The next step in our history of glaciation is to outline the disappearance of the ice sheets. When a decrease in solar activity produced a corresponding decrease in storminess, several influences presumably combined to cause the disappearance of the ice. Most of their results are the reverse of those which brought on glaciation. A few special aspects, however, some of which have been discussed in _Earth and Sun_, ought to be brought to mind. A diminution in storminess lessens upward convection, wind velocity, and evaporation, and these changes, if they occurred, must have united to raise the temperature of the lower air by reducing the escape of heat. Again a decrease in the number and intensity of tropical cyclones presumably lessened the amount of moisture carried into mid-lat.i.tudes, and thus diminished the precipitation. The diminution of snowfall on the ice sheets when storminess diminished was probably highly important. The amount of precipitation on the sheets was presumably lessened still further by changes in the storminess of middle lat.i.tudes. When storminess diminishes, the lows follow a less definite path, as Kullmer's maps show, and on the average a more southerly path. Thus, instead of all the lows contributing snow to the ice sheet, a large fraction of the relatively few remaining lows would bring rain to areas south of the ice sheet. As storminess decreased, the trades and westerlies probably became steadier, and thus carried to high lat.i.tudes more warm water than when often interrupted by storms. Steadier southwesterly winds must have produced a greater movement of atmospheric as well as oceanic heat to high lat.i.tudes. The warming due to these two causes was probably the chief reason for the disappearance of the European ice sheet and of those on the Pacific coast of North America. The two greater American ice sheets, however, and the glaciers elsewhere in the lee of high mountain ranges, probably disappeared chiefly because of lessened precipitation. If there were no cyclonic storms to draw moisture northward from the Gulf of Mexico, most of North America east of the Rocky Mountain barrier would be arid. Therefore a diminution of storminess would be particularly effective in causing the disappearance of ice sheets in these regions.

That evaporation was an especially important factor in causing the ice from the Keewatin center to disappear, is suggested by the relatively small amount of water-sorted material in its drift. In South Dakota, for example, less than 10 per cent of the drift is stratified.[45] On the other hand, Salisbury estimates that perhaps a third of the Labradorean drift in eastern Wisconsin is crudely stratified, about half of that in New Jersey, and more than half of the drift in western Europe.

When the sun's activity began to diminish, all these conditions, as well as several others, would cooperate to cause the ice sheets to disappear.

Step by step with their disappearance, the amelioration of the climate would progress so long as the period of solar inactivity continued and storms were rare. If the inactivity continued long enough, it would result in a fairly mild climate in high lat.i.tudes, though so long as the continents were emergent this mildness would not be of the extreme type.

The inauguration of another cycle of increased disturbance of the sun, with a marked increase in storminess, would inaugurate another glacial epoch. Thus a succession of glacial and inter-glacial epochs might continue so long as the sun was repeatedly disturbed.

FOOTNOTES:

[Footnote 38: This chapter is an amplification and revision of the sketch of the glacial period contained in The Solar Hypothesis of Climatic Changes; Bull. Geol. Soc. Am., Vol. 25, 1914.]

[Footnote 39: R. D. Salisbury: Physical Geography of the Pleistocene, in Outlines of Geologic History, by Willis, Salisbury, and others, 1910, p.

265.]

[Footnote 40: The Quaternary Ice Age, 1914, p. 364.]

[Footnote B: For fuller discussion of climatic controls see S. S.

Visher: Seventy Laws of Climate, Annals a.s.soc. Am. Geographers, 1922.]

[Footnote 41: Many of these alterations are implied or discussed in the following papers:

1. F. W. Harmer: Influence of Winds upon the Climate of the Pleistocene; Quart. Jour. Geol. Soc., Vol. 57, 1901, p. 405.

2. C. E. P. Brooks: Meteorological Conditions of an Ice Sheet; Quart.

Jour. Royal Meteorol. Soc., Vol. 40, 1914, pp. 53-70, and The Evolution of Climate in Northwest Europe; _op. cit._, Vol. 47, 1921, pp. 173-194.

3. W. H. Hobbs: The Role of the Glacial Anticyclone in the Air Circulation of the Globe; Proc. Am. Phil. Soc., Vol. 54, 1915, pp.

185-225.]

