400 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
degree Cent, by the pressure. Here in the lower levels of the ice 
sheet is the true zone of flow, passing gradually upward into the 
colder zone of fracture, which normally terminates upward in the 
in which the fractures are promptly healed by the freezing of 
infiltrating water, by settling, and by fresh snow fall. These con¬ 
siderations clearly suggest a tendency, at least, to a reversal of the 
law governing the vertical distribution of velocity; and the tendency 
of the upper layers of ice to bear down or depress the obliquely ris¬ 
ing englacial drift will be neutralized to the extent or degree of the 
reversal. It is even conceivable that the distribution of velocity 
may favor or accelerate the rise of drift, and the writer feels that, 
in view of these arguments, the transfer of drift in large volume to 
a considerable height in the ice, and over a wide area, may be 
asserted with renewed confidence. Granted a waning ice sheet free 
from crevasses and well supplied with superglacial drift over a broad 
marginal zone, and the conditions are ripe for the development, 
through the agency of superglacial streams, of eskers and esker 
systems or osars equal in length and continuity to any which have 
been described. 
The difficulties of the subglacial hypothesis are here relatively 
much more serious, since it requires us to postulate and maintain 
continuous ice tunnels from five miles or less to one hundred and 
fifty miles or more in length. In fact, the formation, either before 
or after the ice has ceased to move, of subglacial streams and tunnels 
from one hundred to one hundred and fifty miles long, extending 
back to points where the thickness and pressure of the ice must be 
very great and crevasses are practically impossible, can be accepted 
only as a last resort, or when the failure of the alternative explana¬ 
tion has been demonstrated. 
Varying loidth of eskers .— The expansion of superglacial streams 
to form lakes of greater or less breadth is normal, and the most 
extreme variations in the breadth of eskers present absolutely no 
difficulty, while, as the product of subglacial streams, they are 
simply inexplicable, requiring, according to Stone, tunnels of all 
widths up to three fourths of a mile. The broad eskers described by 
Stone (’ 99 , p. 440-444), in which a medial ridge of coarse gravel is 
flanked on either side by, and merges with, a plain of finer gravel and 
sand, the whole being often confined to one side of a valley by an 
ice border, and, like true eskers, crossing ridges from one valley to 
