is rare, even in winter snow—

little falls out, and it is blown away by winds; the processes of physical weathering dominate almost the whole year. On the right bank of the Yenisei in the Minusinsk steppe we find an area of deposition of coarser material in the form of sands, for which, for example, the environs of the town of Minusinsk are famous; farther east begins the area of loess, which covers the eastern part of the basin and even rises onto the loess foothills of the Sayan, reaching here a thickness of up to 10–15 sazhens. According to P. K. Yavorsky, from whom we take the cited data¹, the loess is typical, homogeneous, unstratified, porous, and brownish-yellow in color; in river valleys and in depressions of the soil it has its greatest thickness, while on mountain slopes it gradually thins from the foot toward the summit, where it is sometimes even absent. The area of its distribution generally coincides with the steppe and subtaiga districts; along river valleys of a latitudinal direction the loess penetrates especially far. This investigator even happened to see how, in November 1892, a south-west wind that had been blowing for several days deposited upon snow that had fallen shortly before in the town of Minusinsk, on trees and buildings, a layer of fine dust about 1.5 mm thick. I myself had occasion to see, in December 1910, near the village of Chebaki, already on the northern margin of the basin, how much dust the south-west wind raised together with snow on the steppe, accumulating this mixture of dust and snow in gullies, hollows, ravines, and so forth.

Toward the surface the Minusinsk loess, at 0.4–0.8 m, passes into chernozem, and owing to its presence the eastern part of the Minusinsk district is famous for its fertility; the boundaries of the agricultural region coincide with the boundaries of the distribution of loess.

With regard to the distribution of loess on the western margin of the Minusinsk basin, we find no indications among local investigators; meanwhile, according to the stream hypothesis, there could also be loess here: the eastern foothills of the Kuznetsk Alatau are occupied by steppes; on them there are outcrops of bedrock undergoing weathering; atmospheric precipitation is not lacking either; that is, the conditions for the accumulation of loess according to the stream hypothesis exist—but there is no loess. Consequently, the Minusinsk basin presents yet another example confirming the correctness of the aeolian hypothesis.

Still farther east we encounter loess in the environs of the city of Irkutsk—on the flat heights that border the valleys of the Angara and Irkut rivers and are composed of Jurassic coal-bearing deposits; the loess forms a stratum up to several arshins thick, is unstratified and porous, has a brownish-yellow color, and differs from the typical loess by a somewhat greater content of sand, which makes it looser. Its stratum crowns the heights, for example, Kaiskaya Mountain between the station and the lower reaches of the Irkut River, Jerusalem Mountain on the right bank of the Angara, and

i.e.; the more considerable heights from which this loess might have been washed down, according to the fluvial hypothesis, are at a distance of several versts and are separated from the indicated mountains by valleys or, at the very least, by deep saddles likewise more or less covered with loess; this proves that the present relief of the locality had already been created before the beginning of the loess period, and that the fluvial hypothesis cannot explain the accumulation of the Irkutsk loess.

The aeolian hypothesis, with knowledge of the local conditions, is quite applicable here as well. Irkutsk is situated in the southern corner of that vast amphitheater of the Sayan and Baikal mountains which Ed. Suess named after this city; these mountains are covered with dense forests and have a humid climate, but the arena of the amphitheater along both banks of the Angara for 200–300 versts north of Irkutsk consists of many dry steppes, which is why nomads—cattle-breeding Buryats—have long lived here. We do not, however, find here either deserts or desert-steppes, since the arena is open to the NW, whence the winds bring sufficient precipitation. Nevertheless, the extensive Balagansk steppes constitute an area of deflation; whoever lived in Irkutsk in spring could not help noticing those dust storms which come at this time of year from the northwest and usually end with the fall of more or less abundant snow, and later—rain; the dusty state of the atmosphere in Irkutsk (dry haze) is not rare at the end of spring and at the beginning of summer.

