the Photoelectric Effect. 583 



the sodium strip could be placed in position without coming 

 into contact with the air. But with the best care, the surface 

 of the sodium coated over so rapidly that it gave no photo- 

 electric current whatever. The plan finally adopted was to 

 place a little strip of sodium in position at S, heat the bulb so 

 as to drive off layers of adhering gas, and keep it under the 

 highest vacuum obtainable with the Gaede pump for about 24 

 hours. Then the pump connexion was sealed off, and the bulb 

 was left with a tube of phosphorus pentoxide and one of coco- 

 nut charcoal immersed in liquid air for several hours more. 

 After this a rod turning in a ground-glass joint provided 

 with a mercury trap was turned, thus winding up a wire 

 which drew a piece of Gillette razor-blade along so as to 

 leave a fresh surface exposed to the light. 



Even with these precautions the photoelectric sensitiveness 

 of the sodium fell off rapidly, decreasing from a deflexion of 

 3000 divisions per minute to about 50 per minute in the 

 course of two hours and a half. If photoelectric fatigue is 

 due, as seems most probable, to the chemical effects of the 

 surrounding atmosphere, it is clear that very minute amounts 

 of matter can produce very sensible effects of this kind. 

 The vacuum was unquestionably quite good in these experi- 

 ments. Before applying the liquid air the pressure recorded 

 by the McLeod gauge varied from amounts which were too 

 small to be estimated, up to about 0*00001 mm. 



As has been stated, we were not able to take the observa- 

 tions sufficiently quickly to obtain the complete velocity 

 distribution curves for sodium. However, the maximum 

 reading of the voltmeter at which any deflexion could be 

 detected was quite easy to find. Such points correspond to 

 the feet of the curves in PL XIII. fig. 2. These points were 

 obtained as soon as possible after the sodium had been shaved 

 and while the photoelectric currents were large, so that they 

 should be quite accurate. In this way the following values 

 were obtained : — 



Wave-length 43'6 36*6 31-3 25'4 21 



Apparent energy in 



equivalent volts... — 1-9 —1*5 -l'O —0*2 +0*5 



True energy in equi- 

 valent volts 0-6 1-0 1*5 2-3 3*0 



The true values of the kinetic energy were obtained by 

 taking the contact difference of potential between silver and 

 sodium to be 2*5 volts. This determines the position of the 

 maximum point in the curves like those in PI. XIII. fig. 3. 



2 Q2 



