36 CORA D. REEVES 



When we compare the curves of the fish Md and 17, in 

 which a distinct drop marks failure to discriminate, we 

 find that this failure does not occur at identical slit widths 

 In the two. It occurs about 5.0 mm. for 17 and at 0.9 mm. 

 for Md. It is probable that in this region of decreasing 

 red-brightness the exact point at which discrimination is 

 lost and recovered depends on the chance rushing of the 

 fish toward a red plate similar to the blue in brightness. 

 When this occurred, it was shown by a fall in the percent- 

 age of correct choices — a fall in the curve. The slit- width 

 was then held constant. As hunger increased through 

 failure to get food, it may be that the fish became more 

 alert and looked at the plates before it swam toward them. 

 The percentage of correct choices then increased. Ability 

 to discriminate between the plates was present at all these 

 slit widths, but became manifest only when hunger helped 

 the association blue-food to function in behavior. 



If we take the records of these four fish, more than eight 

 hundred in number, and plot the averages of correct choices 

 at the successive widths of the slit, the curve (fig. 14) shows 

 a region from 5.0 mm. to 0.9 mm. in which discrimination 

 is less accurate. This region includes the point wher'e the 

 blue and red are matched in brightness for the human 

 dark-adapted eye. But even here there is over 70% of 

 correct choice; while the average percentage of correct 

 . choices for the whole series is a little above eighty. The 

 individual curves show that at every intensity of the de- 

 creasing red the fish sooner or later discriminated. The 

 composite curve shows no region of very low discrimina- 

 tion. It appears to follow from these experiments that: 



(1) Dace do not easily learn to differentiate white 

 stimulus patches of considerable difference in intensity 

 (1 to 4). 



(2) They discriminate promptly between red and white 

 with variation in the intensity of the white, which in- 

 cludes a white intensity inducing a similar response to that 

 of the red used. 



(3) They discriminate between blue and red at all in- 

 tensities of red. While there is at first lack of accuracy m 

 response when these two colors were similar in brightness 



