the Size of Molecules. 695 
§ 4. As regards the determination of oc, it is at once clear 
that the kinetic theory stands in a better position. From the 
relations (i1.) and (iil.), we find for air, 
ee ee Oe 
Here we notice there is no multiplication of errors, so that 
we should expect the error of the result to be only one-half 
or one-third of the error in N. 
This is the best determination of which the kinetic theory, 
by itself, is capable. If, however, we call in outside assistance 
from electrical observations, the case is much improved. 
§ 5. Let us agree to adopt the value 
N=4 x 10" 
from the electrical observations already mentioned, then we 
ean use the formule 
1°25 
k='390pe ,/ aNa?? INAH a edu) 
b=2nNo?, 
for separate determinations of o. It is clear that in these 
determinations of o the error of observation in « is reduced 
to one-half, that in 6 to one-third, and those in N to one-half 
and one-third respectively, in the two cases. The error in 
the determination of o ought, therefore, to be only about 
one-sixth of that in our previous determination of N. 
We have already found that for air, 
No? =3306 sq. cm. 
If we take N=4x 10", we find 
= 2 OO. 
If m is the molecular weight of any gas, then p and ¢, 
referred to normal temperature and pressure, are proportional 
respectively to m and m-2, so that as regards « and m, o is 
proportional to 
K-2ma, 
From the value of o already obtained for air, we can 
deduce the values for other gases, the results being given in 
the following table. The values of « are taken, some from 
Landolt and Bornstein’s tables, and some from recent 
observations. 
