242 Prof. E. Taylor Jones on a 



Under these circumstances it is well-known that each circuit 

 oscillates with two periods. In fig. 3 some trace is shown in 

 the early part of the curve of the shorter period oscillation 

 superposed upon the longer wave. The existence of the two 

 oscillations can be very clearly shown by connecting a spec- 

 trum-tube in the secondary circuit and photographing its 

 image formed by the rotating mirror *. 



The second method employed to produce the oscillations 

 consisted in sparking, with an induction-coil actuated by the 

 same break as before, to the plates of the condenser as the 

 spots were crossing the plate, the condenser being connected 

 to coil A and to the electrometer. This method has the 

 advantage of allowing the oscillation-curves for the simple 

 circuit of condenser and A to be obtained without the com- 

 plications arising from the presence of a second coil inductively 

 influencing it. 



In the curves obtained by this method the first wave or 

 two represent the oscillations of the condenser with its plates 

 connected by A and (through the spark-gaps) the induction- 

 coil in parallel. The sparks cause, however, rapid damping 

 in the latter circuit, and the remaining waves represent the 

 oscillations of the. condenser and A alone. Before a photograph 

 was taken, the tension of the strip was adjusted so that the 

 lowest points of these latter waves were in or very nearly in 

 the zero line. Fig. 4 shows a curve obtained in this 

 manner. The period of oscillation was in this case about 

 •00238 second. 



The plates used were the " Flashlight " plates manufactured 

 by the Imperial Dry Plate Co. Ltd. A single transit of 

 the spot of light across one of these plates is sufficient to give 

 a well-defined curve. 



(4) Measurement of Periods of Oscillation. 



If the two spots moved with equal velocities across the plate, 

 a comparison of the wave-lengths of the electrometer curve 

 and of the corresponding part of the tuning-fork curve would 

 give the ratio of the periods of oscillation. It was found, 

 however, that the tuning-fork spot always moved more 

 rapidly than the other in the first half of the plate, and more 

 slowly in the second. This was seen by measuring the 

 horizontal distance between the spots in various parts of the 

 plate. In order to find the ratio of the velocities in any part 

 of the wave, the spots were photographed in a number of 

 positions at intervals of a centimetre or less on each plate. 



* See Colley, Wied. Ann. xliv. p. 120 (1891). 



