COLOR VISION IN FISHES 487 



but not conclusive. Here again Reeves improperly assumed that lights 

 equally bright for one animal would be equally bright for another, which 

 could be true only if their spectral luminosity curves were completely 

 superimposable or at least coincided with, or crossed, each other at the 

 particular wavelengths used. 



Since Reeves' time, all of the training experiments reported have been 

 made on Phoxinus Icevis and, to some extent, on Gasterosteus aculeatus. 

 Schiemenz, Wolff, Ktihn, and Hamburger, working from 1924 to 1926, 

 are the last prominent names in the literature to date. Spectral lights 

 were used ahnost exclusively, and the spectrum of the fish explored with 

 sufficient thoroughness to establish its limits roughly and to yield curves 

 of hue-discrimination — that is, graphs of the closeness of two just- 

 discriminable wavelengths plotted against their position in the spectrum. 



Light-adapted fishes were trained to jump for food held just above 

 the water on glass rods bathed with narrow spectral bands. They would 

 continue to seek food on empty rods in the training color even when 

 twelve intensities of it were interchanged at random, and were never 

 confused by any intensity of any other color except the training color's 

 immediate neighbors in the color circle. Lights too low in intensity to 

 be seen as colored by the human were the only ones confused by the fish. 

 Trained to seek food in a particular colored area among others — a 

 multiple version of Reeves' two-choice presentation — they were never 

 confused by the other colors regardless of intensity-relationships. These 

 species apparently give much more attention to hues than to brightnesses, 

 in contrast to Reeves' material. 



The animals' ability to discriminate hues close together in the spec- 

 trum was better in the short-wave end, a little poorer in the long-wave 

 region, than that of man. They could be trained to ultraviolet as far 

 as A,365m[X; and this region, violet, blue, green, yellow, and red were all 

 qualitatively different for the fish. When offered the whole spectrum 

 on the wall of the aquarium, they gathered in the particular region to 

 which they had been trained, and snapped the air seeking the accustomed 

 food. If the spectrum was moved, the fishes shifted with it. If the in- 

 tensity of the whole spectrum was lowered they still gathered in the 

 training color as long as it had color for the human, despite repeated 

 scattering by hand and shifts of the position of the spectrum. And they 

 could not be trained at all to particular brightnesses of white light. 



These high-school EUritze thoroughly dispose of Hess's contention 

 that all fish are color-blind. Although the English worker, Bull, has very 



