308 R. Sissingh on Kerr's 



In order to determine the influence of a small alteration in 

 the observed quantities upon the calculated numbers, the values 

 of fji and m were reckoned for the rotations denoted by an 

 asterisk and chosen at random. From the observations and 

 the calculated values, we may conclude that 



hp = 4>m <t>i P = *&»> 

 either for zero- or minimum-rotations, and also that 

 p. = fi l} m . = m r 



Two of the zero-rotations, <j>? and $° la , are nearly equal to 

 zero for this angle of incidence. The light reflected from the 

 magnetized mirror can therefore be extinguished by a rotation 

 of one of the two Nicol prisms. The influence of the mag- 

 netization is at this angle of incidence almost equivalent to a 

 rotation of the plane of polarization. The other two zero- 

 rotations, </>° a and <j>° , must therefore differ but slightly from 

 the corresponding minimum-rotations, which is the case. 



It is clear from the tabulated observations that, within the 

 limits of error, the minimum- and zero-rotations give the same 

 results. The degree of accuracy of the amplitudes fi reckoned 

 from <£ m , and of the phases m reckoned from <j>°, is fairly high. 

 The error is not greater than 1 per cent. 



15. In the observations one has to be very careful that the 

 reflexion takes place exactly at the centre of the mirror, since 

 the magnetic spectrum showed that only in this position are 

 the lines of force parallel to the reflecting surface. If by 

 means of an auxiliary reflector, inclined at an angle of 45° to 

 the normal to the mirror, we allowed the light to fall perpendi- 

 cularly on the mirror, the magnetization always produced some 

 effect unless the reflexion took place exactly at the centre of 

 the mirror. This can only arise from the component of mag- 

 netization normal to the mirror. However small was the 

 inclination to the mirror of the lines of force outside its centre, 

 the sign of the magnetization could always be determined by 

 the displacement of the band in the field of the analyser. In 

 order to eliminate entirely the influence of a normal component 

 of the magnetization, the section of the incident pencil of light 

 was reduced to a few square centimetres. 



16. Synopsis of Results. Comparison with Theory. — The 

 zero- and minimum-rotations were determined for various 

 angles of incidence with both mirrors. These angles of 

 incidence were so chosen that the difference in phase of the 

 metallic reflexion varied from one to the next by T ^- of a wave- 

 length, except for angles less than 36°, when the differences 

 were -J wave-lengths. The angle of incidence 61° 30' was 

 chosen in order to more accurately determine the maximum 

 value of the amplitude fi. The observed and calculated 

 quantities are collected in the following Table : — 



