CLOUDS AND EIVEKS, ICE AND GLACIERS. 73 



ice-river we see is not straight but curved, and its cur- 

 vature is from the Montanvert ; that is to say, its con- 

 vex side is east, and its concave side is west (look to 

 the sketch). You have already pondered the fact that 

 a glacier, in some respects, moves like a river. How would 

 a river move through a curved channel ? This is known. 

 Were the ice of the Mer de Glace displaced by water, 

 the point of swiftest motion at the Montanvert would 

 not be the centre, but a point east of the centre. Can 

 it be then that this ' water-rock, 5 as ice is sometimes 

 called, acts in this respect also like water ? 



175. This is a thought suggested on the spot; it may 

 or it may not be true, but the means of testing it are at 

 hand. Looking up the glacier, we see that at les Ponts 

 it also bends, but that there its convex curvature is 

 towards the western side of the valley (look again to 

 the sketch). If our surmise be true, the point of 

 swiftest motion opposite les Ponts ought to lie west of 

 the axis of the glacier. 



176. Let us test this conjecture. On July 25 we fix 

 in a line across this portion of the glacier seventeen 

 stakes ; every one of them has remained firm, and on 

 the 26th we find the motion for 24 hours to be as 

 follows : 



Fourth Line : D D upon the Sketch. 



East "West 



Stake .12 3 4 5 6 7 8 9 10 11 3.2 13 14 15 



Inches . 7 8 13 15 16 19 20 2i 21 23 23 21 22 17 15 



