connecting tubes are required. It might appear at first sight as
though one connexion would serve, but the differences in pressure on
which these instruments depend are so minute, that the pressure of
the air in the room where the recording part is placed has to be
considered. Thus if the instrument depends on the pressure or suction
effect alone, and this pressure or suction is measured against the
air pressure in an ordinary room, in which the doors and windows are
carefully closed and a newspaper is then burnt up the chimney, an
effect may be produced equal to a wind of 10 m. an hour; and the
opening of a window in rough weather, or the opening of a door, may
entirely alter the registration.

[Illustration: FIG. 1 & FIG. 2 Anemometers.]

The connexion between the velocity and the pressure of the wind is
one that is not yet known with absolute certainty. Many text-books on
engineering give the relation P=.005 _v_^2 when P is the pressure in
lb per sq. ft. and _v_ the velocity in miles per hour. The history
of this untrue relation is curious. It was given about the end of the
18th century as based on some experiments, but with a footnote stating
that little reliance could be placed on it. The statement without the
qualifying note was copied from book to book, and at last received
general acceptance. There is no doubt that under average conditions
of atmospheric density, the .005 should be replaced by .003, for many
independent authorities using different methods have found values very
close to this last figure. It is probable that the wind pressure
is not strictly proportional to the extent of the surface exposed.
Pressure plates are generally of moderate size, from a half or quarter
of a sq. ft. up to two or three sq. ft., are round or square, and
for these sizes, and shapes, and of course for a flat surface, the