he
construction of the marine barometer till lately in general use, that
the cistern was not large enough to hold the mercury which descended
from the tube in a low atmospheric pressure.

The practice which has long prevailed of mounting the marine barometer
in wood is objectionable. The instrument recently introduced agreeably
to the recommendation of the Kew Committee, is greatly superior to any
other description of marine barometer which has yet been tested, as
regards the accuracy with which it indicates the pressure of the
atmosphere. The diameter of the cistern is about an inch and a quarter,
and that of the tube about a quarter of an inch. The scale, instead of
being divided into inches in the usual way, is shortened in the
proportion of about 0.04 of an inch for every inch. The object of
shortening the scale is to avoid the necessity of applying a correction
for difference of capacity between the cistern and the tube. The
perfection with which this is done may be judged of from the fact, that
of the first twelve barometers tested at the Liverpool Observatory with
an apparatus exactly similar to that used at Kew (whence these
instruments were sent by railway, after being tested and certified), the
index corrections in the two pressures of 28 and 31 inches in three of
them were the same; two differed 0.001 of an inch; and for the remainder
the differences ranged from 0.002 to 0.006 of an inch. The corrections
for capacity were therefore considered perfect, and, with one
unimportant exception, agreed with those given at Kew.

In order to check the pumping of the mercury at sea, the tubes of these
barometers are so contracted, through a few inches, that, when first
suspended, the mercury is perhaps twenty minutes in falling from the top
of the tube to its proper level. When used on shore, this contraction of
the tube causes the marine barometer to be always a little behind an
ordinary barometer, the tube of which is not contracted. The amount
varies according to the rate at which the mercury is rising or falling,
and ranges from 0.00 to 0.02 of an inch. As the motion of the ship at
sea causes the mercury to pass more rapidly through the contracted tube,
the readings are almost the same there as they would be if the tube were
not contracted, and in no case do they differ enough to be of importance
in maritime use.

The method of testing thermometers is so simple as scarcely to require
explanation. For the freezing