eights at which they would stand at some uniform
temperature. The temperature to which such observations are reduced is
32deg Fahr. or 0deg cent.

If English units be used (Fahrenheit degrees and inches), this correction
is given by the formula

.09T - 2.56
x = -H ------------,
1000

in the centigrade-centimetre system the correction is .0001614 HT (H being
the observed height and T the observed temperature). Devices have been
invented which determine these corrections mechanically, and hence obviate
the necessity of applying the above formula, or of referring to tables in
which these corrections for any height of the column and any temperature
are given.

The standard temperature of the English yard being 62deg and not 32deg, it
will be found in working out the corrections from the above formula that
the temperature of no correction is not 32deg but 28.5deg. If the scale be
engraved on the glass tube, or if the instrument be furnished with a glass
scale or with a wooden scale, different corrections are required. These may
be worked out from the above formula by substituting for the coefficient of
the expansion of brass that of glass, which is assumed to be 0.00000498, or
that of wood, which is assumed to be 0. Wood, however, should not be used,
its expansion with temperature being unsteady, as well as uncertain.

If the brass scale be attached to a wooden frame and be free to move up and
down the frame, as is the case with many siphon barometers, the corrections
for brass scales are to be used, since the zero-point of the scale is
brought to the level of the lower limb; but if the brass scale be _fixed_
to a wooden frame, the corrections for brass scales are only applicable
provided the zero of the scale be fixed at (or nearly at) the zero line of
the column, and be free to expand upwards. In siphon barometers, with which
an observation is made from two readings on the scale, the [v.03 p.0420]
scale must be free to expand in one direction. Again, if only the upper
part of the scale, say from 27 to 31 in., be screwed to a wooden frame, it
is evident that not the corrections for brass scales, but those for wooden
scales must be used. No account need be taken of the expansion of the glass
tube containing the mercury, it being evident that no correction for this
expansion is required in the case of any barometer the height of which is
measured from the surface of the mercury in the cistern.