722 
Proceedings of the Poycd Society 
in taking a series of observations, it is necessary to get rid of the 
bubbles after the passage of each spark. They were attracted gene- 
rally to the positive surface, but sometimes to the negative. The 
attraction was more marked when no jars were attached to the 
Holtz ; it was not so powerful as in oil of turpentine, and was gene- 
rally in the opposite direction. 
The electrostatic force required to pass a spark through a layer 
of paraffin oil or of turpentine is constant, whereas it is variable in 
the case of air and other gases. For the observed differences of 
potential plotted with respect to the thickness of layer give a 
straight line through the origin, while in the case of the gases the 
curve is concave. 
The electric strength of the paraffin oil used was found to be 4, 
of the turpentine 3 m 7 ; air being unity. 
To investigate the effect upon the electric spark of heating the 
electrodes, we constructed electrodes of thick platinum wires placed 
at right angles to one another — a suggestion we owe to Professor 
Clerk Maxwell. When one of the wires was heated by a current 
from a battery of four Bunsen elements, the electrometer deflection 
was diminished by about one-fourth of its amount, and that 
whether the wire heated was positive or negative. A similar 
diminution was observed when the deflection for continued sparks 
was taken. This diminution of the difference of potential must be 
due to change at the surface of the wire ; for the air between the 
wires (the shortest distance between the wires being 4 millimetres) 
cannot be so much rarefied by the heating of the wire as to produce 
the effect. 
We have also investigated the effect upon the electric spark of 
heating the air round the discs, the pressure being kept constant. 
We have observed the deflections of the electrometer for a 
constant spark for temperatures from 20° C. to 280° C., and find 
that they indicate a curve, which slopes down gradually as the 
temperature is increased, while the deflections during cooling give a 
curve which is somewhat lower at the lower temperatures. 
It appeared an important matter to ascertain whether the 
electrometer used in all these observations gives deflections strictly 
proportional to the inducing charge. To calibrate it by means of 
cells would have required a very large number ; hence the following 
