part of its work and to have affected the surface most in the belt of
country a few hundred miles in width around the edges of the sheet. It
was to be expected that in a continental glacier, as in those of
mountain valleys, the most of the _débris_ should be accumulated about
the margin where the materials dropped from the ice. But why the
cutting action should be greatest in that marginal field is not at
first sight clear. To explain this and other features as best we may,
we shall now consider the probable history of the great ice march in
advance and retreat, and then take up the conditions which brought
about its development and its disappearance.

Ice is in many ways the most remarkable substance with which the
physicist has to deal, and among its eminent peculiarities is that it
expands in freezing, while the rule is that substances contract in
passing from the fluid to the solid state. On this account frozen
water acts in a unique manner when subjected to pressure. For each
additional atmosphere of pressure--a weight amounting to about fifteen
pounds to the square inch--the temperature at which the ice will melt
is lowered to the amount of sixteen thousandths of a degree
centigrade. If we take a piece of ice at the temperature of freezing
and put upon it a sufficient weight, we inevitably bring about a small
amount of melting. Where we can examine the mass under favourable
conditions, we can see the fluid gather along the lines of the
crystals or other bits of which the ice is composed. We readily note
this action by bringing two pieces of ice together with a slight
pressure; when the pressure is removed, they will adhere. The adhesion
is brought about not by any stickiness of the materials, for the
substance has no such property. It is accomplished by melting along
the line of contact, which forms a film of water, that at once
refreezes when the pressure is withdrawn. When a firm snowball is