450 GEETRUDE MAREAN WHITE 



neither fork contained food. The probabihty that the dis- 

 crimination was based upon himinosity was said to be lessened 

 by using a much hghter red in some of the tests. Blue forceps 

 were distinguished from red in the same way. Food was later 

 placed in the green forceps, and the fish learned to go to the 

 green first. 



According to Reighard ('08), the gray snapper can distinguish 

 colors. White Atherinas were taken in preference to those 

 that had been stained blue. Though blue, light green, and dark 

 green Atherinas were taken indiscriminately, red Atherinas 

 were snapped at least often. Red Atherinas with the tentacles 

 of the medusa, Cassiopea xamachana, sewed into their mouths, 

 were refused, and very soon all red Atherinas were refused 

 whether tentacled or not. This association of unpalatability 

 with red lasted without any further practice from July 19 to 

 August 8. 



The feeding experiments taken as a whole offer strong evi- 

 dence of a color sense in the fishes observed. The same criticism 

 may be made of all of them, however; pure colors were not 

 used, and the possibility exists that the fishes were reacting to 

 brightness. 



Hess ('09, '12, '13) is the chief opponent of color vision in 

 fishes, maintaining that fishes see colors only as shades of gray — 

 as a totally color-blind person perceives them. He offers the 

 following reasons for this view: 



In spectral light young Atherina hepsetus and young Squalus 

 cephalus congregated in the yellow-green and green, i.e., in the 

 brightest part of the spectrum of the dark adapted totally color- 

 blind human eye. Some were found in the green-blue, very few 

 in the red. By moving a black card along the spectrum and 

 intercepting certain rays, the fishes could be driven into the 

 blue or even into the violet, though they always returned when 

 the card was removed. The longer hght waves in the red pro- 

 duced no more effect upon the fishes than complete darkness. 

 The luminosity values of different parts of the spectrum were 

 worked out. This was done by lighting one-half of the tank 

 with monochromatic light and the other with equivalent mixed 



