Mr Campbell, Discontinuities in Light Emission. 521 



It will be seen from Table I that the value of 0^' varies 

 apparently as some power of the intensity between the first and 

 the second, but considerably nearer to the first. This result 

 shows that, contrary to the hopes of the author, no evidence has 

 been produced against the " spherical wave " theory. Indeed the 

 fact that the power of the intensity is slightly greater than unity 

 might be held to be favourable to that theory as opposed to the 

 "bundle of energy" theory. But it must be remembered that all 

 sources of error tend to increase the fluctuations at the higher 

 intensities in comparison with those at the lower and so to give 

 a divergence from a proportionality between ^^^ and the intensity 

 in the direction observed. 



In the first place there is a possible error from small changes 

 in the magnitude of the constant " drift " of the position of balance, 

 which increases rapidly with the total intensity of the light falling 

 on the cell. The cause of the drift was never discovered. It was 

 always, on the average, in the same direction while the light was 

 acting, and may probably be due to some slight "photoelectric 

 fatigue," notwithstanding the high vacuum, which was different in 

 the two cells. 



In the second place the considerations mentioned in § 5 above 

 must be remembered. The finite area of the source of light is of 

 less importance when the source is further from the cell : hence 

 r]^ is likely to be somewhat less at greater distance. 



It is concluded, accordingly, that these experiments can make 

 no decision between the two theories : they can only say that the 

 " spherical wave theory " is inconsistent with a large value for « — 

 a conclusion which is to be expected on other grounds. 



§ 8. The second point requiring consideration is the variation 

 of the fluctuations with the instrumental constants. These con- 

 stants were altered in two ways : — (1) By changing the potential 

 of the electrometer needle and so altering at the same time the 

 electrometer constants a and b, the sensitiveness s and the 

 capacity G. (See previous paper.) (2) By changing the re- 

 sistance R. 



(1) Table II shows the calculated and observed values of the 

 ratios of ^y^ for different values of the potential on the electro- 

 meter needle, the resistance being kept constant during any one 

 set. a and b were never less than six times p : hence the ex- 

 pression (11') of the previous paper reduces without sensible 

 error to 



/. 



T ^2,2 1 



It will be observed that the agreement between calculation 

 and observation is quite as good as could be expected. Since the 



