the most intense cold.

It is evident that what would be necessary again to envelop the boreal
parts of North America with a glacial sheet would not be a
considerable decrease of heat, but an increase in the winter's
contribution of frozen water. Even if the heat released by this
snowfall elevated the average temperature of the winter, as it
doubtless would in a considerable measure, it would not melt off the
snow. That snowfall tends to warm the air by setting free the heat
which was engaged in keeping the water in a state of vapour is
familiarly shown by the warming which attends an ordinary snowstorm.
Even if the fall begin with a temperature of about 0° Fahr., the air
is pretty sure to rise to near the freezing point.

It is evident that no great change of temperature is required in order
to bring about a very considerable increase in the amount of snowfall.
In the ordinary succession of seasons we often note the occurrence of
winters during which the precipitation of snow is much above the
average, though it can not be explained by a considerable climatal
change. We have to account for these departures from the normal
weather by supposing that the atmospheric currents bring in more than
the usual amount of moisture from the sea during the period when great
falls of snow occur. In fact, in explaining variations in the humidity
of the land, whether those of a constant nature or those that are to
be termed accidental, we have always to look to those features which
determine the importation of vapour from the great field of the ocean
where it enters the air. We should furthermore note that these
peculiarities of climate are dependent upon rather slight geographic
accidents. Thus the snowfall of northern Europe, which serves to
maintain the glaciation of that region, and, curiously enough, in some