Limiting our view to one of the tributaries only, the result is still more impressive. Previous to its junction with the Talefre, the Glacier de Lechaud stretches before the observer as a broad river of ice, measuring 825 yards across: at Trelaporte it is squeezed, in a frozen vice, between the Talefre on one side and the Geant on the other, to a driblet, measuring 85 yards in width, or about one-tenth of its former transverse dimension. It will of course be understood that it is the _form_ and not the _volume_ of the glacier that is affected to this enormous extent by the pressure. Supposing no waste took place, the Glacier de Lechaud would force precisely the same amount of ice through the "narrows" at Trelaporte, in one day, as it sends past the Pierre de Beranger. At the latter place its velocity is about half of what it is at the former, but its width is more than nine times as great. Hence, if no waste took place, its _depth_, at Trelaporte, would be at _least_ 4-1/2 times its depth opposite the Pierre de Beranger. Superficial and subglacial melting greatly modify this result. Still I think it extremely probable that observations directed to this end would prove the comparative shallowness of the upper portions of the Glacier de Lechaud. FOOTNOTES: [A] Great care is necessary on the part of the man who measures the displacements. The staff ought to be placed along the original line, and the assistant ought to walk along it until the foot of a _perpendicular_ from the stake is attained. When several days' motion is to be measured, this precaution is absolutely necessary; the eye being liable to be grossly deceived in _guessing_ the direction of a perpendicular. [B] The details of the measurement of the fourth and fifth lines are published in the 'Philosophical Transactions,' vol. cxlix., p. 261. ICE-WALL AT THE TACUL. VELOCITIES OF TOP AND BOTTOM. (11.) As regards the motion of the _surface_ of a glacier, two laws are to be borne in mind: 1st, that regarding the quicker movement of the centre; 2nd, that regarding the locus of the point of maximum motion. Our next care must be to compare the motion of the surface of a glacier with the motion of those parts which lie near its bed. Rendu first surmised that the bottom of the glacier was retarded by friction, and both Professor Forbes[A] and M. Martins[B] have confirmed the conjecture. Theirs are the only observations which we possess upon the subject; and I was particularly desirous to instruct myself upon this important head by measurements of my own. [Sidenote: FIRST ATTEMPT AT MEASUREMENT.]