ey were occasioned by a lateral overflow of some
thirty miles of its upper portion, first this way and then that, in such a
brief space of time. The change is equal to nearly 1/15 of the weight of
the whole atmosphere, and the cause, whatever it was, existed within two
or three miles of the earth. Mr. Redfield's explanation I give in his own
words, at length:

"One of the most important deductions which may be drawn from the
facts and explications which are now submitted, is an explanation of
the causes which produce the fall of the barometer on the approach of
a storm. This effect we ascribe to the centrifugal tendency or action
which pertains to all revolving or rotary movements, and which must
operate with great energy and effect upon so extensive a mass of
atmosphere as that which constitutes a storm. Let a cylindrical
vessel, of any considerable magnitude, be partially filled with
water, and let the rotative motion be communicated to the fluid, by
passing a rod repeatedly through its mass, in a circular course. In
conducting this experiment, we shall find that the surface of the
fluid immediately becomes depressed by the centrifugal action, except
on its exterior portions, where, owing merely to the resistance which
is opposed by the sides of the vessel, it will rise above its natural
level, the fluid exhibiting the character of a miniature vortex or
whirlpool. Let this experiment be carefully repeated, by passing the
propelling rod around the exterior of the fluid mass, in continued
contact with the sides of the vessel, thus producing the whole
rotative impulse, by an external force, analagous to that which we
suppose to influence the gyration of storms and hurricanes, and we
shall still find a corresponding result, beautifully modified,
however, by the quiescent properties of the fluid; for, instead of
the deep and rapid vortex before exhibited, we shall have a concave
depression of the surface, of great regularity: and, by the aid of a
few suspended particles, may discover the increased degree of
rotation, which becomes gradually imparted to the more central
portions of the revolving fluid. The last-mentioned result obviates
the objection, which, at the first view, might, perhaps, be
considered as opposed to our main conclusion, grounded on the
supposed equability o