[Footnote 42: W. B. Wright: The Quaternary Ice Age, 1914, p. 100.]

[Footnote 43: The description of the distribution of the ice sheet is based on T. C. Chamberlin's wall map of North America at the maximum of glaciation, 1913.]

[Footnote 44: Chamberlin and Salisbury: Geology, 1906, Vol. 3, and W. H.

Hobbs: Characteristics of Existing Glaciers, 1911.]

[Footnote 45: S. S. Visher: The Geography of South Dakota; S. D. Geol.

Surv., 1918.]

CHAPTER VIII

SOME PROBLEMS OF GLACIAL PERIODS

Having outlined in general terms the coming of the ice sheets and their disappearance, we are now ready to discuss certain problems of compelling climatic interest. The discussion will be grouped under five heads: (I) the localization of glaciation; (II) the sudden coming of glaciation; (III) peculiar variations in the height of the snow line and of glaciation; (IV) lakes and other evidences of humidity in unglaciated regions during the glacial epochs; (V) glaciation at sea level and in low lat.i.tudes in the Permian and Proterozoic eras. The discussion of perhaps the most difficult of all climatic problems of glaciation, that of the succession of cold glacial and mild inter-glacial epochs, has been postponed to the next to the final chapter of this book. It cannot be properly considered until we take up the history of solar disturbances.

I. The first problem, the localization of the ice sheets, arises from the fact that in both the Pleistocene and the Permian periods glaciation was remarkably limited. In neither period were all parts of high lat.i.tudes glaciated; yet in both cases glaciation occurred in large regions in lower lat.i.tudes. Many explanations of this localization have been offered, but most are entirely inadequate. Even hypotheses with something of proven worth, such as those of variations in volcanic dust and in atmospheric carbon dioxide, fail to account for localization. The cyclonic form of the solar hypothesis, however, seems to afford a satisfactory explanation.

The distribution of the ice in the last glacial period is well known, and is shown in Fig. 6. Four-fifths of the ice-covered area, which was eight million square miles, more or less, was near the borders of the North Atlantic in eastern North America and northwestern Europe. The ice spread out from two great centers in North America, the Labradorean east of Hudson Bay, and the Keewatin west of the bay. There were also many glaciers in the western mountains, especially in Canada, while subordinate centers occurred in Newfoundland, the Adirondacks, and the White Mountains. The main ice sheet at its maximum extension reached as far south as lat.i.tude 39 in Kansas and Kentucky, and 37 in Illinois.

Huge boulders were transferred more than one thousand miles from their source in Canada. The northward extension was somewhat less. Indeed, the northern margin of the continent was apparently relatively little glaciated and much of Alaska unglaciated. Why should northern Kentucky be glaciated when northern Alaska was not?

In Europe the chief center from which the continental glacier moved was the Scandinavian highlands. It pushed across the depression now occupied by the Baltic to southern Russia and across the North Sea depression to England and Belgium. The Alps formed a center of considerable importance, and there were minor centers in Scotland, Ireland, the Pyrenees, Apennines, Caucasus, and Urals. In Asia numerous ranges also contained large glaciers, but practically all the glaciation was of the alpine type and very little of the vast northern lowland was covered with ice.

In the southern hemisphere glaciation at low lat.i.tudes was less striking than in the northern hemisphere. Most of the increase in the areas of ice was confined to mountains which today receive heavy precipitation and still contain small glaciers. Indeed, except for relatively slight glaciation in the Australian Alps and in Tasmania, most of the Pleistocene glaciation in the southern hemisphere was merely an extension of existing glaciers, such as those of south Chile, New Zealand, and the Andes. Nevertheless, fairly extensive glaciation existed much nearer the equator than is now the case.

In considering the localization of Pleistocene glaciation, three main factors must be taken into account, namely, temperature, topography, and precipitation. The absence of glaciation in large parts of the Arctic regions of North America and of Asia makes it certain that low temperature was not the controlling factor. Aside from Antarctica, the coldest place in the world is northeastern Siberia. There for seven months the average temperature is below 0C., while the mean for the whole year is below -10C. If the temperature during a glacial period averaged 6C. lower than now, as is commonly supposed, this part of Siberia would have had a temperature below freezing for at least nine months out of the twelve even if there were no snowfield to keep the summers cold. Yet even under such conditions no glaciation occurred, although in other places, such as parts of Canada and northwestern Europe, intense glaciation occurred where the mean temperature is much higher.

