e way of correcting the error of capacity is--to ascertain (1) the
neutral point of the instrument, or that height at which the zero of the
scale is exactly at the height of the surface of the cistern, and (2) the
rate of error as the barometer rises or falls above this point, and then
apply a correction proportional to this rate. The instrument in which the
error of capacity is satisfactorily (indeed, entirely) got rid of is
_Fortin's Barometer_. Fig. 3 shows how this is effected. The upper part of
the cistern is formed of a glass cylinder, through which the level of the
mercury may be seen. The bottom is made like a bag, of flexible leather,
against which a screw works. At the top of the interior of the cistern is a
small piece of ivory, the point of which coincides with the zero of the
scale. By means of the screw, which acts on the flexible cistern bottom,
the level of the mercury can be raised or depressed so as to bring the
ivory point exactly to the surface of the mercury in the cistern. In some
barometers the cistern is fixed, and the ivory point is brought to the
level of the mercury in the cistern by raising or depressing the scale.

[Illustration: FIG. 3.--Fortin's Barometer.]

In constructing the best barometers three materials are employed,
viz.:--(1) brass, for the case, on which the scale is engraved; (2) glass,
for the tube containing the mercury; and (3) the mercury itself. It is
evident that if the coefficient of expansion of mercury and brass were the
same, the height of the mercury as indicated by the brass scale would be
the true height of the mercurial column. But this is not the case, the
coefficient of expansion for mercury being considerably greater than that
for brass. The result is that if a barometer stand at 30 in. when the
temperature of the whole instrument, mercury and brass, is 32deg, it will
no longer stand at 30 in. if the temperature be raised to 69deg; in fact,
it will then stand at 30.1 in. [Sidenote: Corrections of the barometer
reading.] This increase in the height of the column by the tenth of an inch
is not due to any increase of pressure, but altogether to the greater
expansion of the mercury at the higher temperature, as compared with the
expansion of the brass case with the engraved scale by which the height is
measured. In order, therefore, to compare with each other with exactness
barometric observations made at different temperatures, it is necessary to
reduce them to the h