14 Messrs. Richardson, Nicol, and Parnell on the 
solution is therefore 
‘ A re eee - na 
KD oie dy, 0-9, Be Gos flee JOE 
n Lee n a log 9 — log ry 
and the quantity diffusing in unit time per sq. cm. at a point 
determined by ris 
| C, Mn € 
= A 
i a ik 
dC | pn a 
Pir nadrtj  logr,—logr, + 
This solution reduces to the other case, of course, on making 
the radii large compared with their differences. As a matter 
of fact, with the thin-walled tube we used the difference for 
the two sets of formule was less than the expected error of 
our absolute measurements, so the first formula, being some- 
what simpler, was used in interpreting the results, 
Applying these formule to the equation H,=2H, n=2 
and 2X, becomes the positive root of the quadratic 
Oy. ah koCo a4 k Po Ms 
ae P[G+43) 1] 
It wili be noticed that in the general case there are two terms 
in the diffusion formula, one being proportional to the external 
concentration of the dissociated, and the other to that of 
the undissociated gas. The relative importance of these two 
terms depends on the ratios of the coefficients of diffusion 
(n, #) and the solubilities (A—1, a-!). There is a relation 
between the latter, viz., a”/A=k,/k. In the case where the 
traction of external gas dissociated is small compared with 
the whole amount A,=C,=(f,P,)# and the rate of diffusion 
becomes 
1 f u2(ho\' pi, \ 
In this case therefore there are two terms in the formula, 
one proportional to the external pressure and the other to its 
square root. If there were an appreciable amount of dis- 
sociation in the hydrogen outside, the first term would 
increase more rapidly with the pressure than as the square 
root, and the second more rapidly than the first power. 
In order to obtain information on these points, the power 
of the pressure to which the diffusion was proportional was 
This is 
