ondenser and the air is
practically nothing; consequently we may consider with safety that the
rate of rise is certainly less than 1 degree per three hours. The
voltage and frequency were about the same in both experiments,
consequently the energy passed is about proportional to the capacity
used in the two experiments.

From this it follows that since the specific heat of both condensers
was the same (nearly), the loss in the present case is a good deal
less than one-tenth per cent. The residual charge is also much less
than when the condenser is simply built up of paper paraffined in an
unsystematic manner, and from which the air and water have been
imperfectly extracted, as by baking the condenser first, and then
immersing it in paraffin or oil.

It is usual to consider that the phenomena of residual charge and
heating in condensers, to which alternating voltages are applied, are
closely allied. This is true, but the alliance is not one between
cause and effect--at all events, with regard to the greater part of
the heating. The imperfect exclusion of air and moisture,
particularly the latter, certainly increases the residual charge by
allowing surface creeping to occur; but it also acts directly in
producing heating, both by lowering the insulation of the condenser
and by allowing of air discharges between the condenser plates.

Of these causes of heating, the discharges in air or water vapour are
probably the more important. Long ago a theory of residual charge was
given by Maxwell, based on the consideration of a laminated
dielectric, the inductivity and resistance of which varied from layer
to layer. It was shown that such an arrangement, and hence generally
any want of homogeneity in a direction inclined to the lines of force
leading to a change of value of the product of specific resistance and
specific inductive capacity, would account for residual charge.

This possible explanation has been generally accepted as the actual
explanation, and many cases of residual charge attributed to want of
homogeneity, which are certainly to be explained in a simpler manner.
For instance, the residual charge in a silvered mica plate condenser,
carefully dried, can be increased at least tenfold by an exposure of a
few minutes to ordinarily damp air. The same result occurs with
condensers of paraffined or sulphured paper; and these are the
residual changes generally observed. The greater part must be due to
creep