Let A B, C D, be the bounding sides of a glacier moving in the direction of the arrow; let _m_, _n_ be two points upon the ice, one, _m_, close to the retarding side of the valley, and the other, _n_, at some distance from it. After a certain time, the point _m_ will have moved downwards to _m'_, but in consequence of the swifter movement of the parts at a distance from the sides, _n_ will have moved in the same time to _n'_. Thus the line _m n_, instead of being at right angles to the glacier, takes up the oblique position _m' n'_; but to reach from _m'_ to _n'_ the line _m n_ would have to stretch itself considerably; every other line that we can draw upon the ice parallel to _m' n'_ is in a similar state of tension; or, in other words, the sides of the glacier are acted upon by an oblique pull towards the centre. Now, Mr. Hopkins has shown that the direction in which this oblique pull is strongest encloses an angle of 45 deg. with the side of the glacier. [Sidenote: LINE OF GREATEST STRAIN.] [Illustration: Fig. 26. Diagram showing the line of Greatest Strain.] What is the consequence of this? Let A B, C D, Fig. 26, represent, as before, the sides of the glacier, moving in the direction of the arrow; let the shading lines enclose an angle of 45 deg. with the sides. _Along_ these lines the marginal ice suffers the greatest strain, and, consequently _across_ these lines and at right angles to them, the ice tends to break and to form _marginal crevasses_. The lines, _o p_, _o p_, mark the direction of these crevasses; they are at right angles to the line of greatest strain, and hence also enclose an angle of 45 deg. with the side of the valley, _being obliquely pointed upwards_. [Sidenote: MARGINAL AND TRANSVERSE CREVASSES.] This latter result is noteworthy; it follows from the mechanical data that the swifter motion of the centre tends to produce marginal crevasses which are inclined from the side of the glacier towards its source, and not towards its lower extremity. But when we look down upon a glacier thus crevassed, the first impression is that the sides have been dragged down, and have left the central portions behind them; indeed, it was this very appearance that led M. de Charpentier and M. Agassiz into the error of supposing that the sides of a glacier moved more quickly than its middle portions; and it was also the delusive aspect of the crevasses which led Professor Forbes to infer the slower motion of the eastern side of the Mer de Glace.