Possible Existence of Mutual Induction between Masses. 597 



This was observed for several hours, so that it could be 

 accurately plotted and the periodic time noted. 



The flywheel A was then rapidly accelerated (say anti- 

 clockwise) and run up to a speed of 2700 revs, per rnin., so 

 as to give an impulse to B, i£ that were possible. The 

 speed was then maintained constant for one-half the natural 

 period of the swing of B. The flywheel was then rapidly 

 slowed down and the direction of rotation reversed, and the 

 speed increased to 2700 revs, per min. (in a clockwise 

 direction). This process was repeated a number of times so 

 as to induce resonance in B. It will be seen that any action 

 on B due to acceleration of A was in phase with B, while any 

 action due to velocity of A was 90 deg. out of phase. As the 

 time taken to reverse the flywheel occupied only If minutes 

 (a time small in comparison with the half-period of swing), 

 the time-integral of the forces acting on B (if any) might 

 be regarded in the nature of an impulse acting on B when 

 it was near the centre of its swing. Any change in the 

 velocity of: B was most easily calculated from the amplitude 

 of the swing. From the change in velocity of B we can 

 calculate the change in angular momentum B mom . of B. 

 This was expressed as a fraction of the total change of the 

 angular momentum A mom . of the flywheel A. 



In the early experiments made in 1913, it was found that 

 if there was any effect of the kind looked for it was of an 

 exceedingly small order, and that the observed movements 

 of B were mainly due to accidental disturbing forces. At 



this time it was possible to assert that the ratio mom ' was 

 certainly less than 2*3 x 10" s . Amom - 



In the later experiments the chief aim was to diminish 

 the disturbing forces as far as possible, so that the negative 

 result might be stated with the smallest possible limit of 

 error. The introduction of the suspended screen and other 

 refinements greatly improved the steadiness of B when A 

 was running. 



Fig. 4 gives the result of a typical experiment. The 

 horizontal scale gives the deflexion in centimetres as observed 

 on a scale at a distance of 6 metres. The vertical scale 

 (read from top to bottom) gives the time in minutes. The 

 actual deflexion in radians is obtained by multiplying the 

 readings by 8*33 x 10" 4 . 



The method of calculating the gain in angular momentum 

 by B is as follows : The amplitude of the swing of B before 

 the acceleration of A is noted. Let this be 6 lm Then, after 



