6.), the interval in air fails to be a measure, as yet, of the
insulating or resisting power of the gas in the vessel, yet we may for
present purposes take the mean interval as representing in some degree that
power. On examining these mean intervals as they are given in the third
column (1388.), it will be very evident, that gases, when employed as
dielectrics, have peculiar electrical relations to insulation, and
therefore to induction, very distinct from such as might be supposed to
depend upon their mere physical qualities of specific gravity or pressure.

1395. First, it is clear that at the _same pressure_ they are not alike,
the difference being as great as 37 and 110. When the small balls are
charged positively, and with the same surfaces and the same pressure,
muriatic acid gas has three times the insulating or restraining power
(1362.) of hydrogen gas, and nearly twice that of oxygen, nitrogen, or air.

1396. Yet it is evident that the difference is not due to specific gravity,
for though hydrogen is the lowest, and therefore lower than oxygen, oxygen
is much beneath nitrogen, or olefiant gas; and carbonic acid gas, though
considerably heavier than olefiant gas or muriatic acid gas, is lower than
either. Oxygen as a heavy, and olefiant as a light gas, are in strong
contrast with each other; and if we may reason of olefiant gas from
Harris's results with air (1365.), then it might be rarefied to two-thirds
its usual density, or to a specific gravity of 9.3 (hydrogen being 1), and
having neither the same density nor pressure as oxygen, would have equal
insulating powers with it, or equal tendency to resist discharge.

1397. Experiments have already been described (1291. 1292.) which show that
the gases are sensibly alike in their inductive capacity. This result is
not in contradiction with the existence of great differences in their
restraining power. The same point has been observed already in regard to
dense and rare air (1375.).

1398. Hence arises a new argument proving that it cannot be mere pressure
of the atmosphere which prevents or governs discharge (1377. 1378.), but a
specific electric quality or relation of the gaseous medium. Hence also
additional argument for the theory of molecular inductive action.

1399. Other specific differences amongst the gases may be drawn from the
preceding series of experiments, rough and hasty as they are. Thus the
positive and negative series of mean intervals do not