tlte Vibration of Light and the Plane of Polarization. 189 



J- and -y- the coefficients of absorption in glass and air, that is 



II ''' . . . . , . 



to say, the logarithms of the proportions in which the longitu- 

 dinal vibrations diminish when propagated in the path ^, 

 a' being the angle of diffraction ; then if n sin «= sin a!, we have 



, tan 7 cos (a — a') 



1.2sin<«)^ 



(1) 



The following Table exhibits a comparison of the results of 



this formula with the experiments of Holtzmann. In it I have 



2/ /'■ 

 assigned ?i its ordinary value 1*53, and have taken —^ as 9"5. 



How well the observed values of <y' agree with those calculated 

 from formula (1), will be seen at once from this Table. The 

 most considerable disagreement, that namely of 1° 5' in the third 

 observation, may be accounted for on the grounds of an error of 

 experiment, since this observation, for a diffraction greater by 



only 18' than the second, gives a totally different value of ^ . 



If, however, no experimental error has really affected this result, 

 then I should imagine that the two observations must belong to 

 two different diffraction-spectra; and I can see therein some evi- 

 dence that, for colours that lie far apart in the spectrum, the mag- 

 nitude j-jfj must have somewhat different values, a conclusion 



which on other accounts also is not without probability; but 

 perhaps it is not safe to rest an hypothesis on so small a number 

 of experiments. 



It might be thought that, at all events, diffraction in a single 

 medium would be governed according to Stokes's laws. In order 

 to observe the plijenomenon, two glass plates, on one of which 

 a diffraction grating had been scratched, must be so pressed 

 together that, if possible, Newton's rings may wholly disap^)car. 

 But by means of formula (1) it may be anticipated that^ in this 



