CROSBY : ORIGIN OF ESKERS. 
393 
face of the ice. A dry stone, if thin enough, transmits heat directly 
to the ice; but with a submerged stone the transmission is through 
-the medium of the water. Hence as soon as the thin stone or 
deposit becomes submerged, the thinness ceases to be an essential 
factor, and it is on a par with the thick stone or dejjosit. The nor¬ 
mal temperature of the water is, of course, that of melting ice, or 
four degrees Cent, below its temperature of maximum density. The 
water absorbs some of the solar radiation directly, and it takes up 
promptly and completely the much larger amount of heat absorbed 
by the submerged or partially submerged detritus. As fast as the 
water gains in temperature it sinks to the bottom, displacing colder 
and lighter water, and expends its surplus heat in melting the ice. 
As has been noted by others, this principle explains the fact that 
the ice shores of the Malaspina lakelets are undercut below the 
water level, the ice, as stated by Russell, melting below the surface 
more rapidly than above, Avhere it is exposed to the direct rays of 
the sun. The thickness of the deposit is no appreciable bar to the 
process, so long as the material is sensibly permeable, but on the 
contrary it enables the water warmed by contact with its surface 
to sink promptly and quickly to the underlying ice; and with 
increasing thickness the pressure cooperates by lowering the melt¬ 
ing point, while with such torrential streams as must have been 
those in which normal eskers were formed, the conversion of 
mechanical energy into heat may not be neglected. A bed of per¬ 
meable gravel not only cooperates in the ways indicated, to favor the 
melting of its ice floor, but it must tend to conserve the heat and to 
prolong the time during which melting can take place. 
In view of these considerations, there seems to be no escape from 
the conclusion that the ice floor of a superglacial stream will be 
lowered by the superficial as well as by the basal melting of the ice ; 
that the superficial melting will be more active and efficient in pro¬ 
portion to the extent of aggrading of the channel and the volume 
of stagnant water saturating the gravel; and that the stream will 
be lowered at least as rapidly as the interstream surfaces, because, 
while these are also, according to our initial assumption, covered 
with drift and thus protected from the direct action of the sun’s rays, 
they lack the standing water essential to the effective indirect utiliza¬ 
tion of the solar radiation. 
Obviously the aggrading of the channel may continue during all 
