26 CORA D. REEVES 



the one of wave-length ^. 4600 to ^ 5245, with maximum 

 in the blue; the other of wave-length from ^ 5890 to beyond 

 the end of the visible spectrum. The two filters were quite 

 different in the degree to which they interrupted the light. 

 When the stimulus patches were observed with the filters 

 in place but without the slit, the red was brilliant, while 

 the blue was somewhat dull but a saturated color. In 

 order to tell whether the illuminations used were effective, 

 dace that were tame but without experience in the experi- 

 ment aquarium were observed when put into compart- 

 ment 2 and presented with the stimulus patches. They 

 were not fed. At first one plate was illuminated with the 

 light passed through the blue filter while the other was 

 unilluminated. The straight approach to the blue as 

 compared with the zigzag swimming on the non-illuminated 

 side showed that the particular illumination was effec- 

 tive in modifying the behavior of the fish. Then with my 

 eyes dark-adapted I matched this blue and a red in bright- 

 ness by varying the width of a slit used with the red. The 

 red thus reduced was presented to the fish in contrast 

 with no-illumination. It was certain that the fish again 

 responded to the lighted area by definite straight approach 

 but moved in a zigzag course in the non-illuminated side. 

 When the slit was removed and the brilliant red presented, 

 while the other plate remained unilluminated, the approach 

 to the red area was in a straight line; but the fish came 

 only half the distance to the stimulus plate from the slide- 

 door, then turned suddenly across the aquarium into the 

 dark half, and retreated. Other tests showed that the 

 brilliant red gave strong stimulation. 



It was evident that both red and blue patches of light 

 were effective as stimuli for the fish, and that without the 

 slit the red stimulated more strongly than the blue. It 

 now remained to discover whether the fish could learn to 

 go to the blue for food, and whether at any intensity the 

 red would give the same stimulation as the blue. If the 

 fish could discriminate the red and blue patches at all 

 intensities of red, it must be concluded that discrimination 

 was on the basis of quality and not brightness. If, at any 

 intensity of red, discrimination of the two patches became 



