17^ The American Geologist septembor, i»9b 
upon or within or beneath the ice body. In 1884 professor 
Shaler gave reasons for believing them to be formed in chan- 
nels of subglacial streams. Mr. Upham has regarded the 
streams as superglacial after the subglacial channels had be- 
come obstructed, or in some cases as flowing in deep ice- 
walled canyons, reaching perhaps to the ground. Down to 
1890 George H. Stone, in his writings upon the drift phenom- 
ena of Maine, regarded the esker streams as mainly supergla- 
cial, but in 1893 he thought them subglacial, at least in the 
coastal region. In a critical study of the origin of certain 
eskers in Massachusetts by W. M. Davis, and in a discussion 
of the mechanics of the phenomena by J. B. Woodworth, both 
authors conclude that the streams were subglacial; and the 
weight of opinion now favors subglacial streams in the rela- 
tively stagnant submarginal portions of the ice sheet. The 
observations of professor Russell in Alaska and of professor 
Chamberlin upon the ice foot in Greenland, give force to this 
view; There seems to be no better explanation for the greatly 
extended, continuous eskers of uniform cross section, or prism, 
which traverse hills and valleys, than that they were formed 
in tunnels in or beneath the ice; the streams in many cases be- 
ing under hydraulic pressure. But doubtless some of the 
broader or shorter, discontinuous, irregular ridges may have 
been deposited in other ways, near the ice front. A question 
of some importance in this connection is the relative amount 
of interglacial or superglacial drift. In an ice body with little 
debris above its base only subglacial streams could acquire 
great burden. 
In northeastern Iowa occur some unique deposits which 
may be regarded as allies to eskers. They have been de- 
scribed by McGee and named "paha." These are elongated 
ridges of aqueo-glacial material upon the surface, deposited 
in canyons in the ice. ''The streams were originally super- 
glacial, so that the modern drainage is 'superimposed from 
ice,' to use Gilbert's expression; these streams were originally 
located by eminences under the ice, which retarded the flow 
and developed incipient lobation; and after the streams had 
deepened their channels so far as to materially reduce the 
thickness of the ice, then the subglacial water was forced to- 
ward the same lines, and the superglacial and subglacial flow 
