COLOR VISION IN FISHES 475 



tion maximum of a fish may be at a wavelength as high as 545 m(l, as 

 contrasted with the human value of about 510m[X. Should the photopic 

 maximum of the same fish happen to fall at the human value of A,557m^ 

 (and it probably would) the Purkinje shift would be only 12m|i, instead 

 of 47m[l; and a downward shift of the photopic maximum could even 

 bring the two peaks into coincidence. Too, the normal human retina is 

 totally color-blind in the extreme periphery, yet even here the brightness 

 values of chromatic stimuli are those characteristic of photopic, not of 

 scotopic, vision. Thus a Purkinje phenomenon occurs here in the absence 

 of color-vision. Human dichromates experience an inversion of the 

 relative brightnesses of red and green, upon a change of adaptation — 

 yet red and green, for them, are the same hue. They thus have an 

 'isochromatic' Purkinje phenomenon as compared with the 'heterochro- 

 matic' one of the normal trichromatic individual. 



The argument from pupilloscopic findings is even shakier; for while 

 in man the pupil is controlled reflexly from the retina and appears to 

 respond maximally to a given color because that color is consciously seen 

 as brightest, in the fish any iris muscles are entirely autonomous and there 

 is no reason to suppose that the wavelength which most stimulates them 

 will also maximally stimulate one or both sets of visual cells in the retina. 

 The teleost pupil moves but little at best, and in his examinations Hess 

 made no attempt to eliminate the passive effects of lens movements upon 

 its size. 



Hess worked largely with very young fishes, apparently in order to be 

 able to have large numbers (up to 60) in the same small tank, so that 

 their distribution in the spectrum thrown in the water would be devoid 

 of crowding-effects, and would also be statistically significant. For this he 

 has been taken severely to task, as also for making too few control tests 

 with thoroughly light-adapted specimens, for disturbing these before test- 

 ing by carrying them for some distance to a darkroom, and for ignoring 

 certain performances when they failed to confirm his ideas. 



With species after species, Hess found that the fry would usually 

 gather in the green or yellow-green portion of the proffered spectrum 

 (A,525-535m[x). He concluded that this region looked brightest to them, 

 since he claimed that they were always step-wise in their preferences for 

 white lights of different intensities. When pairs of spectral lights were 

 offered, the choices of the animals determined a curve of relative bright- 

 ness which simulated that of the scotopic human. Hess claimed to have 

 eliminated the possibility that this was caused by a Purkinje effect, by 



