600 Mr. A. M. Tyndall on the 



A bright glow was also observed on the plate in pure 

 hydrogen. This might be explained either on the view that 

 the plate was bombarded by corpuscles or that the plate was 

 a source of back discharge : a corpuscular bombardment may 

 of course itself be a cause o£ back discharge. At any rate, 

 the glow was connected with the above phenomena in that it 

 gradually disappeared as oxygen was introduced. 



The folio wing argument reconciles the results with Franck's 

 theory : — 



In some work on the combination produced by point dis- 

 charge in hydrogen containing traces of oxygen *, evidence 

 was obtained for the theory that, at low percentages of 

 oxygen much of the oxygen was concentrated in films at the 

 electrode surfaces. Owing to the suggested affinity between 

 oxygen molecules aud negative electricity, the density of this 

 film at the cathode will be far greater than at the anode, and 

 will increase as the percentage of oxygen in the gas increases. 

 Now in negative discharge the negative ions which take part 

 in the discharge will be produced at the point, and if they 

 are ejected from the point surface they will pass through 

 this concentrated layer of oxygen. A certain number of 

 them will immediately take on oxygen and others will 

 escape uninfluenced. The percentage of the clustered to the 

 unclustered will depend on the density of the film, that is to 

 say, on the percentage of oxygen present. Once free of this 

 film they will continue their path without further change, 

 since it may be shown that for the percentages of oxygen 

 considered the chances of further collisions between ions and 

 oxygen molecules in the main body of the gas are very 

 remote. 



Since the clustered ions are wind producing and the un- 

 clustered are comparatively not, the value of ~ will increase 



with increasing percentages of oxygen. Also, since z for 

 these clustered ions is on this view very small, the resultant 

 z for the whole discharge will also decrease with increasing- 

 percentages. To explain the residual z of about 3 millimetres 

 in very impure hydrogen, when all the ions are clustered, 

 some such theory as the doublet theory suggested above is 

 necessary. The unclustered ion is, therefore, a special case 

 for negative discharge in oxygen-free gases only, 



Franck's theory thus offers a very satisfactory explanation 

 of all the wind-pressure results obtained in hydrogen con- 

 taining traces of oxygen. 



* Cliattock and Tyndall, Phil. Mag. [6] xvi. p. 24 (1908). 



