162 American Fisheries Society 



we may make the fish do this for us. After the fish has 

 learned to distinguish the blue plate from the red, we 

 feed him several hundred times before these two plates. 

 But after he has been fed ten times, we change the 

 brightness of the red plate. The fish still discriminates 

 the two plates. Again we change the brightness of the 

 red plate after ten feedings. Thus we continue changing 

 the red plate from time to time until we have tried red 

 of every brightness from very dull to very brilliant. 

 Amongst all the brightnesses of red that we have thus 

 tried there must be one that nearly or quite matches the 

 brightness of the blue. When this condition of matched 

 brightness for the two plates is reached the two colors 

 should appear to a color blind fish to be of the same 

 shade of gray. If the two plates appear to the fish as 

 indistinguishable grays he could not tell them apart and 

 he would then be unable to go regularly to the blue plate 

 for food. He would go as often to the red as to the blue 

 and would remain half the time unfed. If the fish then 

 were color-blind a point would be reached in our two 

 hundred experiments at which he would make mistakes 

 about half the time, and so fail to get food half the time. 

 If he were not color-blind he would continue to distin- 

 guish the blue from the red, no matter what the bright- 

 ness of the red, would make few or no mistakes and 

 would get food nearly every time. 



The fish in the experiments that I am describing 

 learned, after long training, to discriminate the blue and 

 red plates at all brightnesses of the red. Thus with plates 

 of the same size and shape and brightness and with the 

 possibility of distinguishing them by position eliminated, 

 we can conclude only that the fish discriminate between 

 the plates by their colors alone. 



Other methods than that of training with food have 

 been used in the study of color vision in fish. When they 

 are shown two lights of different colors and of presum- 

 ably the same brightness they often show difference in 

 behavior toward them. They are apt to be repelled by 

 red, but not by blue, and to approach a red plate in a 



