1876 PENKNIFE ICE. 69 



are still left unmelted, and the ice has not ceased drain- 

 ing, this measurement will probably be increased. Sir 

 Edward Parry, in 1827, met with some fourteen inches 

 in length on the 12th of July, and eighteen inches in 

 length on the 16th of July. 



'While the formation of "penknife ice" is thus 

 very apparent, a somew^hat similar formation is taking 

 place as the snow decays by reflected heat. 



' Early in the spring, wherever the stratification of 

 the snow covering a floe had become exposed at a 

 newly formed crack, the lower portion of the snow was 

 observed to have granulated, the grains collecting 

 together perpendicularly, the lower ones being the 

 largest and leaving intermediate air-spaces ; the whole 

 structure giving promise that during the summer it 

 would assume the columnar appearance like the so- 

 called " penknife ice," which the surface of many of the 

 Polar floes showed had been formed during a previous 

 season.' 



While the surface of the floes usually consisted of 

 slightly brackish compact ice, in many cases we found 

 it composed of vertical columns of brackish ice half an 

 inch in diameter and about twelve inches in height, 

 rising from a foundation of solid ice, and having light 

 snow intermixed with them ; these were supported at 

 the top by a thin horizontal network of ice, and the 

 whole covered with the usual layer of snow, varying in 

 thickness according to the locality. 



' In a few cases we observed a double set of such 

 inverted icicle-like columns, one above the other, 

 divided by a horizontal layer of clear ice about four 

 inches in thickness, and containing air-drops. In the 