The topography of the lands apparently had much more influence upon the localization of glaciation than did temperature. Its effect, however, was always to cause glaciation exactly where it would be expected and not in unexpected places as actually occurred. For example, in North America the western side of the Canadian Rockies suffered intense glaciation, for there precipitation was heavy because the westerly winds from the Pacific are forced to give up their moisture as they rise. In the same way the western side of the Sierra Nevadas was much more heavily glaciated than the eastern side. In similar fashion the windward slopes of the Alps, the Caucasus, the Himalayas, and many other mountain ranges suffered extensive glaciation. Low temperature does not seem to have been the cause of this glaciation, for in that case it is hard to see why both sides of the various ranges did not show an equal percentage of increase in the size of their icefields.

From what has been said as to temperature and topography, it is evident that variations in precipitation have had much more to do with glaciation than have variations in temperature. In the Arctic lowlands and on the leeward side of mountains, the slight development of glaciation appears to have been due to scarcity of precipitation. On the windward side of mountains, on the other hand, a notable increase in precipitation seems to have led to abundant glaciation. Such an increase in precipitation must be dependent on increased evaporation and this could arise either from relatively high temperature or strong winds.

Since the temperature in the glacial period was lower than now, we seem forced to attribute the increased precipitation to a strengthening of the winds. If the westerly winds from the Pacific should increase in strength and waft more moisture to the western side of the Canadian Rockies, or if similar winds increased the snowfall on the upper slopes of the Alps or the Tian-Shan Mountains, the glaciers would extend lower than now without any change in temperature.

Although the incompetence of low temperature to cause glaciation, and the relative unimportance of the mountains in northeastern Canada and northwestern Europe throw most glacial hypotheses out of court, they are in harmony with the cyclonic hypothesis. The answer of that hypothesis to the problem of the localization of ice sheets seems to be found in certain maps of storminess and rainfall in relation to solar activity.

In Fig. 2 a marked belt of increased storminess at times of many sunspots is seen in southern Canada. A comparison of this with a series of maps given in _Earth and Sun_ shows that the stormy belt tends to migrate northward in harmony with an increase in the activity of the sun's atmosphere. If the sun were sufficiently active the belt of maximum storminess would apparently pa.s.s through the Keewatin and Labradorean centers of glaciation instead of well to the south of them, as at present. It would presumably cross another center in Greenland, and then would traverse the fourth of the great centers of Pleistocene glaciation in Scandinavia. It would not succeed in traversing northern Asia, however, any more than it does now, because of the great high-pressure area which develops there in winter. When the ice sheets expanded from the main centers of glaciation, the belt of storms would be pushed southward and outward. Thus it might give rise to minor centers of glaciers such as the Patrician between Hudson Bay and Lake Superior, or the centers in Ireland, Cornwall, Wales, and the northern Ural Mountains. As the main ice sheets advanced, however, the minor centers would be overridden and the entire ma.s.s of ice would be merged into one vast expanse in the Atlantic portion of each of the two continents.

In this connection it may be well to consider briefly the most recent hypothesis as to the growth and hence the localization of glaciation. In 1911 and more fully in 1915, Hobbs,[46] advanced the anti-cyclonic hypothesis of the origin of ice sheets. This hypothesis has the great merit of focusing attention upon the fact that ice sheets are p.r.o.nounced anti-cyclonic regions of high pressure. This is proved by the strong outblowing winds which prevail along their margins. Such winds must, of course, be balanced by inward-moving winds at high levels. Abundant observations prove that such is the case. For example, balloons sent up by Barkow near the margin of the Antarctic ice sheet reveal the occurrence of inblowing winds, although they rarely occur below a height of 9000 meters. The abundant data gathered by Guervain on the coast of Greenland indicate that outblowing winds prevail up to a height of about 4000 meters. At that height inblowing winds commence and increase in frequency until at an alt.i.tude of over 5000 meters they become more common than outblowing winds. It should be noted, however, that in both Antarctica and Greenland, although the winds at an elevation of less than a thousand meters generally blow outward, there are frequent and decided departures from this rule, so that "variable winds" are quite commonly mentioned in the reports of expeditions and balloon soundings.

