from the top through the oil, many luminous points--air bubbles which
are caught by inserting the primary, and which are rendered luminous
in consequence of the violent bombardment. The occluded air, by its
impact against the oil, heats it; the oil begins to circulate,
carrying some of the air along with it, until the bubbles are
dispersed and the luminous points disappear. In this manner, unless
large bubbles are occluded in such way that circulation is rendered
impossible, a damaging break is averted, the only effect being a
moderate warming up of the oil. If, instead of the liquid, a solid
insulation, no matter how thick, were used, a breaking through and
injury of the apparatus would be inevitable.
The exclusion of gaseous matter from any apparatus in which the
dielectric is subjected to more or less rapidly changing electric
forces is, however, not only desirable in order to avoid a possible
injury of the apparatus, but also on account of economy. In a
condenser, for instance, as long as only a solid or only a liquid
dielectric is used, the loss is small; but if a gas under ordinary or
small pressure be present the loss may be very great. Whatever the
nature of the force acting in the dielectric may be, it seems that in
a solid or liquid the molecular displacement produced by the force is
small; hence the product of force and displacement is insignificant,
unless the force be very great; but in a gas the displacement, and
therefore this product, is considerable; the molecules are free to
move, they reach high speeds, and the energy of their impact is lost
in heat or otherwise. If the gas be strongly compressed, the
displacement due to the force is made smaller, and the losses are
reduced.
In most of the succeeding experiments I prefer, chiefly on account of
the regular and positive action, to employ the alternator before
referred to. This is one of the several machines constructed by me for
the purposes of these investigations. It has 384 pole projections, and
is capable of giving currents of a frequency of about 10,000 per
second. This machine has been illustrated and briefly described in my
first paper before the American Institute of Electrical Engineers, May
20, 1891, to which I have already referred. A more detailed
description, sufficient to enable any engineer to build a similar
machine, will be found in several electrical journals of that period.
30