Aeolian dust is brought now, and was brought formerly—perhaps in greater quantity—from the arena of the amphitheater into its southern corner, a natural place for its deposition; here the arena is closed by mountains, here a change of air currents must occur; in spring, when it is already hot in the steppes, it is still cool in the mountains, the remnants of winter snow are still melting, and the moisture of the upper layers of the soil, shaded by forests, has not yet disappeared.

There is in the Irkutsk amphitheater also a second region of deposition—sands, though not in such a form of development as on the margins of true deserts; north of Irkutsk and south of the Balagansk steppes we find, in the arena of the amphitheater, a belt of sands several tens of versts wide; it begins somewhat north of the Irkut River and stretches to the Belaya River; the terrain here is flat-hilly, for the most part covered with pine forest; the soil is sandy; in places there are also bare sands. I will not assert that true shifting sands formerly existed here, only later overgrown; it is quite possible that, owing to better climatic conditions, forests have always existed here, under whose protection the coarser aeolian materials brought by northwestern winds were deposited.

In general, thanks to climatic conditions which formerly were hardly much worse than the present ones, this loess region has its own peculiarities: the area of deflation is not a desert, but a steppe, with

which the winds carry away, of course, less material; the area of sands is covered with forest, which successfully combats the weak accumulation of sand; the loess covers a comparatively small area and is not distinguished by great thickness.

Let us move still farther east, beyond Lake Baikal. In southwestern Transbaikalia the presence of loess had already been established by Chersky, and I had occasion to study this region in 1895–98 in considerable detail. The loess here is not very thick, very sandy, but nevertheless has its characteristic properties: it is unstratified, porous, contains shells of land mollusks and bones of mammals; it covers places and the floors of valleys (outside the belt of development of modern alluvium), rises onto the slopes and summits of hills and mountains, and often covers the highest points of the relief, onto which it could not have been washed from anywhere, since this relief, apart from minor details, was created before the beginning of the loess period.

The loess region of southwestern Transbaikalia has many peculiar features: the areas of winnowing and deposition are situated here intermingled. The climate of the country is dry; winters have little snow; in winter and spring winds are very frequent, raising masses of dust; the source of the latter is found here in the form of post-Tertiary lake sands, more ancient than the loess, developed along all valleys and rising rather high onto the mountain slopes. When the post-Tertiary lakes, which had drained into Lake Baikal, dried up, their sandy beds, which had occupied vast spaces, were subjected to winnowing; the dust rising from them was deposited on the neighboring heights, covered with forest or steppe; the coarser particles accumulated in the form of shifting sands on the margins of the valleys. This proximity of the areas of winnowing and deposition conditioned the peculiarity of the Transbaikal loess—its strong sandiness, since along with the dust too many fine grains of sand were deposited.

The lake deposits were gradually destroyed by winnowing and erosion, and now in places have already disappeared completely; on the floors of valleys we can not infrequently encounter outcrops of bedrock or a thickness of alluvium. But significant remnants of lake sands continue to be winnowed even now, being transformed partly into loess and partly into shifting sands, which we can encounter quite often next to loess deposits; these shifting sands, blown about by winds, also yield some quantity of dust in addition to that obtained from the lake deposits, from river alluvium exposed on river terraces, on islands and shoals during the dry season, and from the numerous weathering rocks and cliffs. I suppose that a certain quantity of aeolian dust is brought into the region described from afar—from the south, from Eastern Mongolia; this region directly passes southward into the loess region situated between Kyakhta and Urga on the northern outskirts of the Great Gobi; but this dust, having traveled so long a path, is very fine and not

from it alone the Trans-Baikal loess was formed; in it a coarser local material predominates.

