J. 



— e tu- ^ bi cos —7+09 sill =^ — TT sm ^ > 

 I ^ oy^ co^ (o^ ) 



476 Mr. G. W. Walker on </i6 



If we put 



n 



where tan d= -5 , we find that 

 pi ' 



Now for Na light b is of order 10^^, wdiile jO" is of order 10^". 



We shall find later that co'^ is of about the same order as p^. 

 Hence with great numerical accuracy to far beyond the 

 violet we may conclude that 



/•CO _ r^ ^2 



(6) Generalization. 



We have now seen that the molecules of the type selected 

 contribute to refraction a term of which an adequate repre- 

 sentation is 



except in the immediate vicinity of a certain pointj) = ll. For 

 this point we require the additional term which has already 

 been discussed. 



Now the real molecule or atom is more complicated than 

 that which we selected. Instead of one small negative particle 

 we have a great many. Each of these will contribute a term 

 of the above form along with a term wdiich is numerically 

 insignificant except in the immediate vicinity of a certain 

 point. If these particles are all similar the w^ will be the 

 same for all, but it does not follow that the critical values at 

 which they are thrown off from the parent atom is the same 

 for all. 1 conclude that for the real atom we shall have the 

 contribution 



''-'=^{'-^iB 



