22 Messrs. Richardson, Nicol, and Parnell on the 
becomes materially simplified. In the first place, we may 
take our result to mean that the percentage of gas disso- 
ciated externally is negligibly small, and in the second place 
there is no experimental evidence of the existence of the 
term a proportional to the pressure. The formula for 
Q, therefore, which we require is simply 
1 Z k, 2 = ‘ 
Qe vic 
when pz is the coefficient of diffusion of the atomic hydrogen 
and &, the dissociation constant inside the platinum. ( is 
divided by 2 since it has been measured in grammes, whereas 
the right-hand side is in gramme- -molecules. In the original 
deduction of this formula A was not the inverse of the coeffi- 
cient of absorption of the undissociated hydrogen as usually 
defined *, but was the ratio of the internal and external con- 
ntrations ; ; under those circumstances P, was not the 
pressure but the concentration of the external gas. If we 
make P, the pressure and 1/A the coefficient of absorption, 
the ratio of the two will be the same as before and will be 
unaffected by changes of temperature: in order to distinguish 
the new meanings we will write the new quantities P and Ao 
respectively. Now the coefficient of diffusion ~. must evi- 
dently be proportional to the mean velocity of agitation of 
the molecules, and therefore varies as 62, where @ is the 
absolute temperature. It is rather more difficult to specify 
the way in which the solubility 1/A) of the undissociated 
hydrogen will vary with the temperature. It is not probable 
that the variation will be very rapid. At high temperatures 
it seems likely & prior? to resemble the case of a gas dis- 
solving in a liquid without chemical action at ordinary 
temperatures. 'or such cases it has been shown by Bohr + 
that the temperature variation of the absorption coefficient 
is given by 
a (0-0) =Ky 
@ being the absolute temperature and n and K_ constants. 
At high temperatures this becomes much the same as 
9/A,=K’, so that we shall assume that Ay varies as 0. 
Since both w, and A,? vary approximately as 62 it follows 
* Cf, Winkler, Ber. vol. xxii. p. 1773. 
_ + Wied. Ann. 1897, vol. 1xii. p.644. There are a number of mistakes 
in this paper owing to the sign — having been replaced by the sign +. 
