304 



E. Sissingh on Kerr ? s 



of the analyser we have then the telescope of the spectro- 

 meter. Sometimes the polarizer was also removed, because 

 the deviation of the light, although very small (1'), hinders 

 the proper centering and adjustment of the mirror. It is 

 not easy in these operations to avoid every alteration of the 

 mirrors used to read off the rotations. 



The observations gave, for the principal positions of the 

 Nicol prisms : — 



With the Silver Mirror. 



rith Demagnetized Iron Mirror 



Polarizer. 



Analyser. 



388-8 



279-9 



387-0 



280-2 



389-4 



281-4 



388-4 



280-5 



Polarizer. 



388-1 

 388-4 



388-25 



Analyser. 



279-2 

 282-2 



280-7* 



A second series of observations, in which the scale-readings 

 were not quite the same, gavef: — 



Iron Mirror demagnetized. I Silver Mirror. 



olarizer. 



Analyser 



387-3 



283-7 



388-0 



283-3 



387-6 



283-5 



Polarizer. 



387-2 

 387-4 

 387-2 



Analyser 



284-6 

 283-6 

 283-3 



387-3 



283-8 



Since in these observations l/ = l-33, l a '=l'23, we may, 

 taking into consideration the difficulties in the experiment, 

 conclude that the principal positions of the Nicol prisms are 

 the same for the silver mirror and the demagnetized iron 

 mirror. 



13. Now that we have established that the demagnetization 

 is complete, we may proceed to show that the amplitude fi of 

 the magnetic component changes its sign with the direction 

 of magnetization, but that the phase m is not changed by this 

 means. From formulae (1) and (2), § 3, and (4), § 5, it is 

 shown that this is the case if the minimum- and zero-rotations 

 for positive and negative magnetization differ from each other 

 only in their sign and not in absolute value. 



For an angle of incidence of 51° 22' the minimum rotations 

 were determined both for positive and negative magnetization 

 — that is to say, after each series of observations with any 

 given direction of magnetization a demagnetization of the 

 mirror took place. An exact determination of the minimum 

 rotations, however, required the collimator-slit to be reduced 

 by a cross slit to a very short length. If this was not done 



* Each of these numbers is the mean of 8 settings. 



t As the absolute value of the scale-divisions does not influence these 

 results, no notice was taken of its alteration, 



