The M(ir(iin of the Cornell Glacier. 153 
is seen the reason for the formation of roches raoutonees 
forms; some areas are scoured more than otliers, and there is 
a gradual reduction to a generalized surface because the de- 
pressions are protected b}-^ more nearly stagnant ice. while the 
projections are worn down toward the curve of average out- 
line' of tlie irregularity. 
Vo.riations in the Direction of Ice Movement. The ice 
sheet, burying all land so that no sign of it is found on the 
ice top, as is the case in the interior of Greenland, must move 
witli moderate directness and uniformity. Perhaps there is 
some deflection b}- topographic irregularities of great size, 
but sucli variation would be the same in direction so long as 
the thickness of the ice remains constant. For instance, if a 
part of the present irregular border of Greenland were over- 
ridden by the ice sheet the general direction wc^uld be west- 
ward, notwithstanding peninsulas and mountain peaks, which, 
however, might deflect the ice slightly from the general di- 
rection. Any considerable deflection could not be caused, for 
the land irregularities are less in importance than the force 
which is pushing the ice onward. 
When such a land is being abandoned by the ice sheet, 
quite different conditions are introduced. At first the moun- 
tain peaks begin to appear as nunataks, first, merely checking 
the ice movement and making it nearly stagnant on top, while 
the glacier moves around the sides, deflected possibly many 
degrees from the original course. Later, by rising above the 
ice, and dividing it into two streams, which may possibly 
meet on the downstream side, or if the terminus ends in the 
sea failing to join, a similar deflection is caused. Where it 
comes in contact with the land there may be decided deflec- 
tion of the ice currents, possibly toward a valley tongue, as 
frequently is the case with the ice near the margin of the Cor- 
nell glacier. Near the margin there may be a very decided de- 
flection of currents, even in neighboring parts of the glacier, 
and at the exact terminus the deflection m-Aj be even greater. 
A tongue of ice, led away by some favorably situated val- 
ley, will go in whatever direction the valley leads; and one 
of the most astonishing results that I obtained from my study 
of the ice margin, was the discovery of the remarkable man- 
ner in which the ice is thus deflected. Even a tiny depression 
