h is itself attached either to the
metallic foot _e_, or to an air-pump. The aperture within the hemisphere at
_f_ is very small: _g_ is a brass collar fitted to the upper hemisphere,
through which the shell-lac support of the inner ball and its stem passes;
_h_ is the inner ball, also of brass; it screws on to a brass stem _i_,
terminated above by a brass ball B, _l, l_ is a mass of shell-lac, moulded
carefully on to _i_, and serving both to support and insulate it and its
balls _h_, B. The shell-lac stem _l_ is fitted into the socket _g_, by a
little ordinary resinous cement, more fusible than shell-lac, applied at
_mm_ in such a way as to give sufficient strength and render the apparatus
air-tight there, yet leave as much as possible of the lower part of the
shell-lac stem untouched, as an insulation between the ball _h_ and the
surrounding sphere _a, a_. The ball _h_ has a small aperture at _n_, so
that when the apparatus is exhausted of one gas and filled with another,
the ball _h_ may itself also be exhausted and filled, that no variation of
the gas in the interval _o_ may occur during the course of an experiment.

1189. It will be unnecessary to give the dimensions of all the parts, since
the drawing is to a scale of one-half: the inner ball has a diameter 2.33
inches, and the surrounding sphere an internal diameter of 3.57 inches.
Hence the width of the intervening space, through which the induction is to
take place, is 0.62 of an inch; and the extent of this place or plate, i.e.
the surface of a medium sphere, may be taken as twenty-seven square inches,
a quantity considered as sufficiently large for the comparison of different
substances. Great care was taken in finishing well the inducing surfaces of
the ball _h_ and sphere _a, a_; and no varnish or lacquer was applied to
them, or to any part of the metal of the apparatus.

1190. The attachment and adjustment of the shell-lac stem was a matter
requiring considerable care, especially as, in consequence of its cracking,
it had frequently to be renewed. The best lac was chosen and applied to the
wire _i_, so as to be in good contact with it everywhere, and in perfect
continuity throughout its own mass. It was not smaller than is given by
scale in the drawing, for when less it frequently cracked within a few
hours after it was cold. I think that very slow cooling or annealing
improved its quality in this respect. The collar _g_ was made as thin as
could be, that t