Origin of Eskers. — Crosby. 19 
which it is tributary, as long- as the floor of the ice-gorge is 
nowhere below the base level, or the level of the frontal bar- 
rier. Down to this level the gorge has been formed by the 
mechanical erosion or corrasion as well as by the chemical 
erosion or melting of the ice ; and below this level corrasion is 
surely inoperative. But here Upham's suggestion of a farther 
deepening of the gorge by melting alone intervenes, and I hope 
to show that it will continue with increased rather than dimin- 
ished efficiency. 
Basal melting of the icesheet must be in constant prog- 
ress, summer and winter, during this state and is possibly 
an important factor in letting down and indirectly deepening 
the deposits in the superglacial channels ; while the even dis- 
tribution of the effects due to this cause may safely be as- 
sumed. But the main question now is, will the superglacial 
channel persist, or retain its walls, until its bottom reaches the 
ground? In other words, will the melting of its floor beneath 
its increasing load of detritus keep pace with, or at least keep 
ahead of, the general ablation of the interstream surfaces? The 
familiar instances of medial moraines resting on ridges of ice 
and isolated bowlders perched on pillars of ice, through the 
more rapid melting of the drift-free surface, seem to demand a 
negative answer. On the other hand, the innumerable well- 
like holes in the ice, noted by many observers, and varying 
from an inch to several feet in depth, and often containing 
nothing but the purest of water, although commonly a stone 
or a little sand, clay, or cosmic dust is seen resting on the bot- 
tom, point to an affirmative answer. These holes are usually 
explained, however, as dependent upon the absorption of the 
solar heat by a slight thickness of stony or earthy matter ; 
whereas thicker masses protect the ice beneath from the solar 
radiation. The occasional absence of foreign matter indicates 
that this explanation is incomplete ; and the lakelets described 
by Russell* as a common and characteristic feature of the 
moraine-covered marginal zone of the Malaspina glacier clearl} 
demand a different explanation from that usually accepted for 
the relatively small holes ; for the lakelets, which may be fifty 
to one hundred feet or more in depth, and are rarely more than 
one hundred feet in diameter, appearing to differ from the holes 
• Nat. Geol. Mag., vol. iii, p. 120 