With southwestern Transbaikalia, adjacent to the outskirts of the Great Gobi, the principal region of weathering in Asia, we shall conclude our brief survey of the loess regions of this continent. Everywhere—from the Caspian Sea to the banks of the Selenga River—we have seen a regular combination of regions of weathering and deposition and analogous conditions in the genesis of loess; the different thickness and different character of the loess strata found their natural explanation in the local peculiarities of climate, relief, and the areas of weathering. These conditions could not be identical over a stretch of 20 degrees of latitude and 60 of longitude, but the aeolian hypothesis proved applicable everywhere, whereas the aqueous one, for its application, would require the assistance of such unproved assumptions as the disappearance of more considerable heights from which the loess was washed down, the cutting of deep valleys after the formation of the loess, and, in general, a serious change in the relief; resorting to the aqueous hypothesis, we would have to come to the conclusion that in most of the regions described new loess can no longer form, and that only the washing-over of old strata and the transference of loess from place to place is taking place. Meanwhile in all these regions loess is being formed even now, under the present conditions of climate, relief, and vegetation.

Let us now consider in more detail the very process of “siltation,” as described by the defenders of the aqueous hypothesis.

The process of filling steppe depressions with loess, which Richthofen set forth in the first volume of China, lies at the basis of the aqueous hypothesis, but with a more or less complete elimination of the role of wind. This hypothesis assumes that outcrops of bedrock, when weathered, yield fine dust, mineral particles which are washed by rain streams onto the lower slopes, are deposited on the soil surface between the stems of steppe grasses and, accumulating in this way over the course of ages, give thicknesses of unstratified loess.

But with such a local origin of the loess material, the composition of the bedrocks ought to be strongly reflected in the composition of the loess; for one cannot imagine that on a slope of heights composed of pure white quartz there would be obtained a loess of the same composition and color as on a slope of heights of black clayey slate or of red limestone; on the first slope there should be a minimal amount of white loess, on the second—much gray, on the third—a little red; the loess of the first slope will be very poor in clayey material and soluble salts, the loess of the second will contain little quartz and lime, the loess of the third—even less quartz, but much lime and iron. Meanwhile I personally have nowhere seen, nor have I read in the literature, that loess should rapidly change in its color and composition depending on neighboring outcrops of rocks. On the contrary, over immense expanses—

countries, while lying in the vicinity of all sorts of rocks, loess retains one and the same color and approximately one and the same composition. This uniformity of loess over entire regions can be explained only if it consists chiefly of exotic dust; products of the destruction of all possible rocks, borrowed from extensive areas of deflation and mixed with one another in air currents, alone can yield such a uniform dust. In the example of China, where the reddish loess of the lower horizons is replaced by a yellowish-brown upper one, we saw how a strong change in the composition of the predominant rocks in the area of deflation is reflected in the composition of the dust; south-western Transbaikalia gave us an example of the strong influence of the composition of the prevailing deflated rocks on the composition of the nearby loess.

The example adduced by A. P. Pavlov from Turkestan¹ seems to me wholly unconvincing. The author traced how a hard white granite is replaced by a white mass of disintegrated granitic minerals; farther on, their accumulation becomes enriched with a powdery substance and acquires a yellowish color; still farther on, the powdery loess material becomes predominant, and the rock assumes the usual color of loess and effervesces strongly with acid. This observation proves only that near granite outcrops the products of its destruction predominate in the composition of the soil, which is entirely natural not only for loess countries but for eluvium and deluvium in any locality; the farther from the granite outcrops, the more the granitic gruss, already in a more decomposed state (since it has been subjected longer to weathering processes), is mixed with loess dust brought by winds, which finally becomes dominant. To prove that the given loess originated from the local granite, it is necessary to make analyses of the fresh granite, of its gruss at various stages of decomposition, and finally of the pure loess; to trace and explain the regularity of those changes in chemical and mineralogical composition which would certainly be found in that case, if the process of destruction of the granite really yielded loess as its final result.

A. P. Pavlov did not give such proof. But he goes on to say: “rather considerable masses of loess, but with frequent intercalating layers of unrolled minerals of the disintegrated granite, are already observed in the valleys of Ak-tau.” Why, then, did the granitic gruss remain in these interlayers and not also turn into loess? This must be explained by the preceding words of the author: “among the examples of the formation of loess in this way, one proved especially interesting in that it made it possible to trace the different stages of fragmentation or disintegration of the granite forming the Ak-tau mountains, the gradual change

the greater part of the products of this disintegration into loess-like material; next the stage of sorting or washing out of this material by proluvial streams¹) and the formation of the loess beds of the Transcaspian ravines.”

