88 Prof. F. Allen on the Persistence of 



and my own, it would appear that the lower portions of the 

 curves for X'AtiQ jjl, X *460 yit, and A '430^ are all nearly 

 horizontal, though probably only one is exactly so. My 

 curve for A'430yu, shows a slight upward tilt of the lower 

 part, though the data are too scanty to give a reliable 

 conclusion. 



In regard to the meaning of the abrupt change of direction 

 shown by the curves, there is probably no reason to question 

 the interpretation based on the duplicity theory of von Kries*, 

 that the upper part relates to vision by the retinal cones, 

 the lower part by the rods, and the sharp change in 

 direction marks the sudden transition from the one to the 

 other. 



It is evident from the curves that the critical frequency 

 of flicker, or persistence of vision, is not simply dependent 

 on the luminosity alone, but that colours per se have peculiar 

 and independent effects of their own |. 



In those curves which have a portion horizontal it appears 

 that the effect of the diminishing intensity of the light is just 

 balanced by some restraining influence of the colour. 



The Purkinje phenomenon, or the displacement of the 

 brightest point of the spectrum towards the blue when the 

 intensity of the light is diminished, has been shown by 

 the curves in the preceding figures. The tables, however, 

 contain data which can be used to exhibit this important 

 phenomenon in a more specific manner. For each value of 

 cos 2 a there is a series of* relative values of the luminosity, L, 

 for the various spectral lines studied. Taking the red of 

 A, * 665 fi as a standard of reference, and plotting its lumi- 

 •nosities horizontally, the corresponding values of L for the 

 green of X'520/jl, the blue of X '4:60 fi, and the violet of 

 A,*430yu,, can be plotted as ordinates so as to give curves 

 of equivalent luminosities. If the luminosities of these 

 colours rise or fall at the same rate, the resulting curves 

 ought to be straight lines. The data from the preceding- 

 tables are arranged in Table III., and are displayed gra- 

 phically in fig. 5. 



The three curves in the figure alike show the relative 

 increase in brightness of the green, blue, and violet lights 

 compared with the red, since they deviate from straight 

 lines at the lower intensities. As the violet is approached it 

 is to be expected that the phenomenon should manifest itself 

 over an increasing portion of the curve, as it does in the 



* < Colour Vision,' J. H. Parsons, p. 208. 

 t Ibid., p. 98. 



