436 Dr. F. G. Donnan on the Relative Rates of 



to re-examine the viscosity-relations of the apparatus. Re- 

 sults : — m. s. 



Oxygen 11 16 



Hydrogen 2 51*6 



Oxygen 11 16 



Calculated for hydrogen 



by densities .... 2 49* 7 



These results indicate the presence of a small viscosity- 

 effect. This was probably caused by the gradual narrowing 

 of the aperture, which is shown by the successive results for 

 oxygen. Now all the previous calculations have assumed the 

 absence of any disturbance due to viscosity. The best way to 

 determine their value is to re-calculate the experimental results, 

 introducing a correction for viscosity as determined by these 

 last measurements, i. e. introducing a maximum correction. 

 This could be done directly by means of the formula 



t = K\ \/d + k 2 fJ>, 



provided k x and tc 2 remained constant. But the results for 

 oxygen show that this was not the case. It is probable, 

 however, that any change in the size of the hole will chiefly 

 affect /t 1? since the outflow is mainly an " effusion " flow. 

 These considerations suggest the following method of making 

 the viscosity-correction. Calculate values for /c x and k 2 from 

 the last measurements for oxygen and hydrogen. Then in any 

 particular case correct the value of k x so obtained by assuming 

 that the change in k ± is proportional to the change in the 

 observed time of effusion for oxygen. That this mode of 

 procedure is justifiable is shown by the fact that the times 

 calculated for oxygen with the values of /Cj and k 2 so obtained 

 agree very closely with the observed times in each particular 

 case. 



The last results for oxygen and hydrogen give k x = 166*5, 

 k 2 =10. Accordingly the corrected time for CO (see p. 435) 

 is given by 



t = ( 166-5 x |^| x s/ u) + ('875 x 10), 



whence 



t = 10 m 32 s -3. 



The following short table will serve to show the extent to 

 which the correction affects the results : — 



