CRYSTALLISATION OP WATER. 73 



obvions that the particles in immediate contact with an enclosed 

 bubble are, in this respect, in circumstances similar to those on the 

 exterior of the ice. The heat of a body is at the present day attributed 

 to a motion of the molecules composing that body. When the motion 

 of the particles of a solid reaches a certain limit, the solid becomes a 

 liquid; when the motion is further increased, the liquid becomes a gas. 

 And we can easily see that a molecular motion, which is incapable of 

 liberating the particles of a solid mass, may be propagated through 

 ♦•hem without prejudice to its solidity ; yet, when this same motion 

 reaches the particles bounding a cavity, it maj' suffice to liberate 

 them. Professor Tyndall, from whom I have derived this explanation, 

 made a number of experiments, which entirely confirm it ; and those 

 who have studied his theory of glacier motion know that the same idea 

 has been applied to explain the effect of regelatiou upon which the 

 theory is based. 



There is one other point of view in which the ice-flowers teach us 

 a lesson. Why should one portion of a solid block of ice melt sooner 

 than another? The heat which liquefies these flowers must have 

 passed through the surrounding ice before it reached them. We con- 

 clude, to quote Professor Tradall's words, that " the absorption is 

 fitful, and not continuous ; and there is no reason to suppose that in 

 other solids the case is not the same, though their constitution may 

 not be such as to reveal it. There is no such thing as absolute 

 homogeneity in nature. Change commences at distinct centres, instead 

 of being uniformly distributed ; and in the most apparently homo- 

 geneous substance we should discover defects, if our means of 

 obseiwation were fine enough." 



