514 PROCEEDINGS OF THE AMERICAN ACADEMY 



served in the cooling of large cylindrical iron castings. From such 

 cavities I have plucked out detached octahedral skeleton crystals of 

 iron. 



The phenomena of ice structure vary so greatly with the circum- 

 stances under which congelation takes place, that they seem almost 

 infinite. Large areas of ice seem to me under certain conditions to be 

 produced in considerable degree from what might be regarded as a 

 snow-storm from below — the whole body of ice seeming to be made up 

 of snow-flakes — of marvellous beauty. I have seen a snow-flake in the 

 interior of a block of ice not less than two inches in horizontal diame- 

 ter, and perfect in all its parts. 



Tyndall figures a block full of these snow-flakes revealed by direct- 

 ing a concentrated sunbeam successively on different points in the in- 

 terior of a block. In others I have found numerous fern-leaf crystals 

 (segments of the great snow-flake) in vertical position, and several of 

 my casts of decaying ice show the pseudomorphs of fern leaves which 

 have dissolved out therefrom. 



I have found in little cavities, commonly known as air-holes, the 

 under surface of the thin film of ice roofing over the air space and its 

 floor below, minute six-sided tabular crystals of dimensions great enough 

 to be readily recognized with the unaided eye. 



It is not the purpose of this note to embrace a detailed account of the 

 observations I have made upon the phenomena of ice, but is mainly, as 

 I have said, to present the conception I entertain of the cause of the 

 columnar structure of decaying ice. 



I conceive it to be due primarily to the alternate expansion and con- 

 traction of the ice, consequent upon cycles of temperature, resulting 

 first in minute fissures from the shrinking, induced by cold, which, fill- 

 ing with air, yield a sort of honeycomb structure ; and, secondarily, to 

 the cellular structure produced by the intersection of vertical or nearly 

 vertical blades of ice crystals, which protruded from the continuous ice 

 above into the water below, as the thickness of the ice increased, and 

 which vertical blades determined the surfaces where the sun's rays at 

 the time of reflection are resolved into dark heat ; and, thirdly, the 

 elongation of the columns enclosed by the honeycomb structure conse- 

 quent on the pressure caused by expansion. 



The second phenomenon is like that afforded by subjecting blocks of 

 wax to pressure. The aggregations of imperfect crystals in the ease of 

 wax are flattened out, the surface planes or planes of separation of the 



