819 



known values (for helium and hydrogen e.g.) or for the correction 

 of already determined values (among others for oxygen). 



Now for H^ has been found s'= ± 2,9, ƒ =4,83; so this agrees 

 very well with our type, where v = 0,6 (.? ==: 3, ƒ ' = 4,8). Ac- 

 cordingly we may expect for H^ a straight diameter, the reduced 

 coefficient of direction of w4iich will amount to 0,6. 



For helium has been found .9 = 3,13, ƒ ^4,46; this too may be 

 correct. If s is really 3,13, ƒ w^ould even be greater than 5. But if 

 ƒ is no more than 4,5, then y would be = 0,56, and s = 2,9, just 

 as for H^. 



And, at last, for 0^ has been found s = 3,346,/= 5,76, y := 0,813, 

 where the values of s and ƒ are in good harmony, but y deviates 

 greatly. For with ƒ =: 5,76 would correspond y=0,72, 5 = 3,349, 

 so that s as has been said, agrees beautifully, but y ought to be 

 considerably loAver than 0,813. But here the fact may also have 

 influence that liquid oxygen is an associated liquid, and that hence 

 (just as for ethyl and methylalcohol, acetic acid, etc.) y is higher 

 than the normal value. 



7. We will not drop this subject before having set forth a few^ 

 points. First: the circumstance, to which putting X^:X^z=:l really 

 comes, is this. According to (r) X^:l.,=zs:ii{l — V.r) ^"d so when 

 /\=:).^ is assumed, this comes to this that s =: Sf' -. {8 -\- f') (.see 

 above (7)). This is experimentally satisfied. But further /' = 8y — 

 also in consequence of the new relation (12) — and according to 

 (21) b'k = {f)k — b^y : bic vie. The latter is in connection with the form 



1) Substance with high molecular weight. 



") Mean of w-pentane, i-pentaue, fluorbenzene, etc. 



