| 
} 
} 
the Size of Molecules. 693 
we can eliminate N, and so determine co. Conversely, of 
course, we could, if we wished, eliminate o and determine N. 
§ 2. Let us begin by considering the determination of N. 
The best observations in the first three classes are probably 
those on the viscosity of air. The best in the last two classes 
are those on the deviations of air from Boyle’s Law. 
As regards observations on the viscosity of air, the coefti- 
cient of viscosity at normal temperature and pressure is given 
in Landolt and Bérnstein’s tables as ‘0001714, this being a 
mean value derived from the consideration of a large number 
of experiments. Meyer (‘ Kinetic Theory of Gases,’ p. 190), 
also considering a great number of experiments, takes the 
value :000172. The usual formula for the coefficient of 
viscosity derived from the Kinetic Theory of Gases is 
ie oO0 pel, Sn eee lp 
where / is the mean free path, ¢ the mean velocity, and p the 
density of the gas. 
We may take the values of p and ¢ at normal temperature 
and pressure to be 
p— 00293 
c=45100 cm. per sec. 
The value of /, as will be proved in the paper immediately 
following this, must be taken to be 
differing from Maxwell’s value by the presence of the factor 
1°255.. in the numerator. 
Substituting these values in formula (i.), we obtain 
0001714=-350 x “0001293 x 48500 x ==? 
/ 27No? 
from which we derive | 
Ne? aaUbred, mie. A sa a, a 
Turning to deviations from Boyle’s Law, Van der Waals * 
deduces from Regnault’s experiments the value | 
b=-00198, 
where 6 is the usual 6 of Van der Waals’ equation referred 
to normal temperature and presssure, the value of b being 
* ‘Continuity of the Liquid and Gaseous States, p.400 (English trans.). 
