wo exhaustions, each of about two hours' duration, have, however, in
my experience succeeded very well, provided, of course, that the
dielectric is prepared as suggested. At the end of the exhaustion
process the clamping screws are tightened as far as possible, the
condenser remaining in its bath until the paraffin is pasty.

Condensers made in this way resist the application of alternating
currents perfectly, as the following tests will show. The dielectric
consisted of about equal parts of hard paraffin and vaseline. A
condenser of about 0.123 microfarads capacity and an insulation
resistance of 2000 megohms, [Footnote: As tested by a small voltage.]
having a tin-foil area of 4.23 square metres (about), and double
papers each about 0.2 mm. thick, designed to run at 2000 volts with a
frequency of 63 complete periods, was tested at this frequency.

The condenser was thoroughly packed all round in cotton-wool to a
thickness of 6 inches, and its temperature was indicated more or less
by a thermometer plunged through a hole in the lid of the containing
box and of the condenser box, and resting on the upper surface of one
set of tin-foil electrodes, from which the soft paraffin mixture had
been purposely scraped away. The following were the results of a four
hours' run at a voltage 50 per cent higher than that for which the
condenser was designed--i.e. 3000 volts.

Times. Voltage Temperature Temperature Difference
in Condenser. in Air.

Hrs. Min.

2 10 2750 22.8 deg. C. 23.0 deg. C. + 0.2 deg.

3 10 2700 23.0 deg. C. 23.3 deg. C. + 0.3 deg.

3 18 3200 23.1 deg. C. 23.0 deg. C. -0.1 deg.

4 10 3200 23.3 deg. C. 23.7 deg. C. + 0.4 deg.

5 10 3100 23.6 deg. C. 23.4 deg. C. -0.2 deg.

6 10 3000 23.8 deg. C. 23.35 deg. C. -0.45 deg.

An idea of the order of the amount of waste may be formed from the
following additional experiment.

A condenser similar to the one described was filled with oil of a low
insulating power. It was tested calorimetrically, and also by the
three voltmeter method, which, however, proved to be too insensitive.
The temperature rise in the non-conducting box in air was about 0.3 deg.
C. per hour, and the loss of power was found to be less than 0.1 per
cent. In the present case the actual rise was only 1 deg. in four hours,
and the integral give and take between the c