Mr Campbell, Discontinuities in Light Emission. 319 



rubidium cells made by the Polyphos Elek. Ges. of Munich : but, 

 though these were more sensitive than any made subsequently, 

 the differences between various cells were too great to permit of 

 their use. It should be noted that these cells, unlike those made 

 usually by the firm, were exhausted to the highest possible 

 vacuum. It is necessary in these experiments to avoid the 

 presence of any gas in the cells which might give rise to ' ionisa- 

 tion by collision,' for two reasons. (1) If there is ionisation by 

 collision it is very improbable that the fundamental assumption 

 of the theory given above, that the current through the cells is 

 simply proportional to the number of light disturbances incident 

 upon it, is fulfilled. (2) As was pointed out in the previous 

 paper*, if the current is not saturated, the effective value R of 

 the resistance between the quadrants of the electrometer varies 

 with the intensity of the light : there is a portion of the con- 

 ductivity which does not depend on the properties of the conductor 

 inserted specially, the magnitude of which it is almost impossible 

 to measure, and the constaacy of which cannot be assured. 



Cells were then made containing the alloy of sodium and 

 potassium. The form ultimately adopted is shown in fig. If. 

 After the vessel had been evacuated the alloy was sucked into 

 the bulb A. When the evacuation had been completed it was 

 allowed to flow into the vessel 5 by inclining the structure. 



The difficulties which attend the use of the alloy have been 

 discussed recently so fully by Elster and GeitelJ that little more 

 need be said. Of the two great difficulties which they found, the 

 distillation of the alloy on to the other electrode and the con- 

 tamination of its surface with particles of oxide, the former was 

 of no importance in this work and the latter of but slight 

 importance. The presence of oxide would only increase the value 

 of rf. But a method was found of getting rid of the small specks 

 of oxide by a process much simpler than filtering. If, when the 

 alloy is in its final position, one point at its edge is warmed 

 slightly (the neighbourhood of the hand is sufficient) the change 

 in the surface tension is sufficient to drive all impurities to the 

 opposite wall of the cell, leaving a perfectly clean surface. The 

 warm body must be kept in position while the experiment is in 

 progress, but the most careful experiments failed to show any 

 changes in the conductivity of the cells due to its presence. 



The cells were exhausted by charcoal immersed in liquid air 

 before being placed in position. A second tube of charcoal was 

 always kept in liquid air while observations were in progress. 



* Campbell, loc. cit. § 16. 



+ The tube joining A and B should have been drawn so that it joins B between 

 the copper plate and the surface of the alloy. 



+ Elster and Geitel, Pliys. Zeit. x. 1909, p. 457. 



