34 CORA D. REEVES 



of discrimination is thus shown. This return of discrim- 

 ination is probably due to the failure to secure food, which 

 has made the fish hungry and alert. With further reduc- 

 tion in the intensity of the red, discrimination is main- 

 tained in varying degree by all the fish. Discrimination 

 continues when slit widths of red are reduced to 0.6 mm., 

 0-.4 mm., and 0.2 mm., and finally when the interposition 

 of the photographic negative has made the red so dull 

 that it has scarcely any color value for the human eye. 

 That the difference in distribution of light over the plates 

 does not control the behavior of the fish appears from the 

 results. For it is at slit width of 35 mm. for both plates 

 that discrimination is easiest and at that width the light 

 distribution is identical for the two plates. On the other 

 hand discrimination is more difficult at slit width 5. to 

 0.9 mm. for the red plate and in this region there is a dif- 

 ference varying from 0.1 per cent to 10 per cent in the 

 relative intensities of center and border in the variable 

 plate. The ease or difficulty of discrimination does not 

 vary with variation in illumination of the variable plate 

 (Note the graph of fish H.) It is shown elsewhere that 

 the dace did not learn in my own experiments to dis- 

 criminate intensity differences less than 1 :4. The most 

 accurate brightness discrimination found by any worker is 

 that reported by Hess (1909) in which differential response 

 was obtained to two white plates illuminated as 1.0 to 1.23. 

 When the graphs of the four dace are compared, it is at 

 once evident that they are not alike. Identity is not to 

 be expected since fish differ in the degree to which they 

 are influenced by external stimuli, in readiness of response, 

 and perhaps in ability to discriminate wave-lengths. If 

 we compare first the ' red maximum ' part of the curves, 

 it is seen that the initial fall is more gradual in Md and H 

 than in the other two fish. In both of them it involves 

 100 trials as compared with 20 in YP and 17, and in the 

 case of H there is no descent of the curve for the- first 40 

 trials. Both H and Md had to be taken from the aquar- 

 ium during the first part of this series of trials in order 

 to perfect the adjustment of the apparatus. This prob- 

 ably accounts for the greater number of trials required 



