Measuring the Luminosity of the Spectrum, 



607 



chosen standard, which was ultimately the violet colour of 

 wave-length '414^. Since, however, this standard was of 



Fte. 2. 



■^c 



very low intensity, it need not be considered except in the 

 ends of the spectrum where the luminosity is small. The 

 accuracy of the measurements is influenced, especially in the 

 extreme red and violet, by stray white light, which was 

 unavoidably present in sufficient amount to influence the 

 measurements of the brightness of the spectrum in those 

 feebly luminous regions. 



The method as originally devised was to adjust the speed 

 of the sectored disk until the critical frequency of flicker of 

 the standard violet was reached. The disk was then to be 

 maintained constantly at this rate of rotation, and each colour 

 in succession reduced in intensity by rotating the polarizing 

 prism until critical frequency was reached. The luminosity 

 could then be determined as before. By this method the 

 auxiliary white standard could be dispensed with, and the 

 luminosity of the spectrum obtained directly. It was found, 

 however, impossible to maintain the motor at a sufficiently 

 constant speed, and the method modified as described in this 

 paper was substituted. 



