Principles underlying the Flicker Photometer. 71 



Tt may be supposed that the impression received by the 

 cones would be the first to weaken, and therefore the duration 

 of undiminished impression, on which the point of balance of a 

 flicker photometer presumably depends, mainly depends upon 

 a purely cone-effect. But at a later stage the rod-impression 

 would presumably begin to die away were it allowed to 

 do so. 



The theory of the " recurrent image " investigated by 

 Shelford Bidwell in this country, von Kries in Germany, and 

 others, is very interesting in this connexion. According to 

 these observers, if a luminous surface, seen against a dark 

 background, is moved before the eyes, a second ghostly image 

 may, under suitable conditions, sometimes be seen, following 

 the main image at a constant distance away. This image 

 might presumably be explained by the later dying away of 

 the rod impression presupposed above. 



Now the recurrent image has been found by many observers 

 not to occur when the central portion of the retina (where 

 there are only cones) is used in vision, to be most easily seen 

 by greenish light, and to be entirely absent when red light is 

 used. All these facts strengthen the belief that this recurrent 

 image is due to the rods, for, let it be remembered, there are 

 believed to be no rods in the centre of the retina ; the rods 

 are also most sensitive to greenish-blue light, and practically 

 insensitive to red light (except under very exceptional 

 conditions). 



The subject of after-images is admittedly very complicated 

 and not to be explained on the basis of the rod-action alone. 

 Even here, however, the experiences of Prof. G. J. Burch *, 

 referred to previously seem to furnish possible evidence in 

 favour of the view that an image received through the rods 

 tends to be retained for a longer time than one received through 

 the cones. This investigator finds that, after remaining in a 

 darkroom for several hours, all the after-images constituting 

 " dazzle " disappear. During the' period of disappearance, 

 it is found that coloured images disappear in a regular suc- 

 cession, red being the first colour to vanish and violet the last. 

 This seems suggestive, for these colours are at the ends of 

 the spectrum to which the rods are least and most sensitive 

 respectively. 



Admitting, therefore, that there is a certain amount of 

 evidence in favour of the suggestion that the rods retain an 

 impression in an undiminished form longer than the cones, 

 and have therefore a lower critical speed, let us now deal 

 with several other facts which support the suggestion that 

 * Proc. Roy. Soc. London, 1905. 



