of Light hy the Discharge of a Leyden Jar. 317 



band, because then one gets the effect of the separate oscilla- 

 tions ; and they are very bright though very momentary. 



Looking at the formula (4), it is clear that to get large 

 eflPects it is desirable to use a large battery of jars, and to 

 charge to a high potential, only that one is afraid of breaking 

 down the coil. Large capacity and a great number of turns 

 of Avire are the safest ways of increasing the effects. 



One sees also that extra self-induction in the circuit does 

 neither good nor harm to the resultant effect. It diminishes 

 the effect of each oscillation, but it prolongs the time during 

 which they last. And it is the total " action " of the series of 

 decaying swings which the eye perceives. One sees that 

 extra resistance in the circuit is wholly bad. 



As to the cause of what must now be regarded provisionally 

 as the erroneous conclusion of Yillari I see no reason to doubt 

 his experiments, though he does not give sufficient details to 

 enable one to arrive at a perfectly satisfactory judgment on all 

 points ; it is practically certain that he did get a much dimi- 

 nished effect on spinning the flint-glass drum between the 

 poles of the magnet and sending the light along successive 

 diameters of the drum. 



But the cause of this I venture to suggest is possibly to be 

 found in the state of strain into which the glass will be thrown 

 by centrifugal force. It may be said that, if so, the light 

 ought to have been similarly affected even when no magnetic 

 field was employed ; and Villari expressly says that this was 

 not the case. 



But then it is to be noticed that, when no magnetic rotation 

 is attempted, the aspect of the plane of polarization to the 

 stress remains constant throughout the journey; and if light 

 happens to enter with no component modifiable by the stress, 

 it w^ill go out in the same condition. Whereas when rotation 

 has taken place inside the glass this constancy of aspect is 

 destroyed, and the hght on exit has a different component 

 modifiable by the strain to what it had on entrance. 



I do not profess to be able to give a coherent account of 

 how this cause shall give rise to a reduction in the rotation 

 instead of to an elliptical polarization. But then neither am 

 I able to extract from Villari^s account of his experiments 

 any assurance that some elhptic polarization was not pro- 

 duced, and that the reduction of the rotation of the polarized 

 plane was anything more than a mixture of small effects not 

 easily analysable nor precisely defined. 



It is in any case a most interesting experiment, and should 

 be repeated so as to really get at the bottom of the cause of 

 the observed phenomenon. There are many other experiments 

 on whirling glass which may likewise be made. 



