72 CRYSTALLISATION OF WATER. 



" 111 my greenhouse there was au aquarium, some 5ft. long, with 

 about Gill, deep of water in it. 



" The doors being closed, the temperature was prevented from 

 falling below 32° during the night, but early on the morning of 

 Christmas-day, as I entered, the cold air rushed in ; and my attention 

 was first called to the intensity of the frost by the fact that some water, 

 accidentally spilt, was frozen solid before I had time to walk the length 

 of the greenhouse. 



" On looking into the aquarium to see how the fish stood the cold, my 

 attention was attracted by some little stars rising from the sides and 

 bottom, and floating up steadily to the surface of the water. They 

 sometimes came up singly, and sometimes in clusters of two or three 

 or more. 



" They at first formed little islands of ice on the surface, by joining 

 together at their points ; others rising under them soon filled up the 

 spaces, and formed thin sheets of fragments of ice, with rough, jagged 

 edges. 



" In a short time the entire surface was covered with ice, and in the 

 course of a few hours the aquarium was frozen solid." 



Before concluding, I must refer to two other phenomena, which any- 

 one, who tries to reproduce the ice-flowers, will probably meet with. 

 Sometimes, instead of liquid flowers, we get only liquid discs, that is, 

 extremely thin layers of water, which are unattended by a " bubble." 

 I have seen these discs hexagonal, instead of circular, (see Figs. 5 

 and 6,) presenting a close resemblance to the hexagonal plate snow- 

 crystal which is frequently seen. Their average diameter is the same 

 as that of the flowers, and they also lie in the plane of freezing. If, 

 instead of the ice being perfectly clear, it contains bubbles of air, then, 

 on exposure to heat, we shall find the portion of ice immediately 

 surrounding the bubble to melt. Thus we obtain composite cavities, 

 as drawn in Fig. 7, where the central bubble represents the air, and 

 the clear area surrounding it the water. The cavity occupied by the 

 water does not, in this case, generally assume the form of au ice- 

 crystal. It is usually rounded, but sometimes slightly crimped at the 

 edges. Since the ice in melting shrinks to a smaller volume, it follows 

 that the air composir-g the bubble must now be rarefied ; and this is 

 confirmed by observing that when the ice surrounding it is melted, and 

 the bubble set free, it invariably collapses to a much smaller size. 



The question arises, why should the ice melt immediately around 

 the bubble in preference to any other part? The answer to this 

 question depends upon an interesting property of the particles of 

 matter. We know that a liquid in its ordinary condition is continually 

 giving off vapour from its surface ; but why from its surface only ? 

 Simply because the molecules there have greater freedom of action, 

 are less hampered by the surrounding molecules than those within the 

 bodv. So, in a mass of ice, the particles on the surface yield to the 

 influeuce of the heat sooner than those within the mass. But it is 



