the Lines of various Spectra, 35 



Then, the ratio of these variations, 



~ 7 0283" " l6 Zb > 



must be a constant for all the natural families. 



The spectrum of germanium is partially known ; and it can 

 be shown chemically that it belongs to the same family as 

 silicon and tin. So, treating its atomic weight as an unknown 

 quantity, and assuming certain values for the wave-lengths 

 and atomic weights of silicon and tin, we have : — ' 



X. 



AX. 





A 2 X. 



" Variation." 



Si. 4010 











Ge. 4453 



443 

 624 





181 



S=- 4051 



Sn. 5077 











Atomic weight. 











Si. 28 











X 



Ge. x 



118 



-28 



— x 



ut 



\ — 2x 



146-2*; v 

 tf-28 - X 



Sn. 118 











By his law, 













•4051 

 X 



= 



•37584 

 •0283 " 



= 13-28, 



.*. 



X 



= 



•03051, 





.-. 



X 



= 



72-32. 





So M. de Boisbaudran succeeds in getting identically the 

 same value for the atomic weight as Winckler afterwards 

 found by chemical means. This is surely a wonderful con- 

 firmation of the law, if the lines assumed to be homologous are 

 so in reality. It is true that the atomic weights of the ele- 

 ments are not known so accurately as to justify M. de Bois- 

 baudran in attributing such exactness to his deductions ; and 

 so the method ought not to be applied to the determination of 

 atomic weights. But, if the law is correct, it may well serve 

 to identify homologous lines. Let us apply it to the case of 

 magnesium, zinc, and cadmium. The ratio of the " varia- 

 tions," which is a constant for all natural families, is not 

 known more accurately than 13 '0 ; but this will do very well 

 for our purposes. The variation in the increment of the 



D2 



