~ 
ELECTRICAL CONDUCTIVITY OF CERTAIN SALINE SOLUTIONS. 59 
feet were so adjusted as to be capable of dipping into the six mercury pools 
c, c', d, d’, e, e’, but not all at once, for the two feet which dipped into d and e 
were longer than the other four, so that the rocker could oscillate about an 
imaginary line joining d and ¢ as an axis. Its natural position was one inclined 
to the left, that is to say, with feet dipping ¢, c’, d, and e, and the other two legs 
suspended in the air above d@’ and e’. In this position of the rocker no current 
was passing through the tube, but the ends of the tube were connected together 
through ¢ and ¢’. Hence so long as the rocker remained thus, the tube was 
“short circuited,” and depolarised itself much more rapidly than it would have 
done had its terminals been left insulated,—that is to say, the polarisation pro- 
duced in it by previous currents disappeared much more rapidly than it would 
otherwise have done. 
When g was rocked to the right this connection between the ends of the tube 
was broken by the left hand part of the rocker being lifted out of the pools 
eand ¢’; and the battery and galvanometer circuits were completed by con- 
nection being established between d and d’, and e and ¢é’ respectively. This 
state of things lasted for an instant only, as the rocker was immediately allowed 
to fall back to its natural position, that is, inclined to the left. During that 
instant a current was passing through the tube, and it was by its instantaneous 
effect upon the galvanometer that the resistance of the liquid was ascer- 
tained. But this current, however short its duration, produced a sensible and 
even considerable polarisation in the tube. This polarisation soon reduced itself 
when the rocker g fell back to its natural position, in virtue of the connection 
between ¢ and c’, and the tube returned in a few seconds to a neutral state, in 
which it was ready for another rock of g. 
We found, after a time, that it was best to connect ¢ and e’ by a separate 
key, because if by the previous method d and d’ were connected, even a very 
small fraction of a second before ¢ and ¢ (as might easily happen through an 
inequality in the length of the rocker’s legs), polarisation went on to such an 
extent before the galvanometer circuit was closed as to render the measurement 
incorrect. In order to make the rate of polarisation as small as possible, the 
current was diminished by inserting 10,000 B. A. units into each of the two 
arms of the bridge (R, and R,). The resistance of the tube itself varied 
from 1000 to 10,000 B. A. units. By these high resistances the current was so 
weakened as to make the amount of electrolysis in a short time quite inappre- 
ciable, while the rate at which the electrodes became polarised was made so 
slow, that the effect of a current upon the solution could be noted before polari- 
sation had time to interfere with the result. 
The mode of testing will be best understood by an example. Suppose that 
the resistance R of the coils is nearly equal to that of the tube. Insert 7 and 
VOL. XXVII. PARTI. - Q 
