406 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
we are asked to suppose that the suhglacial tunnel, often many miles 
in length, is formed at a time when the ice still has a definite motion, 
to the controlling influence of which the tunnel is supposed to owe 
its general trend; that the meanders and transverse reaches of the 
tunnel are not obliterated by the motion of the ice; and that the 
tunnel as a whole survives the necessarily slow cessation of the ice 
movement. This insures to the tunnel and the stream which formed 
and maintains it a good degree of longevity and time for the slow 
building of the esker and terminal plain. But these deposits testify, 
in composition and structure, to rapid work by torrential streams, 
and the stationary ice mai-gin during their formation points unequivo¬ 
cally to the same conclusion. It is an interesting question, therefore^ 
as to what the subglacial stream was doing during the relatively long 
period of its existence anterior to this brief period of tremendous 
constructive activity, and also as to why it should then abruptly 
abandon the channel to which it had adhered so long. In brief, the 
trend of the subglacial stream demands ice control and a long life; 
its deposits demand a short and intensely active life followed by a 
sudden disappearance from the scene of its labors. But whether it 
be long- or short-lived, its existence should be recorded in the ground 
moraine, in the form of interbedded gravels before the ice ceased to 
move, and of erosion channels after the ice became stationary. 
These phenomena, however, are of rare occurrence, and still more 
rarely can they be correlated with subglacial streams or eskers. 
This important problem will receive further consideration in con¬ 
nection with the source of the material of eskers and sand plains. 
Relations of eskers to fro7ital and delta plains .— The main facts 
under this head are well determined and there is little question or 
controversy concerning them. My purpose is simply to call attention 
to the very numerous plains of modified drift, including many true 
delta plains, which have no tributary eskers and apparently never 
have had. They have, however, undoubtedly been formed through 
the agency of glacial streams. If these were superglacial streams, 
the explanation of the absence of associated eskers is simply, as pre¬ 
viously noted, that the plain was formed and the stream diverted 
before its channel was sufticiently base-leveled to permit any notable 
aggrading. But if, instead, they were subglacial streams, the aggrad¬ 
ing of their beds must have been in progress during their entire exis¬ 
tence, or at least during all the time required for the formation of 
