394 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
the time that the deposit is being let down upon the ground, and 
the resulting esker will then conform in elevation with the terminal 
plain, at least as closely as eskers commonly do, and show the char¬ 
acteristic lack of sympathy with the ground topography. Should the 
stream be diverted to another course before the subsidence of its 
deposits is completed at all points, the process will continue with¬ 
out essential change, unless a crevasse should draw off the water 
saturating the gravel, the result being one of the eskers falling in 
whole or in j^art below the normal elevation. The exceptionally 
high points or knolls may, according to this hypothesis, represent 
the detritus discharged by hanging tributaries into the main canyon 
after the diversion of the headwaters from the latter. All the con¬ 
ditions favor vertical rather than lateral melting of the ice ; and 
there is, apparently, no tendency to scatter the deposit in getting it 
down to the ground. At the same time all observed variations in 
width are fully provided for, and more especially the gradual widen¬ 
ing which commonly marks the junction with the terminal plain. 
Certainly nothing is more probable than the widening of the ice 
canyon at its mouth in the thin and frayed or lobate southern margin 
of the ice, where its walls are bathed by a large body of standing 
water. Stone’s maps (’99) show that this terminal widening is a 
particularly characteristic feature of the great esker systems of 
Maine; and some even of the subglacial tunnels of the Malaspina 
glacier become, at the last, canyons with rapidly diverging walls. 
The constant movement and readjustment of the deposit during the 
settling process keep it loose and permeable, and enable the water 
to wash out some of the filling of the coarsest gravels and thus give 
rise to the open-work gravel specially noted by Davis. 
In the quite exceptional case where crevasses or other accidents 
drain the esker channel before its subsidence is completed, the dif¬ 
ferential melting is likely to be reversed, the interstream surfaces 
going most rapidly because less protected from the sun’s heat; the 
esker deposit is left on a ridge of ice, and, sliding down on both 
sides, gives rise to the occasional double esker, of which I have 
noted several good examples in the vicinity of Boston (’94, p. 278- 
284). The quite common failure of the esker to connect properly 
with its terminal plain, a weak place or break intervening, may be 
in part attributed to the deposition of the head of the plain over the 
sloping margin of the ice and the subsequent melting of the latter; 
