306 
has taken place. This spark is due to two causes. One is the tendency of the 
Leyden jars forming the capacity C to take on what is known as a residual 
charge. The other results from the oscillatory character of at Leyden jar 
discharge, the jars having a charge after each spark depending on the direc- 
tion of the last oscillation. With a charge on the capacity C varying as to 
both sign and magnitude, one can not expect a constant time interval between 
the sparks L and S. In my later experiments I have been able to eliminate 
much of this trouble by short-circuiting the terminals of the capacity C 
through a high resistance R and an inductance I. The resistance R 1s merely 
a tube of water with wires passing through corks at either end of the tube. 
The inductance I is an electromagnet of about a thousand turns of wire. 
The result may be obtained with either a resistance or an inductance, if suffi- 
ciently large. Using both one can, without reducing the intensity of the 
illuminating spark, reduce the resistance R by shortening the water resis- 
tance until the jars discharge themselves completely very soon after every 
spark. Thus the condenser is brought into the same electrical condition before 
every spark and consequently the time required to charge it to sparking 
potential is made constant. 
The arrangement here described does not completely eliminate all varia- 
tions in the time interval between the sparks because much of the variation is 
due to change in the effective resistance of the spark gaps themselves, some- 
thing the writer has been unable to control. The arrangement does, however, 
reduce the variation about 50 per cent. 
Physics Laboratory, Indiana University, November, 1915. 
