LIGHT OF DIFFERENT WAVE-LENGTHS BY FISH 



43 



slit was reached (tests 121-140). There was then, at first, 

 a very great drop in correct discrimination, but with the 

 slit held at this width the fish learned to go regularly to 

 the blue for food, and when tested further with the intensity 

 of the red lessened, showed no confusion. 



d. Sunfish SS2. — The small male fish, 552, was first 

 tested with the slit at 7 mm. (graph 55.2 in fig. 15). For a 

 time there was avoidance of this relatively weak red as 

 shown by the 70 per cent average of correct choices for 

 the first twenty trials. As the red was weakened, the 

 tendency to approach it is shown by a marked fall in the 

 percentage of correct choices at 2 mm. slit (trials 21-40), 

 but the learning of the food association followed (trials 

 41-60). The record then gives a curve of learning more 

 typical than that of the other fish (fig. 15, trials 21-160) , 



^^ ^^. / _- 



X-/0 2 



/s as as-rri^i 



7//>^e 



Fig. 16 



Graph made by averaging all the percentages of correct choice of the three sunfish 

 at the successive widths of slits indicated. 



with fluctuations as the red became duller, but no lack 

 of discrimination after the food association had been estab- 

 lished. When the negative already described was used 

 to further reduce the light, there was at first a few red 

 choices (trials 121-140); but, as with the other fish, there 

 was recovery, and discrimination became more accurate. 

 The percentage of correct choices was then 75 or over. 

 When the graphs (fig. 15), made by plotting the per- 

 centage of correct choices for each twenty tests for each 



