30 
The American Geologist. J"^^'^ i^'^^. 
approximation of the crest to the level of the terminal plain, 
with a gradual rise above that level to the northward if the 
stream was sufficiently long-lived to permit the cbmplete ag- 
grading of its bed. The superglacial hypothesis, on the other 
hand, explains coincidence in hight with the terminal plain, 
but requires it only where aggrading has continued during 
the entire period occupied in letting the esker down upon the 
ground. AX'hen aggradation has ceased earlier than this, the 
final adaptation of the esker to the uneven surface of the 
ground may rise to pretty nearly -all the observed irreg- 
ularities of elevation. The high points or knolls are, how- 
ever, as previously noted, best explained as dne to late depos- 
its contributed by hanging lateral valleys ; and this harmonizes 
well with the fact that they occur usually at a bend in the gla- 
cial river. The subglacial hypothesis, on the other hatid, ap- 
pears to leave the exceptionally high points unexplained. 
Hight of the esker above its base, or elevation above its term- 
inal plain, are not crucial tests; but the superglacial .stream 
appears, on the whole, more competent than the subglacial 
stream to account for the observed facts. 
Branching of Eskcrs. — The subglacial streams of a stag- 
nant ice-sheet should branch in essentially the same fashion 
as the existing or postglacial streams of the same region ; and 
the branches should, in general, be aggraded almost equally 
with the main stream. But this is clearly not the fashion of 
eskers. for they are little given to true, river-like branching, 
especially in their lower or more southern courses ; and 
branches making large angles with the main esker 
are almost unknown. In fact, there are practically no 
branches, in the sense of minor tributaries to a main 
line ; but we observe instead an occasional confluence at very 
oblique angles of eskers of approximately equal size and 
length, suggesting the primary rather than the secondary or 
lateral branching of rivers. This, as previously noted, is all 
precisely what we should expect in the case of superglacial 
streams on the comparatively high gradient of the marginal 
slope of the ice-sheet, with 'SO great a thickness of easily-erod- 
ed ice above the base level that the main stream had no diffi- 
culty, with its higher gradient, in cutting its canyon below 
the canvon of its lateral tributaries and making of the latter 
