264 



H. E. J. G. du Bois on Kerr's 



All parts of this combination were rigidly put together, and 

 its optical axis adjusted perpendicular to the medial line (§ 2) 

 of the optical apparatus. Instead of Lippich's slitted screen 

 a " slitted mirror"" of adjustable breadth was set up vertically 

 at 45° to the spectral screen in about the same place where 

 the zirconia disk previously stood. This mirror, therefore, 

 projected a sufficiently homogeneous beam of light into the 

 polarizer. 



§ 21. The positions of the lines Li a, D, b, F, G were 

 marked on the spectral screen and checked by means of 

 spectra of Li, Na, and H. The slitted mirror was afterwards 

 set to these. The " red " light used in the preceding experi- 

 ments was not quite homogeneous, but owing to the slight 

 dispersion sufficiently exact measurements could be made 

 with it. Strictly speaking a definite wave-length cannot be 

 assigned to the " red " light. But in order to allow a re- 

 duction of the spectral measurements to absolute values the 

 question arises : which part of the spectrum is rotated through 

 the same angle as was found for the u red " light ? To this 

 a definite answer may be given ; by trial and interpolation I 

 found for the wave-length \— 62 x 10 -6 centim. 



§ 22. The rotation was determined with each metal for six 

 colours, each angle being obtained from thirty observations. 

 The values of K (in minutes per unit magnetization) are 

 given in Table II. ; the numbers marked with an asterisk are 



Table II. 



Colour. 



Line. 



\Xl06cm. 



Cobalt. 



Bed. 



Li a 



671 



-0208 



" Red." 





62 



-•0198* 



Yellow. 



D 



58-9 



-•0193 



Green. 



b 



51-7 



-0179 



Blue. 



F 



48-6 



-•0181 



Violet. 



Q 



431 



-•0182* 



Nickel. 



Iron. 



0173* --0154 

 •0160* --0138* 



-•0154* 

 -•0159 



-•0163* 

 -•0175* 



-•0130 

 -0111 

 -•0101 

 -•0089 



Magne- 

 tite. 





+ •0096 



4- -0120* 





+ •0133* 





+•0072* 





+ •0026* 





the means of two separate series of observations. In plotting 

 K = funct. (X) perfectly continuous curves of dispersion are 

 obtained, showing the following characters : — 



IV. Cobalt has a faintly pronounced minimum for bluish 

 green. Nickel a minimum for yellow. The iron curve runs 

 almost straight down from red to violet. Lastly, for magnetite 

 a maximum occurs in the yellow. 



