﻿1)66 Dr. L. Silber stein and Mr. Trivelli on the 



to n = 0'286 per /a 2 or about 29 million darts per square 

 centimetre of the plate; but if, say, only one-thousandth of the 

 area of each grain were vulnerable, we should conclude that 

 29 milliards of darts were thrown upon each cm. 2 of the plate. 

 But it would be idle to speculate upon this subject and, as 

 far as we can see, the only way of deciding whether that 

 suggestion is correct or not and of determining the value of 

 the fraction e is to measure the exposure energy in absolute 

 units *. Now, in none of our experiments thus far reported 

 was the energy value of the exposure even roughly estimated, 

 not to say measured. But in order to decide this important 

 question, preparations for measurements of this kind are now 

 in progress in this laboratory, and their results will be 

 published in due time. 



5. Effect of finite breadth of size-classes of targets. — The 

 short name u target " will now be used for either a single 

 grain or a clump of grains in sufficient contact to act as a 

 photographic unit. 



In the three sets of observations hitherto reported, the 

 targets were classified according to the number of grains 

 contained in them (from 1 to 33), and for each class the 

 average size (area) was used as a in the theoretical formula, 

 without taking account of the finite breadth of any such 

 class, i. e. of the interval, a x to a 2 say, over which its indi- 

 viduals ranged. It was not possible with the said classi- 

 fication to secure reliable estimates of this breadth, which, 

 however, for some classes might have been considerable 

 (perhaps of the order 1/x 2 ), and at any rate varied from class 

 to class. It is likely that some of the outstanding dis- 

 crepancies are due to these neglected factors and especially 

 to the latter. 



To eliminate this source of error, and at the same time to 

 avoid the laborious planimetrization of targets within very 

 narrow limits, we propose henceforth to divide the whole 

 material of targets into deliberately broad classes, all of equal 

 breadtli, say 2a. 



If, then, the average size of any of these classes of targets 

 is used as the variable a in our formula, a correction has to 

 be made for the finite value of 2&. This correction can easily 

 be found. 



Disregarding for the moment the <7-term, the number of 

 targets of a class of breadth 2a.=.a 2 — a x affected by n darts, is 



* Although even then the final result would be made doubtful by the 

 uncertainty whether the total light energy (as required by Einstein) or 

 only a fraction of it is conveyed in discrete quantum parcels. 



