PHYSICAL PARADOXES. 



time to do it is when a thaw wind has brought 

 the snow back to zero C. 



Similarly with the formation and flow of 

 glaciers. The crystals beneath superincumbent 

 masses of snow, getting warmed by contact 

 with or proximity to the ground, are at last not 

 too far below freezing-point for those under 

 the heaviest strains to be partly melted by the 

 pressure. As their yielding and taking up 

 positions of less strain diminishes the pressure, 

 the surrounding cold, at or slightly below zero, 

 causes re-freezing or regelation. 



So we get masses of ice beneath the surface, 

 and movement and flow of these masses much 

 as if they were a slow stream. This is visible 

 in the glacier shown in Plate IV. 



4. Ice unmelted in a Vessel of Boiling Water. 



If a test-tube containing water be held in 

 the flame of a spirit lamp or of a Bunsen burner, 

 the water will in due course be heated to 

 boiling-point, with the production of steam ; 

 but if it be so held that the flame is applied 

 to the middle of the tube, as in Fig. 29, only 

 the upper part of the water will be heated, the 

 lower part remaining so cool that it is easily 

 held by the hand while the boiling proceeds. 

 The reason is that the heat of the flame causing 

 the water to expand (the usual effect of heat) 

 at the point where it is applied, makes it there- 



126 



