for the greatest amount of time near the front
or outer rim of the ice field, for there, we may presume, that for
the longest time the cutting rim would rest upon the bed rock of the
country. As the ice receded, this rim would fall back; thus in the
retreat as in the advance the whole of the field would be subjected to
a certain amount of erosion. On this supposition we should expect to
find that the front of a continental glacier, fed with pressure-molten
water from all its interior district, which became converted into ice,
would attain much warmer regions than the valley streams, where all
the flow took place in the state of ice, and, furthermore, that the
speed of the going on the margin would be much more rapid than in the
Alpine streams. These suppositions are well borne out by the study of
existing continental ice sheets, which move with singular rapidity at
their fronts, and by the ancient glaciers, which evidently extended
into rather warm fields. Thus, when the ice front lay at the site of
Cincinnati, at six hundred feet above the sea, there were no glaciers
in the mountains of North Carolina, though those rise more than five
thousand feet higher in the air, and are less than two hundred miles
farther south. It is therefore evident that the continental glacier at
this time pushed southward into a comparatively warm country in a way
that no stream moving in the manner of a valley glacier could possibly
have done.

The continental glaciers manage in many cases to convey detritus from
a great distance. Thus, when the ice sheet advanced southwardly from
the regions north of the Great Lakes, they conveyed quantities of the
_debris_ from that section as far south as the Ohio River. In part
this rubbish was dragged forward by the ice as the sheet advanced; in
part it was urged onward by the streams of liquid water formed by the
ordinary process of ice melting. Such subglacial rivers appear to have
been formed along the margins of all the great glaciers. We can
sometimes trace their course by the excavation which they have made,
but more commonly by the long ridges of stratified sand and gravel
which were packed into the caverns excavated by these subglacial
rivers, which are known to glacialists as _eskers_, or as serpent
kames. In many cases we can trace where these streams flowed up stream
in the old river valleys until they discharged over their head waters.
Thus in the valley of the Genesee, which now flows from