The appearance of crevasses is often determined by circumstances more local and limited than those above indicated; a boss of rock, a protuberance on the side of the flanking mountain, anything, in short, which checks the motion of one part of the ice and permits an adjacent portion to be pushed away from it, produces crevasses. Some valleys are terminated by a kind of mountain-circus with steep sides, against which the snow rises to a considerable height. As the mass is urged downwards, the lower portion of the snow-slope is often torn away from its higher portion, and a chasm is formed, which usually extends round the head of the valley. To such a crevasse the specific name _Bergschrund_ is applied in the Bernese Alps; I have referred to one of them in the account of the "Passage of the Strahleck." (18.) The phenomena described and accounted for in the last chapter have a direct bearing upon the question of viscosity. In virtue of the quicker central flow the lateral ice is subject to an oblique strain; but, instead of stretching, it breaks, and marginal crevasses are formed. We also see that a slight curvature in the valley, by throwing an additional strain upon one half of the glacier, produces an augmented crevassing of that side. But it is known that a substance confessedly viscous may be broken by a sudden shock or strain. Professor Forbes justly observes that sealing-wax at moderate temperatures will mould itself (with time) to the most delicate inequalities of the surface on which it rests, but may at the same time be shivered to atoms by the blow of a hammer. Hence, in order to estimate the weight of the objection that a glacier breaks when subjected to strain, we must know the conditions under which the force is applied. The Mer de Glace has been shown (p. 287) to move through the neck of the valley at Trelaporte at the rate of twenty inches a day. Let the sides of this page represent the boundaries of the glacier at Trelaporte, and any one of its lines of print a transverse slice of ice. Supposing the line to move down the page as the slice of ice moves down the valley, then the bending of the ice in twenty-four hours, shown on such a scale, would only be sufficient to push forward the centre in advance of the sides by a very small fraction of the width of the line of print. To such an extremely gradual strain the ice is unable to accommodate itself without fracture. [Sidenote: NUMERICAL TEST OF VISCOSITY.]