The undoubted anti-cyclonic conditions which Hobbs thus calls to the attention of scientists seem to him to necessitate a peculiar mechanism in order to produce the snow which feeds the glaciers. He a.s.sumes that the winds which blow toward the centers of the ice sheets at high levels carry the necessary moisture by which the glaciers grow. When the air descends in the centers of the highs, it is supposed to be chilled on reaching the surface of the ice, and hence to give up its moisture in the form of minute crystals. This conclusion is doubtful for several reasons. In the first place, Hobbs does not seem to appreciate the importance of the variable winds which he quotes Arctic and Antarctic explorers as describing quite frequently on the edges of the ice sheets.

They are one of many signs that cyclonic storms are fairly frequent on the borders of the ice though not in its interior. Thus there is a distinct and sufficient form of precipitation actually at work near the margin of the ice, or exactly where the thickness of the ice sheet would lead us to expect.

Another consideration which throws grave doubt on the anti-cyclonic hypothesis of ice sheets is the small amount of moisture possible in the highs because of their low temperature. Suppose, for the sake of argument, that the temperature in the middle of an ice sheet averages 20F. This is probably much higher than the actual fact and therefore unduly favorable to the anti-cyclonic hypothesis. Suppose also that the decrease in temperature from the earth's surface upward proceeds at the rate of 1F. for each 300 feet, which is 50 per cent less than the actual rate for air with only a slight amount of moisture, such as is found in cold regions. Then at a height of 10,000 feet, where the inblowing winds begin to be felt, the temperature would be -20F. At that temperature the air is able to hold approximately 0.166 grain of moisture per cubic foot when fully saturated. This is an exceedingly small amount of moisture and even if it were all precipitated could scarcely build a glacier. However, it apparently would not be precipitated because when such air descends in the center of the anti-cyclone it is warmed adiabatically, that is, by compression. On reaching the surface it would have a temperature of 20 and would be able to hold 0.898 grain of water vapor per cubic foot; in other words, it would have a relative humidity of about 18 per cent. Under no reasonable a.s.sumption does the upper air at the center of an ice sheet appear to reach the surface with a relative humidity of more than 20 or 25 per cent. Such air cannot give up moisture. On the contrary, it absorbs it and tends to diminish rather than increase the thickness of the sheet of ice and snow. But after the surplus heat gained by descent has been lost by radiation, conduction, and evaporation, the air may become super-saturated with the moisture picked up while warm. Hobbs reports that explorers in Antarctica and Greenland have frequently observed condensation on their clothing. If such moisture is not derived directly from the men's own bodies, it is apparently picked up from the ice sheet by the descending air, and not added to the ice sheet by air from aloft.

The relation of all this to the localization of ice sheets is this. If Hobbs' anti-cyclonic hypothesis of glacial growth is correct, it would appear that ice sheets should grow up where the temperature is lowest and the high-pressure areas most persistent; for instance, in northern Siberia. It would also appear that so far as the topography permitted, the ice sheets ought to move out uniformly in all directions; hence the ice sheet ought to be as prominent to the north of the Keewatin and Labradorean centers as to the south, which is by no means the case.

Again, in mountainous regions, such as the glacial areas of Alaska and Chile, the glaciation ought not to be confined to the windward slope of the mountains so closely as is actually the fact. In each of these cases the glaciated region was large enough so that there was probably a true anti-cyclonic area comparable with that now prevailing over southern Greenland. In both places the correlation between glaciation and mountain ranges seems much too close to support the anti-cyclonic hypothesis, for the inblowing winds which on that hypothesis bring the moisture are shown by observation to occur at heights far greater than that of all but the loftiest ranges.

II. The sudden coming of glaciation is another problem which has been a stumbling-block in the way of every glacial hypothesis. In his _Climates of Geologic Times_, Schuchert states that the fossils give almost no warning of an approaching catastrophe. If glaciation were solely due to uplift, or other terrestrial changes aside from vulcanism, Schuchert holds that it would have come slowly and the stages preceding glaciation would have affected life sufficiently to be recorded in the rocks. He considers that the suddenness of the coming of glaciation is one of the strongest arguments against the carbon dioxide hypothesis of glaciation.

