68 Mr. J. S. Dow on the Physiological 



This would explain why the Purkinje effect was relatively 

 insignificant, and why, with a flicker instrument, the yellow- 

 spot effect, as shown in fig. 3, was less marked. In addition 

 it might throw some light upon the fact that flicker instru- 

 ments, however effective at ordinary high illuminations, do 

 seem very difficult to use satisfactorily when the illumination 

 is very low. 



The appearance of the field of view of a photometer of the 

 type shown in fig. 2, as seen under certain conditions, also 

 seems to throw some light on the suggested behaviour of the 

 rods. At low illuminations something like the following 

 experience may sometimes be noticed. When comparing red 

 and green, we may set the photometer so that the coloured 

 flickerless surfaces on either side of the band of flicker appear 

 about equally bright. The band of flicker in between will 

 then be grey in colour, and probably, at the correct speed, 

 the flicker can just be made to disappear. 



If now the photometer is moved towards the red light the 

 flicker reappears, and at the same time the red surface on one 

 side of the intermediate band of flicker becomes obviously 

 brighter than the green surface on the other ; the flicker band 

 meanwhile changes from grey to a reddish tinge, and soon 

 also becomes unmistakably brighter than the adjacent green. 



But if we move the photometer towards the green surface 

 we sometimes have a different experience. We again 

 observe a distinct increase in brightness of the green, and 

 we notice, as before, that the central grey band becomes 

 green in tint and brighter than the adjacent red surface. 

 But this gain in brightness is not accompanied by flicker; 

 apparently the only effect is to superimpose a steady luminous 

 impression produced by rod-impulses which succeed each other 

 too rapidly to produce a flicker-sensation. 



In short, it appears as though the speed which suited the 

 production of flicker due to the red illumination was too 

 high to do so in the case of the green illumination. 



The effect (which, however, was only found to occur at 

 very low illuminations, presumably on the borderland between 

 rod- and cone-vision, and was not always easy to reproduce) 

 seems to afford additional evidence in favour of the suggestion 

 that the speed which is adapted to the luminous impressions 

 received through the cones may be sometimes too high for 

 those received through the rods. 



It might, however, be well to guard against one possible 

 source of misunderstanding as to the suggestion implied 

 above regarding the Purkinje effect. In suggesting that the 

 flicker photometer is less subject to the Purkinje effect than 



