LIGHT OF DIFFERENT WAVE-LENGTHS BY FISH 33 



while in the third the red is duller than the blue. The 

 introduction of a negative to further reduce the intensity 

 of the red is also indicated at the top of the graphs. 



The general features of the entire graphs may first be 

 noted. At first there was avoidance of the bright red plate. 

 This shows in all the graphs in the high percentage of blue 

 choices at the beginning. After the initial avoidance, 

 there follows in the record of each fish a time when the 

 bright red and the blue were approached with nearly equal 

 frequency, as shown by the fall in the curves toward 50 

 per cent. This drop in the curves probably indicates that 

 the fish were becoming accustomed to the bright red so 

 that they no longer avoided it, while the association blue- 

 food had not yet been formed. After the curves have 

 descended they again rise while the red is still at maximum 

 value. The curves thus reach a maximum indicating 80- 

 100 per cent of correct or blue choices. Their rise is evi- 

 dence of the formation of the association blue-food. With 

 reduction of intensity of red the curves again descend to 

 a level indicating but 50 to 70 per cent of correct choices. 

 Thus as the brightness of the red patch is reduced to that 

 of the blue for the human dark-adapted eye (slit width 

 about 1.0 m), the ability of the fish to discriminate the two 

 diminishes, but in differing degrees in different individuals. 

 Failure to discriminate is evident in Md at a slit width of 

 about 1 mm. (trials 201-220). In Yl the failure comes 

 at slit width 5 mm. (trials 101-140). If no other trials 

 than these had been made at these slit widths the curves 

 would show for these two fish failure to discriminate at 

 a certain value of the red variable. It would be fair to 

 conclude that at this red value the red and blue were 

 matched in brightness for the fish and were indistinguish- 

 able by means of any quality difference. The entire 

 curves might then be interpreted in terms of brightness 

 differences in the two stimulus patches. But when the 

 fish failed to discriminate at a brightness of the red corre- 

 sponding to slit width 1.0 mm. to 0.9 mm., in the case of 

 Md and width 5 mm. in the case of Yl, the trials were 

 continued at these widths (trials 221-240 for Md. 121-160 

 for Yl). The curves of Md and Yl then rise, and a return 



