Magneto-optic Phenomenon. 307 



values of the minimum rotation agree for positive and negative 

 magnetization. The mean error in the principal positions of 

 the Nicol prisms in which the planes of polarization are 

 parallel and perpendicular respectively to the plane of incidence 

 has about the same value. As already previously indicated in 

 § 8, this direct method of determining the principal positions 

 gives far more accurate results than the indirect methods 

 usually employed hitherto. 



14. Since we may obtain the amplitude fi and the phase m 

 of the magneto-optic component both from the minimum- 

 rotations and the zero-rotations, we must now show that the 

 results agree. For this purpose the zero-rotations were also 

 determined for an angle of incidence of 51° 22', because with 

 large and small angles of incidence the accuracy of the 

 observations is considerably decreased, and the minimum- 

 rotations, which are then capable of being determined with 

 the greatest approach to accuracy, are still sufficiently well 

 defined. As we have already shown that only the amplitude jjl 

 and not the phase m of the magnetic component of the light 

 changes its sign with the direction of magnetization, we only 

 require to measure the angle -\jr between the zero-positions of 

 the Nicol prisms for positive and negative magnetization (cf. 

 § 5). In the following tables the values of fju and m are given 

 for positive magnetization. The observed rotations were : — 



2*i- 2 C, W ¥ 2«, 



_0'<3 -6'-4 -6'-61 6 , , -0'-4| , 



_0'-6 -7'-3 -6'-9 -0''3 



-0'-45 -6'-85 -6'-8 -0'-35 



A discussion of the different series of observations shows 

 that the mean error in these numbers is smaller than 0''5. 

 From the minimum- and zero-rotations the following values of 

 fju and m were obtained f : — 



Tip 



20°. 



%• 



2 C 



fx v 



ff 



m r 



m r 



0'-4 



-6'-8 



-6'-8 



-0'-3 



+079 xlO -3 



H-0-81xl0 -3 



180°-1° 9' 



180° -0° 52' 



l'-2* 



-6'-6* 



-6'-8* 



-0'-5* 



+0-69* „ 



+0-78* „ 



180° -3° 40'* 



180°- 1° 17'^ 



<b m . 

 r ip 



20™. 



r ia 



2 ^« 



2 C« 











7'-9 



-6'-2 



-6'-8 



8'-0 



0-754 „ 



0-756 „ 



180° 4-5° 35' 



180°4-7° 21' 



7'-8* 



~6'-2* 



-6'-4* 



7'-8 



0-750* „ 



0-765 „ 



180° 4-3° 47'* 



180° 4-0° 37' 



t The formulas (3) and (4), § 4 and § 5, leave the sign of the amplitude /* 

 undetermined, and the quadrant in which the phase m lies depends upon 

 this. In the table fi i and fx L are always taken as positive. 



