"Let _A C_ be one side of the glacier and _B D_ the other; and let the direction of motion be that indicated by the arrow. Let _S T_ be a transverse slice of the glacier, taken straight across it, say to-day. A few days or weeks hence the slice will have been carried down, and because the center moves more quickly than the sides it will not remain straight, but will bend into the form _S' T'_. Supposing _T i_ to be a small square of the original slice near the side of the glacier; in the new position the square will be distorted to the lozenge-shaped figure _T' i'_. Fix your attention upon the {128} diagonal _T i_ of the square; in the lower position this diagonal, _if the ice could stretch_, would be lengthened to _T' i'_. But the ice does not stretch; it breaks, and we have a crevasse formed at right angles to _T' i'_. The mere inspection of the diagram will assure you that the crevasse will point obliquely _upward_."[29] [Illustration: FIGURE 4--_Diagram illustrating the Formation of marginal Crevasses_.] [Footnote 29: The Forms of Water: International Scientific Series, New York, 1875, pp. 107-108.] The explanation given above applies especially to the lower or icy portion of a glacier; above the snow-line other facts appear. When a glacier flows through fields of snow on a level with its surface, crevasses are formed in the adjacent banks. These trend down stream for the same reason that the crevasses in the glacier proper trend up stream--that is, the friction of the moving stream against its banks tends to carry them along, while the portions at a distance are stationary. Fissures are thus opened which trend in the direction in which the glacier moves. The angle made by these crevasses with the axis of the glacier is about the same as those of the marginal crevasses, but in an opposite direction. They are widest near the margin of the glacier and taper to a sharp end towards the stationary snow-banks above. The crevasses in the two series thus fall nearly in line, but are separated by a narrow band of irregularly broken snow, marking the actual border of the glacier.[30] [Footnote 30: Crevasses in snow-fields through which ice-streams flow will be mentioned again in describing the Seward glacier.] After leaving Blossom island the party was divided, and we began a new series of numbers for our camp above the snow-line, although in this narrative and on the accompanying map a single series of numbers for all the camps will be used. While in the field the camps in the snow were usually termed, facetiously, "sardine camps," in allusion to the uncomfortable manner in which we were packed in our tent at night. {129} ACROSS PINNACLE PASS.