1892.] 11 [Upham. 
for a century or more, brought increased snowfall, which sufficed 
to hold the ice boundary nearly stationary, perhaps frequently 
first having pushed it again a considerable distance forward. 
The thick ice lying far back from the border may then have 
flowed over its previously thin and drift-covered outer belt, aiding 
with the new snowfall to envelop the once superglacial drift 
stratum within the ice-sheet. These halts or re-advances, if the 
front of the ice had a nearly constant position during several 
years, became marked by terminal moraines, of which I have 
mapped a series of eleven in consecutive order from south to 
north in Minnesota, North Dakota, and Manitoba, while Mr. 
Frank Leverett has traced a still larger number through Illinois, 
Indiana, and Ohio. With the increased thickness and steeper 
gradient of the outer belt of the ice-sheet while the recession of 
its boundary was slackened, wholly stopped, or changed to a re- 
advance, due mainly to very abundant snowfalls, much drift 
which had been formerly exposed on the ice surface would become 
again englacial, so that a stratum of drift several feet thick might 
be enclosed in the ice at an altitude increasing inward from less 
th^n 50 feet to 500 feet or more. 
The upper current of the thickened ice above the englacial bed 
of drift would move faster than that drift, which in like manner 
would outstrip the lower current of the ice in contact with the 
ground. Close to the glacial boundary, whether it halted and 
even re-advanced or merely its retreat was much slackened but 
did not entirely cease, which last seems probably to have been 
the case Avith the drumlin areas of Massachusetts and New Hamp- 
shire, the upper part of the ice must have descended over the 
lower part. This differential and shearing movement, as I think, 
gathered the stratum of englacial diift into the great lenticular 
masses or sometimes longer ridges of the drumlins, thinly under- 
lain by ice and overridden by the upper ice flowing downward to 
the boundary and bringing with it the formerly higher part of the 
drift stratum to be added to these growing drift accumulations. 
The courses of the glacial currents and their convergencies to the 
places occupied by the drumlins were apparently not determined 
so much b}^ the topography of the underlying land as by the con- 
tour of the ice surface, which under its ablation had become 
sculptured into valleys, hills, ridges, and peaks, the isolation of 
the elevations by deep intervening hollows being doubtless most 
conspicuous near the ice-margin. 
