ry in the
winter months, the wind frequently shifts on a sudden from S.W. to
N.W., commonly about an hour before sunset, and blows with great
impetuosity in the latter direction, attended with a severe frost,
and sometimes with a heavy fall of snow. The effects of this
frost, in places exposed to the wind, are extremely rapid, so as
to render the ground impenetrably hard in about a couple of hours
from its commencement. Situations that are not so much exposed
seem comparatively little affected--at least, I have repeatedly
observed that a small sheltered pond in a field was nearly free
from ice, while the current of a large and rapid river at no great
distance was nearly choked up by it. I believe that the phenomenon
under consideration seldom occurs except during such frosts as
these, and the following are the principal circumstances connected
with it which I am able to state from my own observation.

It may here be premised that ice of this description is seldom
seen adhering to anything beside rock, stone, or gravel, and that
it is more abundantly produced in proportion to the greater
magnitude and number of the stones composing the bed of the river,
combined (as will be further noticed) with the velocity of the
current. I have been informed by a friend that he has occasionally
seen it attached to solid wooden piles at a considerable depth
below the surface of the water, but I never saw or heard of any on
earth, mud, or clay. It is not easy to ascertain the precise time
at which the process begins to take place. It appears, however,
almost invariably to commence during the first night of the frost,
and probably within a few hours after sunset. On the ensuing
morning the first thing which strikes an observer is an immense
quantity of detached plates of ice floating down the stream. Mr.
Knight naturally enough supposed these to have been formed at the
surface by the influence of the freezing atmosphere, and
afterwards absorbed by the current; but I think that a minute
inspection would have led him to form a different conclusion--
viz., that they are first formed in the bed of the river, and
afterwards rise to the surface. It is true that none are to be
seen in situations where there is no sensible current, and that
they abound most in rough and rapid places; but on closely
examining any stream of moderate velocity, yet smooth, equable,
and free from all appearance of eddy or rippling, a great number
of these plates o