512 PROCEEDINGS OF THE AMERICAN ACADEMY 



greatest density, the circulation from above downward, and the reverse, 

 ceases. This takes place at a temperature of some 39.2 degrees Fahr. 

 Thenceforward further reduction of temperature causes the water to 

 congeal and expand at the surface. As still further reduction of tem- 

 perature at the surface goes on, the thickness of the ice increases from 

 below, and with this cooling comes a contraction of the superior por- 

 tions of the ice. This contraction, like that of lava shrinking from loss 

 of heat, or like that of large blocks of moist starch shrinking from loss 

 of water, is accompanied by separation of the mass into slender columns 

 by thin blades of space. In the case of ice, thin blades extend from 

 above downward, and divide the ice into a kind of cellular structure. 

 This structure in the ice, though actual, may be inappreciable to the 

 ' eye, except in cases where the cells are sufficiently pronounced to be- 

 come visible through the filaments or blades of air which fill the spaces 

 left by the shrinking ice. The ■ cracks, by which I mean the visible 

 openings in the ice, which are readily commensurable, being frequently 

 a considerable fraction of an inch across, I have observed in many, 

 perhaps in most instances, are not readily traced to the bottom of the 

 ice, but are wedge-shaped and broadest at the top. This is not in- 

 variable, however. Under some conditions I have seen the opposite 

 sides of the cracks displaced, and the whole neighboring surface, stretch- 

 ing away from the crack, slightly curved downward, while the walls of 

 the crack are lifted. This conformation would naturally follow the re- 

 duction of temperature and shrinking from above. At the bottom the 

 ice must of course remain at 32°. In areas of ice which have been 

 thus subjected to contraction producing visible cracks, the openings fill 

 with water from below, which freezes, and marks by its somewhat con- 

 trasted clearness the site of the cracks. I have noticed in these cracks 

 a peculiar transverse columnar structure, the explanation of which does 

 not seem difficult, as will presently appear. 



When, to a period of low temperature, and the production of cracks 

 and their filling with water and freezing, there succeeds a period of 

 warm weather, the whole body of the ice expands. Now r , under the 

 pressure consequent on this expansion, if the temperature rises to 32°, 

 the space of the cracks filled with water must be provided for ; and as 

 the only relief for this is by increasing the thickness of the ice, or, in 

 other words, .in an elongation of the rods of ice in the honeycomb 

 frame, there must be produced a slight vertical molecular movement 

 which still further individualizes the columns. Thus contraction and 



