VARIED STRUCTURE OF FLOES. 393 
that, in addition to the actions of simple infiltration, 
forces allied to endosmosis are called into play. I 
ohserved, during the month of May and in the last 
of April, that the surface snows, heated hy the sun, 
formed pools in the most dependent portions of our 
ice. Where this occurred in ices formed either early 
or late in the season, and which presented, therefore, 
the prismatic arrangement I have described, the work 
of destruction went on with wonderful rapidity. Al- 
though our mean temperature was greatly below the 
freezing point, and these little pools were themselves 
coated nightly with a pellicle of ice, a very few days 
would render them unsafe. A boat-hook could be 
thrust nearly through them, and they were even dan- 
gerous for pedestrians. We had thus all the indica- 
tions, except that of a membraneous interspace, which 
might invite endosmotic action; fluids of different dens- 
ities above and below, and an intermediate structure 
abounding in capillary ducts. 
The presenting face of the hummocks, approaching 
more or less nearly to the vertical, opposes them to the 
direct instead of the oblique rays of the sun ; and their 
sides begin to thaw in consequence, while the more 
horizontal floes remain unchanged : and as these hum- 
mock ridges represent the lines of previous cementa- 
tion, they are soon prepared to become those of first 
separation. Floes break up most readily along such 
lines of old cementation. 
Before passing from these causes of disintegration 
and destruction in the pack, I would refer again to the 
fact which I have mentioned already of its being a 
great mosaic work, composed of tables of various thick- 
nesses, and, of course, of varying resistance. Such 
ice, therefore, when subjected to mechanical pressure, 