Thus, consequently, in addition to the processes of weathering, there occurs a process of washing out its products by proluvial streams. But both the one and the other processes would lead to quite different results than the author supposes; let us examine them in somewhat more detail.

In countries with a dry climate, physical weathering predominates; such a heterogeneous and coarse-grained rock as granite first breaks down into its constituent mineral parts—quartz, feldspars, mica and its equivalents, magnetite, titanite; then these minerals in turn break down into pieces along cracks of cleavage and pressure; moisture, soaking the granite detritus during rain, melting snow, and dew, makes chemical processes possible as well; the destruction of feldspars begins—their transformation into clay with removal of $K_2O$, $Na_2O$, $CaO$; biotite and hornblende acquire a rusty color, owing to the formation of iron oxides extracted from the bisilicate; magnetite turns into brown iron ore and ocher. The soluble salts of alkalis and alkaline earths that have formed are gradually leached out by water and carried by it either downward or along the surface in rain rills; in the end, granite, a rock rich in quartz, is transformed into clayey sand, colored a more or less bright yellow depending on the amount of bisilicates and magnetite that yielded hydrous iron oxides.

What can the processes of washing out by proluvial streams do with this ocher-clayey sand? They must carry away the lighter clay particles, leaving the heavier grains of sand; they must also carry away soluble salts. Consequently, near the granite there should remain sand, while the clay will be deposited lower down the slope. If it is assumed that only rains of slight intensity are at work, then there will be a gradual transition from coarser sand lying near the granite to clay situated farthest away. But rains are, of course, of varying intensity; somewhat stronger ones wash away not only clay but also sand, and still stronger ones—also pieces of more or less decomposed grus; as the rain rills split up and lose water by infiltration into the soil, they also lose their material; grus and sand settle earlier and nearer, clay later and farther away; on a layer of clay deposited by weak rains there will settle a thin layer of sand and grus or of sand alone; then for some time clay will again be deposited in the given place; a heavy downpour will drag downward

¹) Italics mine.

On the Question of the Origin of Loess

along the slope not only grus, but also crushed granite, and deposit them in a thin, tapering layer over sand or clay. In short, when proluvial streams operate, a deposit must result that is layered; the stratification may be not very clear, but nevertheless noticeable, owing to the intermingling of material of different grain sizes, brought by streams of different strength to a given place; a homogeneous, non-stratified loess cannot result.

There are no signs of that sorting of material of which A. P. Pavlov speaks in the example he describes, except for tapering interbeds of granite grus; these interbeds are indeed of proluvial origin; they were created by heavy rains, as Richthofen long ago explained; they occur in loess everywhere near mountain slopes and outcrops of bedrock. But if there had been a more delicate sorting, then near the granite we should have had to see a predominance of light, coarser sand, passing downslope into finer and more clayey sand with indistinct interbeds of clay, and still lower—predominant clay or loam with interbeds of fine sand. In short, the color and composition of the soil would have had to change with distance from the bedrock, even if the roots of steppe grasses had ultimately destroyed the original stratification of this soil. But this destruction would have to be stronger in the deeper horizons, whereas near the surface the stratification of the soil, in any case, should be noticeable.

There is no doubt that rain streams carry onto the loess steppe of a mountain slope the products of destruction of the rocks lying above; weak ones bring grains of sand, clay, flakes of mica, stronger ones—grus and rubble. That is why in the loess of slopes one can always find interbeds of rubble, interbeds of sand—thin, tapering, as the results of the work of a single rain. But this heterogeneous proluvial material, so to speak, sinks into the mass of homogeneous, ash-gray material brought from regions of deflation, deposited over the entire slope, from day to day, in greater or lesser quantity, whereas rains in these localities work only rarely, with interruptions of weeks and months. Moreover, it is more often not heavy but weak rains that occur; they merely moisten the dust accumulated on the soil surface, wash the dust from grass stems, and deposit it nearby on the soil, but are not able to move even this fine material downslope, because the rainwater is greedily absorbed by the dry and porous soil. These weak rains play a major role in the accumulation of loess, since dust particles lying freely on the surface and washed off the grass coalesce with one another and, losing their mobility, adhere to the underlying loess mass; therefore a subsequent, stronger rain can no longer wash them away, and a strong wind can no longer carry them off.

