398 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
derings transverse to the direction of ice flow certainly add some 
difficulties to the hypothesis of subglacial streams.^’ This difficulty, 
together with the fact that the formation and maintenance of tun¬ 
nels transverse to the ice movement is well-nigh inconceivable, and 
the further fact that eskers and the plains to which they are tributary 
were clearly formed at a time when the ice was so far wasted as to 
have a very irregular and fragmentary margin, have led Davis and 
other adherents of the subglacial theory to hold with the supergla- 
cialists that eskers were formed, mainly at least, after the ice became 
stagnant. But it is obvious that the control of the subglacial drain¬ 
age demanded by the subglacial theory must originate in the motion 
of the ice, although it may, conceivably, survive the cessation of 
that motion. In other words, the subglacial theory of eskers requires 
us to suppose that the subglacial streams were established before the 
ice ceased to move, but that, as a rule, eskers were formed by these 
streams after the ice became stationary. This nice adjustment of 
conditions made the streams relatively long-lived, and makes it 
appear the more improbable that they promj^tly abandoned their 
courses when the eskers were finished. The meanders of eskers, 
which are certainly among their most constant and striking features, 
rather seldom exhibit any definite or causal relation to the topo¬ 
graphy of the bed rock and till; and since they cannot be correlated 
directly with the movement of the ice, the subglacialist can only 
suggest that they may have been determined by one or more systems 
of crevasses, of the existence of which there is little or no independ¬ 
ent evidence. In view of all these considerations, it is perhaps not 
too ijiuch to say that the superglacial theory affords at once the 
simplest and most natural explanation alike of the general trend of 
eskers, the major deviations from this trend, and the minor devia¬ 
tions or meanders. 
Length of eshers and esker systems. — In the absence of cre¬ 
vasses, superglacial streams are limited in length only by the breadth 
of the zone of ablation, and superglacial eskers only by the breadth 
of the zone of englacial drift which has become superglacial by abla¬ 
tion ; and the last, in turn, depends upon the height to which the 
englacial drift has risen in the ice. We have seen that observa¬ 
tions on existing glaciers are practically valueless as evidence of 
crevassing in the Pleistocene ice sheet; that the Greenland ice cap 
is free from crevasses except near tlie margins, where it breaks over 
) 
