337 
Now for this wave-length John Koch finds 
v = 1,0002661. 
If we start from Koch’s value of the index for À = 86,9 and if 
we assume the index for sodium-light to be 
= 1,00027155 (Lorenz) ] . . ( A = 3.95 
\ we obtain 
== 1,0002702 (Ramsay and Travers) J ( A = 2,98 
In cases where the refractivities from which A is being calculated 
differ but slightly, the value of zl, as could indeed be expected, 
changes considerably with the assumed value of the dispersion. 
Accepting (1) we obtain for oxygen 
A 0 = 0,82 
K = 1.000533; 
I have not been able to find data for the dielectric constant of oxygen. 
§ 18. Atmospheric Air. For the reason that air is a mix¬ 
ture and that its composition is variable x ), it is not a suitable sub¬ 
stance for use in our computation. The dispersion of atmospheric 
air appears, however, to have been well investigated and the mea¬ 
surements extend over a considerable range of wave-lengths; I have 
thought therefore that the purpose of the present investigation will 
be best served by including air into the discussion. 
Let us try to apply to this case equation (3) of § 14.: 
R = a 1 L 1 -{-a 2 L 2 . ( 1 ) 
Let oq, refer to oxygen; a 2 , L 2 to nitrogen. In § 19. below we 
shall find that i 02 (for nitrogen) probably does not differ much from 
A 01 (for oxygen, see § 17); so that A 0 being an intermediate value 
lying between A 01 and A 02 we can take, as an approximation, 
L\ — L 2 — 
io 2 ^ 2 
A 2 -Ao 2 ' 
Then from (1) we get 
K*' — 2) = («i + «s) L - 
( 2 ) 
(3) 
1 ) Lorenz and Perreau deprived air of H 2 0 and of C0 2 ; in Scheel’s experi¬ 
ments air was dried but its C0 2 was not absorbed; in Kayser and Kunge’s expe¬ 
riments air was not dried at all. 
