LIGHT OF DIFFERENT WAVE-LENGTHS BY FISH 85 



gives evidence that some fish can and do change color to 

 match the background upon which they are resting. Mast 

 presents pictures made by color photography of fish which 

 have become like the background. These were taken after 

 three months of adaptation. 



c. Errors may arise from inhibition of response. — Hess 

 (1911) shows that fish fail to eat when red light is thrown 

 upon them. Of his experiments in which a worm was 

 upon white sand on the bottom of the aquarium illuminated 

 by a Nernst lamp with a red glass filter in front of it, he 

 says, in effect, that more convincing than all else to the 

 lay mind is the fact that the fish left the worm undisturbed, 

 but on illuminating with blue light, the fish snatched the 

 worm at once though it was then less conspicuous to the 

 human eye. Here Hess observed the same response that 

 Bauer got and called " Rotscheu," but Hess misinterprets 

 and says that the red was of low illuminating value. He 

 adds that when brightness of the red was slowly increased, 

 the worm was seized and eaten. Many times my fish have 

 failed to feed upon a bit of white worm when it lay wrig- 

 gling on the black aquarium bottom in front of the colored 

 plates, but when I turned off the lights for the stimulus 

 plates and left only the general illumination from the 

 ceiling light the fish have fed at once, though the total 

 illumination was reduced. The fact was, the colored patches 

 so stimulated the fish that the feeding behavior was in- 

 hibited. When Hess slowly increased his red, it is not 

 stated how long the fish was allowed to become used to 

 the red illumination. It may be that the time required 

 for the slow increase in the red sufficed for the fish to be- 

 come used to it, if fear of red was originally present. 



d. Errors may arise through lack of training at matched 

 brightness. — When stimuli of different wave-lengths are 

 employed in training experiments, it is always essential 

 to vary the brightness of one of the two, for the purpose 

 of finding a point at which the two stimuli match in bright- 

 ness. If at this point discrimination fails, it may be argued 

 that the discrimination shown in other parts of the series 

 has been due to brightness difference and that such a series, 

 therefore, affords no evidence of color vision. If discrim- 