Typical loess, by the manner of its deposition, is a stratified rock, because it is deposited not continuously; every rain that frees the atmosphere from dust causes an interruption in the process of deposition; that same rain, beating the accumulated dust down to the soil, creates a surface of lamination of the loess. But since the material—exotic dust—is always deposited homogeneously, at least over long periods, the thin layer of loess deposited in the interval between two rains cannot be sharply delimited from the overlying and underlying thin layers of identical composition; only an extraordinary downpour, which may occur once in several years, will carry gravel, gruss, and sand onto the surface of the steppe and, owing to the sharp difference in composition, grain size, and mode of deposition of the material, will produce a noticeable layer.

It should not be forgotten that the process of accumulation of aeolian loess is very slow. The only observation that makes it possible to express the annual increment of loess in figures is given to us by L. Loczy: in the city of Si-an-fu, in the province of Shen-si, he saw a layer of loess 2 meters thick on Chinese graves having an antiquity of 2000 years¹); consequently, in a year the loess increases by only one millimeter. But in the course of a year there may have been several rains that beat down the dust; consequently each of the loess layers can have a thickness only in tenths of a millimeter. It is understandable that such fine stratification in a perfectly homogeneous material cannot be noticeable, all the more since the dust particles, moistened by rain, adhere not only to one another, but also to the surface of the soil, i.e. to the preceding plane of lamination, thereby destroying this plane as a boundary between two layers.

In proluvial formation of loess there can be no such destruction of the planes of lamination: deposition takes place from water, on a surface wetted and smoothed by water, while the material, as indicated above, is not entirely homogeneous in two adjacent layers. After all, one cannot suppose that a given rain washes down from the outcrops of granite only those products of destruction that are in one and the same stage of alteration; it will wash down both a particle of fresh feldspar and a particle of clay, a grain of sand, flakes of mica in various stages of decomposition, a particle of ocher, and a particle of magnetite, and may deposit side by side particles of the same specific gravity but of different composition. In the deposited material the processes of weathering must continue, and as a result banding and mottling of the soil are obtained, for example a rusty coloration around an introduced particle of magnetite, biotite, or hornblende. In loess we observe such phenomena extremely rarely; in alluvium, eluvium, and deluvium—often.

¹) According to I. Walther: Das Gesetz der Wüstenbildung, p. 138.

Finally, one cannot but emphasize once more that, in the proluvial origin of loess, the features of the composition of the bedrock emerging in the neighborhood and providing material for this loess must inevitably be reflected in its color and composition; the products of mechanical and chemical weathering of different rocks, especially in the intermediate stages of destruction, are different, and the formation of homogeneous loess upon heterogeneous bedrock by a proluvial process is entirely inexplicable.

I have still to say a few words concerning the loess of Southern Russia; its origin, as is known, is explained by P. A. Tutkovsky with the aid of the same aeolian hypothesis: the great Finnish-Scandinavian glacier, retreating to the north, left behind vast moraine areas, real silty-sandy-boulder deserts, upon which the material was winnowed by cold winds blowing from the glacier, as from places of permanent barometric maximum, to the south, whither they carried the fine dust and deposited it on the steppes lying south of the belt of these moraine deserts¹.

The orderly hypothesis of this author, apparently, must undergo certain modifications in view of the results of the most recent work; thus, for example, N. I. Krishtafovich established, by detailed investigations of the environs of Lublin, that loess must have been formed not during the retreat, but during the advance of the glacier, both of the Saxonian epoch and, in particular, of the Polish-Mecklenburg epoch². This, of course, changes the picture drawn by P. Tutkovsky.

