488 
Proceedings of the Royal Society 
I®’-** 
and the discharge will begin at that point of some conductor of the 
system whose surface density first reaches the value^/^I.. 
In this view the beginning of disruptive discharge is conditioned 
solely by the nature of the dielectric in the immediate neighbour- 
hood of a certain point on the surface of the discharging body, and 
by the electrical surface density at that point. This is, in the first 
instance, unquestionably the simplest and most scientific analysis of 
the phenomena, but it has not as yet, I think, been shown that it 
will account for all the observed facts. Experimenters have mainly 
themselves to blame for this ; for they have in too many cases 
simply worked at random, and have consequently fallen among 
circumstances so complicated that a theoretical examination of their 
results is out of the question. Some, for instance, have apparently 
imagined that the breaking tension or dielectric strength of a 
medium could be inferred from the difference of potential required 
to produce a spark of given length between two bodies in it, no 
matter what their form or surroundings. The most casual observa- 
tion of course proves the contrary. I am not aware whether any 
definite experiments have ever been made to show the influence of 
the electrical condition of surrounding conductors. The following 
extracts from notes of some experiments made last summer,* may 
perhaps be of some interest as illustrating this point. P ancPQ are 
two rounded terminals between which the spark was taken. P was 
charged by means of a Holtz machine to a potential, wdiich was 
measured by means of an electrometer, the method employed being 
a modification of that used by Dr. Macfarlane in his experiments 
* With the valuable assistance of Dr. Macfarlane. 
