694 Mr. J. H. Jeans on 
= N a. For the same quantity Rose-Innes * deduces from 
some work of Callendar the value 
b=-00209. 
If we take 6 to be the mean of these, we obtain 
2 
3 No*=-002035, 
so that 
No*=:00097 cubic cm. . . .. 4am 
From relations (ii.) and (i1.) we obtain 
ya (he? 2909x100). 
(No*)? 
This is the value of N deduced by the kinetic theory trom 
observations on a single gas, air. From calculations on a 
variety of other gases, “I have found values of N differing 
slightly from the foregoing, the average value being per- 
ceptibly lower—about 4°6 x 10" . 
§3. There is, however, no question that the value of N 
deduced from the kinetic theory is not likely to be accurate. 
A glance at equation (iv.) will show that before the kinetic 
theory can arrive at a value of N, it has to double the errors 
of observation, already large, in the determination of 6, and 
to treble those in the determination of the coefficient of 
viscosity. 
A better way of determining N is by electrical ‘methods, 
since N can be determined at once from the ionic charge. 
By this method there is no multiplication of observational 
errors, and, moreover, there is no assumption necessary as to 
the shape of the molecules. For the ionic charge we have 
the following determinations :— 
J.) Themen. 3024 3°4x 10-29, 
Ay A. Wileen- . wate 31x 10-20, 
J, §: Bownsend 2 2aey, 3°0 x 10-29, 
The value for the ionic charge deduced from the value 
N=4'6 x 10" given above would be about 
rt Mea gee 
If we determine N from the three above observations, we 
obtain respectively 
N=3'6x 10%, 40x10", 4:1 x 10. 
* Phil. Mag, v. p. 48. 
