these beds which have been cut back so as to leave the broad shelf or plain on either side between the edges of the central ravine and the lines of cliff which bound the upper part of the valley. The outline of the whole valley is then similar to that in fig. 35, only that the edges of the upper portion are not sloping, but nearly vertical.1

The reason why these great ravines are such excellent instances of river erosion is because they were made almost entirely by the rivers alone, with little assistance from any other detritive agencies. The district drained by the Colorado and its tributaries is a series of elevated plateaux: and there is little doubt that, when the course of the river was first determined, the country stood at a much lower level, and the tributary streams ran in channels of no great depth. The cañon district is now from 4,000 to 8,000 feet above the sea, and the mountains which form the watershed rise to 10,000 and 14,000 feet. These snow-clad summits furnish a perpetual supply of water, which on reaching the plateaus above the cañons becomes charged with sand and mud.

Two circumstances have combined to maintain the verticality of the cañon walls during the process of erosion: the first is a decrease in the rainfall of the cañon district, for at the present time it is practically rainless, and we have seen that gorges are steep and narrow in proportion as the rocks are hard and the rainfall small; secondly, the beds through which they are cut are for the most part horizontal, or very gently inclined, and are traversed by strong vertical planes of jointing, so that they break away naturally in vertical lines.

The conditions which have enabled rivers to cut their way through what are now ridges and mountains, the forces which retard or increase their power of erosion, and the limits which are set to their action, are matters that will be discussed in the third portion of this volume.

"Ex

1 For descriptions and illustrations of these cañons, see ploration of the Colorado River," by Ives and Newberry, 1861, and the later "Memoirs" by J. W. Powell, 1875, and C. E. Dutton, 1882.

FORM

ORMATION of Glaciers.-On high mountains and in high latitudes, as in Greenland or Lapland, the water precipitated from the atmosphere falls as snow instead of rain. This occurs even in the Torrid Zone wherever the land exceeds an altitude of 15,000 or 16,000 feet above the sea, and in all other latitudes at a less height in proportion to their distance from the Torrid Zone. In the Alps the snow-line in summer occurs at a height of 8,800 feet above the sea. Near the North and South Poles it comes down to the level of the sea. In regions of perpetual snow even the hottest summer's sun fails to melt all the snow that falls in a year, hence there is a continual accumulation of it, and, if there were no way of getting rid of it, all the water in the world would be eventually transferred to these regions and remain there in a solid form. There is, however, a drainage from these snow-covered regions by means of glaciers, just as there is a drainage. from other regions by means of brooks and rivers. The lower part of all perpetual snow is being constantly converted into ice; the weight of the superincumbent mass forces out the air from between the snow-flakes and binds them together, just as a snowball is made harder by the pressure of the hand. This firm, granular kind of snow is known as névé or firn, and as it is pushed down the mountain slopes it passes into a kind of soft ice, partly in consequence of the pressure from above, and partly from the alternate melting and freezing of the surface layers: the

sun melts the surface in the daytime, and the water so formed percolates down into the mass and becomes frozen again below.

The ice thus formed is gradually pushed down into the valleys, along which it continues to travel in the form of ice-rivers or glaciers. The glaciers of the Alps are sometimes as much as 600 feet in depth, and they fill the valleys for many miles below the snow-line, till they come down to a level of about 3,500 feet below it, where they terminate, in consequence of the ice melting into water and proceeding as a brook or a river. In the New Zealand Alps, where the summer snow-line is at about 5,000 feet, the glaciers on the Mount Cook range, near lat. 44, descend to about 2,300 feet on the eastern side, and one on the western side comes down to within 670 feet of the sea, discharging its load of débris in the midst of a sub-tropical vegetation.' In high latitudes, where the snow-line is near the sealevel, the ice is pushed into the sea, and the portions which break off from time to time are floated off as icebergs.

Movement of Glaciers.-Glaciers creep along their beds at a rate which varies with the supply of ice, the slope of the valley, and the season of the year, but their movement is seldom so rapid as to be perceptible without careful measurement. Like a river, the ice moves faster in the centre than near the sides, and the surface layers move faster than the lower portions, and the whole moves as if it were a plastic or viscous body. The lower parts of the Swiss glaciers, where the gradient is not steep, only move at the rate of one or two feet in 24 hours; thus the Mer de Glace in summer moves at an average rate of 27 inches a day in the centre, and from 13 to 19 inches a day near the sides; in the winter the rates are only about half as much. Where slopes are very steep ice-falls are formed, the ice descending in huge steps or slices, with long fissures or crevasses between, but on reaching a gentler slope the fissures close and are frozen together again. Such ice-falls are frequent in Norway.

Where there are large snow-fields, and the glaciers pass

1 Green's "High Alps of New Zealand," 1883, p. 75.

down fairly steep slopes, their movement is much faster than any of the Swiss glaciers. Thus at Glacier Bay, in Alaska, nine large and seventeen small glaciers unite to form an ice-stream which has a front of 5,000 feet in width and a depth of 700 feet where it enters the bay: this stream moves during August at the rate of 70 feet a day in the middle, and 10 feet near the sides.

Transport by Glaciers.-We have seen that, in the case of rivers, two processes are going on at the same time, namely, erosion and transportation; the rivers of ice perform the same kind of work, but in a different way. Wherever a glacier passes beneath a crag or cliff of rock it receives on to its surface a continued supply of blocks and stones which are detached by the action of frost and weathering, described on p. 102. A line of such rock-fragments may generally be seen at each side of the glacier. Where two glacier-streams unite the two lines of transported stones on their adjacent sides come together and form a double median line along the centre of the glacier below the junction of the two valleys. Thus, if a glacier happen to have many important tributaries proceeding from different valleys, and uniting to form one large icestream, this may come to have many separate lines of stones and rubbish, which are all carried along to the place where the glacier terminates. Here they are all thrown down, and form an irregular mound or pile of rough, angular débris mixed with earth, which in Switzerland is called a moraine. This term is also applied to the lines of stones on the glacier, so that there are three kinds of moraines, lateral, medial, and terminal. (See fig. 38.)

Very large blocks of rock are sometimes transported by glaciers to great distances from their parent sources, and in those countries where the glaciers were once larger and longer than they are now, such blocks are found scattered over the area formerly covered by the ice; they are often perched on eminences, or left stranded on the sides of a valley. Rock-masses found in such positions are called perched blocks or erratics. (See fig. 39.)

Some of the clearest and best-known instances of the transport of such blocks are found in Switzerland. There was a time, generally known as the Glacial Period, when