There are, then, in the upper snow-fields three kinds of horizontal deposits: the beds of snow, the sheets of dust, and the layers of ice, alternating with each other. If, now, there were no modifying circumstances to change the outline and surface of the glacier,--if it moved on uninterruptedly through an open valley, the lower layers, forming the mass, getting by degrees the advance of the upper ones, our problem would be simple enough. We should then have a longitudinal mass of snow, inclosed between rocky walls, its surface crossed by straight transverse lines marking the annual additions to the glacier, as in the adjoining figure. [Illustration] But that mass of snow, before it reaches the outlet of the valley, is to be compressed, contorted, folded, rent in a thousand directions. The beds of snow, which in the upper ranges of the mountain were spread out over broad, open surfaces, are to be crowded into comparatively circumscribed valleys, to force and press themselves through narrow passes, alternately melting and freezing, till they pass from the condition of snow into that of ice, to undergo, in short, constant transformations, by which the primitive stratification will be extensively modified. In the first place, the more rapid motion of the centre of the glacier, as compared with the margins, will draw the lines of stratification downward toward the middle faster than at the sides. Accurate measurements have shown that the axis of a glacier may move ten- or twenty-fold more rapidly than its margins. This is not the place to introduce a detailed account of the experiments made to ascertain this result; but I would refer those who are interested in the matter to the measurements given in my "Systeme Glaciaire," where it will be seen that the middle may move at a rate of two hundred feet a year, while the margins may not advance more than ten or fifteen or twenty feet. These observations of mine have the advantage over those of other observers, that, while they embrace the whole extent of the glacier, transversely as well as in its length, they cover a period of several successive years, instead of being limited to summer campaigns and a few winter observations. The consequence of this mode of progressing will be that the straight lines drawn transversely across the surface of the glacier above will be gradually changed to curved ones below. After a few years, such a line will appear on the surface of the glacier like a crescent, with the bow turned downward, within which, above, are other crescents, less and less sharply arched up to the last year's line, which may be again straight across the snow-field. (See the subjoined figure, which represents a part of the glacier of the Lauter-Aar.) [Illustration] Thus the glacier records upon its surface its annual growth and progress, and registers also the inequality in the rate of advance between the axis and the sides.