504 



Thus l^a = e.g. for carbon, which is surrounded symmetrically 

 on all sides by atoms or atom groups, as in CH^, CCl^, C,Hg etc. 

 For doubly bound C, as for C,H„ C.H,, C,H,S, we find y/ a=l,55.W-^; 

 whereas for triple bound C, as for CjH, (likewise for CO, CO,, CS,, etc.), 

 the full residual value y/a = 3,1 . IQ--^ is found (see 1). But 

 only for the free atoms in the element carbon (CJ we find the so 

 much higher, ten times higher value \.^ a = 32.10—^. 



And the small deviations between theory and experiment which 

 still remain cannot detract from this fact — not for the other elements 

 either. Whether the value 32 will perhaps have to be replaced by 

 33 in the end, or the value 40 by 41 or 42 — this does not affect 

 the above in the least. And it is noteworthy that also the elements 

 of the minor-group Ti, Zr, Ce, Th., of which so little is known, yet 

 confirm this important fact in the clearest way. Besides we found 

 this already indubiously expressed for Antimonium and Bismuth (see 

 IV) with \/ak = 32,5 , resp. 36 . 10"^ 



As far as the values of bk are concerned, they appear to be the 

 same as those which are also calculated from the compounds (if 

 present) — which might have been expected beforehand. 



In my next paper I hope to treat the exceedingly important 

 elements of the group of the alkali-metals, besides those of the 

 minor-group Cu, Ag, Au. 



Clarens, May 1917. 



