of Metal Rods with rounded ends. 
273 
where i is the current, V the voltage, and L the self-induction of 
the circuit. In order that the current should attain a value 
V 
sensibly equal to , before the end of the contact, it is necessary 
ill 
that 
R 1 
L 
shall be as large as possible. 
The self-induction, L, is 
almost entirely due to the coil of the galvanometer, and, owing to 
the peculiar construction of the latter, was very small indeed, 
so that, with the large value of R 1 employed, the above condition 
was easily fulfilled. 
Integrating the above equation, we get: — 
J j ^ 
Now and e L are both very small quantities, and we may 
Xtl 
certainly neglect their product, so that we get: — 
V 
q = R, 
T — 
L 
Ri 
or 
Ri 
( 2 ). 
This gives the correction for self-inductive growth of the current, 
which is a constant addition to the time. There may also be a 
correction for spark at break, and this remains to be investigated. 
The simplest method of getting an absolute value of the time 
would be to charge a condenser to a known voltage, partially 
discharge it through the rods during impact, and then compare 
the flings produced on the galvanometer by the whole charge, and 
by the remainder after impact. By this means we should have 
no self-induction or spark effects accompanying the discharge 
through the rods, and the self-induction of the galvanometer 
would not matter. It was, however, quite impossible to do this, 
owing to leakage between the various operations, and a modified 
method had to be adopted. For this purpose the standard con- 
denser (C.,, Fig. 1 A) was used, in the same configuration as in 
calibrating the galvanometer. The resistance round the main 
circuit from X to Y being very small, it follows that if the 
resistances, R lt R„, are both kept large, the potential difference 
between X and F, when the rods are in contact and currents 
flowing, wil 1 be very small compared with the whole potential 
employed. The currents flowing in the circuit C 2 K. 2 Xr 2 r 1 YC 2 , 
and in the main circuit, B 1 GXr i r 1 YB u will therefore be 
entirely independent of each other. Thus if we charge the 
condenser, C 2 , and then hold down both the keys, K 1 and 7i._, while 
the impact takes place, we shall get the same current running 
