82 CORA D. REEVES 



The training procedure used in my own work belongs 

 in the third category outlined. It resembles the procedure 

 of Frisch with the colored and gray tubes, in that a posi- 

 tive stimulus of constant intensity is presented to the fish 

 along with a variable. In Frisch's work the constant was 

 a colored tube, while the variable was a series of gray or 

 colored tubes, all of which were shown at the same time 

 with the constant. In my own experiments the constant 

 is a patch of blue light, the variable is a series of like patches 

 of red light of different intensities shown one at a time 

 in contrast to the constant. The intensity of the illumina- 

 tion used for the patches is known, and in the case of the 

 variable the intensities form a series the members of which 

 are close together. At one point in this series, usually at 

 the intensity of the variable corresponding to slit width 

 0.9 miji., the fish temporarily fail to discriminate (cf. 

 graphs figs. 13 and 15). This failure is believed to indi- 

 cate the region of matched brightness of the two stimuli 

 for the fish. If this is true, the fish themselves show by 

 their behavior when the colored patches are matched in 

 brightness for them. As the variable is held at this matched 

 brightness, the fish again come to discriminate the two 

 patches. The conclusion seems inescapable that the quality 

 of the light is the basis of the discrimination. The method 

 thus appears to meet the difficulties encountered in the 

 other training methods, in that it affords a behavior-indi- 

 cator for matched brightness. It has not hitherto been 

 used in training experiments on fish. 



c. Change of color in fish in response to background was 

 long ago shown by Pouchet (1876) to be response to stimu- 

 lation received through the eye. It does not occur in 

 blinded fish (see also Mast, 1916). The use of this method 

 to afford evidence of color vision in fish has been discussed. 

 Compared with the time required for discrimination of 

 colored stimuli by trained fish, a relatively very long time 

 is required for these color changes in the integument to 

 occur. While Mast's experiments with blinded flounders 

 show that the color changes are mediated through the eye, 

 Frisch has shown that in the minnow, Phoxinus, they are 

 influenced by illumination of the integument of the pineal 



