THE THRESHOLD OF VISION FOR DIFFERENT COLOURED LIGHTS. 
123 
dispersion and absorption of his instrument. Koenig’s* values (riven are those 
recalculated by Nutting, and in this case also the energy distribution was not directly! 
measured. Houston's! values are the mean of those obtained by a number of 
observers and the energy was again calculated. It is fairly evident that in this case 
some serious error has been made in determining the energy distribution. At first 
sight it looks as if the intensity of the spectrum used by Houston was so low 
that he was dealing mainly with rod vision. That this is not the true explanation is 
shown by the fact that he also publishes results obtained with a much less intense 
spectrum, and the maximum he obtains for this rod visibility curve is as much below 
the value obtained by other observers for the rod curve as is the maximum he obtains 
for the bright or cone visibility curve. § Our curve lies considerably below those 
obtained by other observers in the blue and violet. This is probably almost entirely 
due to the fact that we are dealing with pure' cone vision, while other observers, who 
apparently used a much larger photometric field, were probably dealing with a mixed 
* ‘ Bull. Bureau of Standards,’ 7, 238, 1911. 
t ‘Phil. Mag.,’ p. 715, May, 1913. 
+ The calculations of energy can only be considered as close appi-oximations of the radiation from 
peifectly black bodies. If not black deviation from absolute measurement of energy is to be expected. 
§ Displaced from the mean of all the other observers. 
VOL. CCXVI.—A. s 
