580 Proceedings of the Royal Society 
the decomposing action has ceased, the resistance of the films (of 
gas or oxide) which are deposited on the electrodes may change in 
value. That neither of these circumstances produces any marked 
effect is, however, amply proved by the numbers which follow, 
which, though given only as first approximations, are within the 
limits of difference of the results given (from galvanometric deter- 
minations) by former experimenters. 
The experiments were all made in my laboratory, mainly under 
my own direction, but sometimes under the eye of my assistant, 
Mr W. R. Smith. Able assistance was rendered by several of m} r 
practical students, — two months ago by Messrs Russell Smith and 
J. C. Young, more recently by Messrs Browning and Nichol. 
As the polarisation in most cases diminishes with very great 
rapidity from the instant of breaking contact with the decomposing 
battery, and as (for this and other reasons) the mode of measure- 
ment by the first swing of the index-needle of the electrometer is 
not deserving of much confidence, it was necessary to devise a 
process by which the electrometer could he charged at leisure up 
to any desired potential, and then, for an instant only, placed in 
connection with the electrodes. The apparatus I employed bears 
a certain analogy to the wippe of Poggendorff, but differs from it 
in some essential particulars, both of construction and mode of 
working. 
In a plate of vulcanite, or other good insulator, ten holes are cut 
as below, and filled with mercury. Those marked E are connected 
with pairs of opposite quadrants of the electrometer, P with the 
electrodes, Bj with the decomposing battery, and B 2 with the 
auxiliary (or charging) battery. Also metallic connection, as indi- 
