CROSBY : ORIGIN OF ESKERS. 
395 
and the very numerous instances of plains of all the various types 
without associated eskers or feeder channels are simply the cases 
where the superficial stream was diverted before its channel was 
base-leveled, and consequently before it was aggraded to any 
important extent. Finally, the lateral position characteristic of 
valley eskers follows naturally from the tendency of the ice, before 
it ceased fiowing;, to become concentrated in the valleys or along the 
lines of freest movement, giving rise to what were virtually valley 
glaciers with arched profiles in what may still have been at the sur¬ 
face a continuous ice sheet, and possibly with lines of shearing 
between the free-moving ice and the relatively stagnant ice of the 
uplands. When the ice becomes wholly stagnant, these lateral lines 
of weakness and the arched profile still remain to influence the 
courses of superglacial streams. 
Subglacial Hypothesis. 
As formulated by Davis (’92), this explanation of eskers also 
presumes a stagnant and decayed marginal zone of the ice sheet. 
The water resulting from the basal melting of the ice, together with 
that due to superficial melting which becomes subglacial through 
the medium of crevasses, is gathered into streams which, it is stated, 
hold to meridional courses or, more exactly, conform with the gen¬ 
eral trend of the ice movement, in obedience to a direct control 
exerted by the ice. This control being sufficient to force the streams 
to flow uphill and over elevations of 100 to 400 feet, we must, 
apparently, suppose that their courses were determined before the 
motion of the ice ceased and while it wa-s still thick enough to hold 
to its normal trend regardless even of quite strongly accentuated 
topographic features. How the subglacial channel or tunnel origi¬ 
nated is not considered, beyond the suggestion that it may be due to 
the enlargement of longitudinal crevasses ; but the point is not vital, 
for it is obvious that the subglacial water must escape and that, no 
matter how closely pent, there must always be lines of least resist¬ 
ance to its flow. 
At and beyond the margin of the ice, the phenomena must be 
essentially the same as for superglacial streams, and a frontal barrier 
or sill of any kind causing the stream to rise, perhaps fountain-like. 