Without entering into the details of both views and without denying the role of the retreating glacier in the process of formation of the loess of Southern Russia, I wish only to point out the possibility of another source for part of the aeolian dust that created this loess. If one looks at the soil map of European Russia, published by the Geological Committee, one cannot fail to notice that the sands of the lower Terek, Kuma, Volga, and Ural are bordered on the northwest by steppe brown soils—at first sandy loams, then light-brown loams, and farther on chestnut soils, on which, along the watersheds, there appears light loamy and clayey chernozem with 4–6% humus. Already beyond the line from the mouth of the Don to Samara, chernozem becomes medium, with 6–8% humus, and occupies an extensive area in the form of a belt extending WSW—ONO, wider in the west, in places only passing into rich chernozem with 8–10% humus.

The question involuntarily arises: is not, in the present case as well, the neighborhood of sands and steppe soils, i.e. loess, regular? Was not part of the aeolian dust brought from the southeast, from the Ust-Urt deserts, and part

of the Caspian sands from the same source? Is the difference in the composition of the steppe soils to the east and west of the loess belt not connected with this, since in the east the amount of Asiatic dust must have been greater? It seems to me that this question deserves the attention of geologists studying the loess of Southern Russia.

But in the further development of the question of the origin of loess in general on the basis of the study of South-Russian loess, one should not forget that this loess has already undergone many vicissitudes. Owing to changed climatic conditions, to the comparatively abundant precipitation, and also to human activity, we see in Southern Russia much loess that is secondary, redeposited—proluvial, deluvial, alluvial—which undoubtedly greatly obscures the matter and complicates the question. Is this not why the principal defenders of the fluvial and other non-aeolian hypotheses are geologists who have worked exclusively in southern Russia and in Western Europe, where the conditions for studying loess are just as complex? It seems to me that, in order to assess the full plausibility, the full persuasiveness of the aeolian hypothesis, it is necessary to study loess in Asia, in those countries where the climate and other conditions have essentially not changed, where the formation of aeolian dust and the accumulation of loess are still taking place. Thanks to travels in Asia, the aeolian hypothesis arose and was developed; in that same country one must seek arguments and facts for its refutation—if only such are to be found, of which I strongly doubt. A. P. Pavlov made the first attempt to test the fluvial hypothesis under Asiatic conditions; how successfully—I have tried to show above.

Tomsk,
20 February 1911.

V. A. Obruchev.

Fig. 1.
\(k\) — bedrock; \(l\) — ice.

Fig. 2.
\(k\) — bedrock; \(l\) — ice; \(\rho\) — morainic mass; \(a\) — [[unclear: handwritten label]]; \(a'\) — [[unclear: handwritten label]].

Fig. 3.
\(k\) — bedrock; \(l\) — ice.

Fig. 4.
\(k\) — bedrock; \(l\) — ice.

Fig. 5.
\(k\) — bedrock; \(s\) — sands; \(o\) — lacustrine deposits.

Fig. 6.
Area of erosion.

Loess \< moraine \< ground moraine → sands, loess.

Fig. 7.
This scheme was taken by S. A. Obruchev with the question of the origin of loess.

Dammed lakes of East [[unclear: Barranco/Baranko]].

Loesses; terminal moraine; sands; ground moraine.

Labels visible on the map: Danube valley; Pannonian Basin; Germany; Copenhagen; Scandinavia; England; Baltic Sea; White Sea; Ukraine; Black Sea; Azov Sea; Caspian Sea; Ural; Zap.-Sibirskoye [[unclear: lake/sea]]; [[unclear: several place names]].

1. Zemlevedenie, 1899, books I–II, pp. 250–298. ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

2. Transactions of the Novo-Alexandria Institute of Agriculture and Forestry, vol. XV, issue 3, 1902, pp. 184–192. ↩↩↩↩↩↩↩↩

3. Verh. d. Ges. f. Erd. zu Berlin, 1892, vol. XIX, pp. 52–65. ↩↩↩↩