﻿592 Mr. Karl T. Compton on Contact Difference of 



for the entire shift accompanying photo-electric fatigue ? 

 Or does the shift indictate a real change in the velocities of 

 the electrons? 



In investigating this question, the aluminium plate was 

 used for D and the nickel box for C. As has been previously 

 noted, this box had stood unpolished so long as to have 

 a practically constant electropositive character. 200 volts 

 was used between C and B. The aluminium plate was 

 polished only at the beginning of the series of observations. 

 Measurements were made of the contact difference of 

 potential between C and D, and readings for " distribution 

 of velocity " curves were taken at different intervals. The 

 results are shown in fig. 9 (PL IX.) and Table V. Curves 

 (a), (b), (c), and (d) were taken at intervals of 1, 7, 21, and 

 31 hours from the time of polishing. Curve (e) shows the 

 variation of the contact difference of potential with time. 

 Of course curves (a), (b), (c), and (d) do not mean that the 

 size of the current was the same in each case. It will be 

 remembered that the method of taking the readings causes 

 each curve to reach its maximum ordinate at 1. This enables 

 the shapes of the curves, and therefore the distributions of 

 velocities, to be compared. But actually each curve repre- 

 sents a smaller current than does the curve preceding it. 

 Curve (/) is plotted with time as the abscissa and the 

 largest deflexion obtained in each case as the ordinate. It 

 should give a rough indication of the rate of photo-electric 

 fatigue in the present case. 



Table V. 



Time since 

 polishing. 



Contact difference 

 of potential. 



Shift of 

 curves. 



Maximum 

 deflexion. 



1 br. 



■90 volt. 



•00 



300 



7 



•81 



•10 



152 



21 



•73 



•10 



148 



31 



•68 



•05 



97 



The shift of the curve as the time increases is in the 

 direction corresponding to the decrease in the electropositive 

 character of the aluminium plate, and the two are equal in 

 amount within the limits of experimental error. Thus it 

 seems that (at least when the plates have received no ab- 

 normal treatment such us making them the electrodes for an 



