Diffusion of Hydrogen through Hot Platinum. 15 
tested by plotting log Q against log P. The result is shown 
on the accompanying diagram (fig. 4). The points marked 
with dots were obtained from the numbers in Tables I. and hh, 
Fig. 4 
Tae eae 
oeeeee ia 
_ DSS Reeee sa 
eee eae ae 
ee ree i 
ERED ae pil 
peal aa) ¥ 
Seale of log a 
SSE 
Es. 
bas 
= 
ARES Kg eapelles | 
eee he ae) | 
94 
SerlOme-2n-4n 6 16.5) O0N)-2 50-4 0-8 7-6, 10. -2. -4 0). 8 
Scale of log P. 
those marked with crosses from another series of experiments 
which will be described later. The points on these curves 
represent absolutely every observation taken at the specified 
temperatures by the method which has been described above. 
It will be seen that they all fall, within the experimental 
error, on straight lines whose slope appears to be independent 
of the temperatur e. The rate of diffusion is therefore a power 
of the pressure which may be obtained from the slope of the 
lines in question. The magnitude of this power for the 
various temperatures is: for é= WbAG" Sols} 9997)) 520s 
8= 856°, 500; and 0=717°, 508. The mean of all Oe Gies 
is 15. Since none of the values differs from *5 by more 
than the expected error, this appears to be the true value ; 
the fact of the average being 3 per cent. above this is prob- 
ably due to some constant source of error which has not been 
corrected for. 
To the order of accuracy of our experiments then we 
conclude that the pressure variation of the diffusion is given 
by a formula of type Q=AP®, where A is ide eae of 
pressure. This is not in agreement with the result of Win- 
kelmann, who makes the diffusion proportional to the concen- 
tration of the dissociated hydrogen outside, the amount of 