According to the cyclonic hypothesis, however, the suddenness of the oncoming of glaciation is merely what would be expected on the basis of what happens today. Changes in the sun occur suddenly. The sunspot cycle is only eleven or twelve years long, and even this short period of activity is inaugurated more suddenly than it declines. Again the climatic record derived from the growth of trees, as given in Figs. 4 and 5, also shows that marked changes in climate are initiated more rapidly than they disappear. In this connection, however, it must be remembered that solar activity may arise in various ways, as will appear more fully later. Under certain conditions storminess may increase and decrease slowly.

III. The height of the snow line and of glaciation furnishes another means of testing glacial hypotheses. It is well established that in times of glaciation the snow line was depressed everywhere, but least near the equator. For example, according to Penck, permanent snow extended 4000 feet lower than now in the Alps, whereas it stood only 1500 feet below the present level near the equator in Venezuela. This unequal depression is not readily accounted for by any hypothesis depending solely upon the lowering of temperature. By the carbon dioxide and the volcanic dust hypotheses, the temperature presumably was lowered almost equally in all lat.i.tudes, but a little more at the equator than elsewhere. If glaciation were due to a temporary lessening of the radiation received from the sun, such as is demanded by the thermal solar hypothesis, and by the longer periods of Croll's hypothesis, the lowering would be distinctly greatest at the equator. Thus, according to all these hypotheses, the snow line should have been depressed most at the equator, instead of least.

The cyclonic hypothesis explains the lesser depression of the snow line at the equator as due to a diminution of precipitation. The effectiveness of precipitation in this respect is ill.u.s.trated by the present great difference in the height of the snow line on the humid and dry sides of mountains. On the wet eastern side of the Andes near the equator, the snow line lies at 16,000 feet; on the dry western side, at 18,500 feet. Again, although the humid side of the Himalayas lies toward the south, the snow line has a level of 15,000 feet, while farther north, on the dry side, it is 16,700 feet.[47] The fact that the snow line is lower near the margin of the Alps than toward the center points in the same direction. The bearing of all this on the glacial period may be judged by looking again at Fig. 3 in Chapter V. This shows that at times of sunspot activity and hence of augmented storminess, the precipitation diminishes near the heat equator, that is, where the average temperature for the whole year is highest. At present the great size of the northern continents and their consequent high temperature in summer, cause the heat equator to lie north of the "real" equator, except where Australia draws it to the southward.[48] When large parts of the northern continents were covered with ice, however, the heat equator and the true equator were probably much closer than now, for the continents could not become so hot. If so, the diminution in equatorial precipitation, which accompanies increased storminess throughout the world as a whole, would take place more nearly along the true equator than appears in Fig. 3. Hence so far as precipitation alone is concerned, we should actually expect that the snow line near the equator would rise a little during glacial periods. Another factor, however, must be considered. Koppen's data, it will be remembered, show that at times of solar activity the earth's temperature falls more at the equator than in higher lat.i.tudes. If this effect were magnified it would lower the snow line. The actual position of the snow line at the equator during glacial periods thus appears to be the combined effect of diminished precipitation, which would raise the line, and of lower temperature, which would bring it down.

Before leaving this subject it may be well to recall that the relative lessening of precipitation in equatorial lat.i.tudes during the glacial epochs was probably caused by the diversion of moisture from the trade-wind belt. This diversion was presumably due to the great number of tropical cyclones and to the fact that the cyclonic storms of middle lat.i.tudes also drew much moisture from the trade-wind belt in summer when the northern position of the sun drew that belt near the storm track which was forced to remain south of the ice sheet. Such diversion of moisture out of the trade-wind belt must diminish the amount of water vapor that is carried by the trades to equatorial regions; hence it would lessen precipitation in the belt of so-called equatorial calms, which lies along the heat equator rather than along the geographical equator.

