168 
highly improbable that in the case of a terrace due to olacial 
ghiy g 
molding we should pass abruptly, over a well-defined edge, 
from the steep erosion slope or scarp behind the terrace to the 
normal slope of the drumlin above the scarp; for it is incon- 
ceivable that the erosive action of a continuous sheet of ice 
could be so sharply differentiated. The scarps are not simply 
a little steeper than the normal slopes, but they are in most 
cases as steep as the material will maintain, as steep and 
sharply defined, in fact, as the scarps of till on the modern 
shore which have been protected for a few years from the action 
of the sea. 
We are thus, apparently, forced to refer the terraces to the 
erosive action of running water. It is improbable that strictly 
subglacial streams would cross the drumlins at all; and if they 
did, the natural result would be to produce channels rather than 
terraces. The same would be true of superglacial streams, 
wherever they cut through the ice into the underlying drumlins ; 
and such channels, notching their summits, have been observed 
on some of the drumlins of eastern Massachusetts. The only 
alternative view that now seems worthy of consideration is that 
which refers the terraces to the erosive action of lateral streams. 
When the ice had retreated from the summits of the higher 
drumlins and, presumably, had ceased to flow, the heat reflected 
from the ground would naturally melt away the edge of the ice 
sufficiently to afford pathways for the superglacial streams, and 
drainage channels for glacial waters impounded north of or 
behind the drumlins. These lateral streams, having banks of 
iee on one side and till on the other, would necessarily erode 
the latter in the manner indicated by the existing terraces. 
This explanation accounts for the longitudinal direction of the 
terraces ; for the observed irregularities in height and level; for 
the accumulations of modified drift sometimes observed at the 
astern ends of the terraces ; and especially for the occurrence 
of the terraces chiefly оп the southern slopes of the drumlins, 
since it is on these slopes that the heat of the sun would be 
] 
— 
w 
De 
= 
A == 