Another phase of the vertical distribution of glaciation has been the subject of considerable discussion. In the Alps and in many other mountains the glaciation of the Pleistocene period appears to have had its upper limit no higher than today. This has been variously interpreted. It seems, however, to be adequately explained as due to decreased precipitation at high alt.i.tudes during the cold periods. This is in spite of the fact that precipitation in general increased with increased storminess. The low temperature of glacial times presumably induced condensation at lower alt.i.tudes than now, and most of the precipitation occurred upon the lower slopes of the mountains, contributing to the lower glaciers, while little of it fell upon the highest glaciers. Above a moderate alt.i.tude in all lofty mountains the decrease in the amount of precipitation is rapid. In most cases the decrease begins at a height of less than 3000 feet above the base of the main slope, provided the slope is steep. The colder the air, the lower the alt.i.tude at which this occurs. For example, it is much lower in winter than in summer. Indeed, the higher alt.i.tudes in the Alps are sunny in winter even where there are abundant clouds lower down.

IV. The presence of extensive lakes and other evidences of a pluvial climate during glacial periods in non-glaciated regions which are normally dry is another of the facts which most glacial hypotheses fail to explain satisfactorily. Beyond the ice sheets many regions appear to have enjoyed an unusually heavy precipitation during the glacial epochs.

The evidence of this is abundant, including numerous abandoned strand lines of salt lakes and an abundance of coa.r.s.e material in deltas and flood plains. J. D. Whitney,[49] in an interesting but neglected volume, was one of the first to marshal the evidence of this sort. More recently Free[50] has amplified this. According to him in the Great Basin region of the United States sixty-two basins either contain unmistakable evidence of lakes, or belong to one of the three great lake groups named below. Two of these, the Lake Lahontan and the Lake Bonneville groups, comprise twenty-nine present basins, while the third, the Owens-Searles chain, contained at least five large lakes, the lowest being in Death Valley. In western and central Asia a far greater series of salt lakes is found and most of these are surrounded by strands at high levels.

Many of these are described in _Explorations in Turkestan_, _The Pulse of Asia_, and _Palestine and Its Transformation_. There has been a good deal of debate as to whether these lakes actually date from the glacial period, as is claimed by C. E. P. Brooks, for example, or from some other period. The evidence, however, seems to be convincing that the lakes expanded when the ice also expanded.

According to the older glacial hypotheses the lower temperature which is postulated as the cause of glaciation would almost certainly mean less evaporation over the oceans and hence less precipitation during glacial periods. To counteract this the only way in which the level of the lakes could be raised would be because the lower temperature would cause less evaporation from their surfaces. It seems quite impossible, however, that the lowering of temperature, which is commonly taken to have been not more than 10C., could counteract the lessened precipitation and also cause an enormous expansion of most of the lakes. For example, ancient Lake Bonneville was more than ten times as large as its modern remnant, Great Salt Lake, and its average depth more than forty times as great.[51] Many small lakes in the Old World expanded still more.[52]

For example, in eastern Persia many basins which now contain no lake whatever are floored with vast deposits of lacustrine salt and are surrounded by old lake bluffs and beaches. In northern Africa similar conditions prevail.[53] Other, but less obvious, evidence of more abundant rainfall in regions that are now dry is found in thick strata of gravel, sand, and fine silt in the alluvial deposits of flood plains and deltas.[54]

The cyclonic hypothesis supposes that increased storminess accounts for pluvial climates in regions that are now dry just as it accounts for glaciation in the regions of the ice sheets. Figs. 2 and 3, it will be remembered, ill.u.s.trate what happens when the sun is active. Solar activity is accompanied by an increase in storminess in the southwestern United States in exactly the region where elevated strands of diminished salt lakes are most numerous. In Fig. 3, the same condition is seen in the region of salt lakes in the Old World. Judging by these maps, which ill.u.s.trate what has happened since careful meteorological records were kept, an increase in solar activity is accompanied by increased rainfall in large parts of what are now semi-arid and desert regions. Such precipitation would at once cause the level of the lakes to rise. Later, when ice sheets had developed in Europe and America, the high-pressure areas thus caused might force the main storm belt so far south that it would lie over these same arid regions. The increase in tropical hurricanes at times of abundant sunspots may also have a bearing on the climate of regions that are now arid. During the glacial period some of the hurricanes probably swept far over the lands. The numerous tropical cyclones of Australia, for example, are the chief source of precipitation for that continent.[55] Some of the stronger cyclones locally yield more rain in a day or two than other sources yield in a year.