Washburn and Bentley, Color-Discrimination. 119 



marked that only eight experiments could be made, as it wholly 

 refused to rise after the eighth. 



The obvious source of error which would wholly invalidate the 

 results if they stood alone is that food w^as actually in the red for- 

 ceps and not in the green, so that both sight and smell might have 

 led the fish in the right direction. It demanded some care to 

 eliminate this error, for, on the one hand, it was impossible, with 

 the forceps fixed in the support, to have them both baited and, at 

 the same time, to prevent the fish's getting food from the green as 

 well as from the red pair; and, on the other hand, we could not 

 perform a large number of tests where neither fork should be baited 

 without weakening the association between red and food. We 

 adopted the plan of performing each day a certain number of 

 experiments, usually eight or ten, with the red fork baited and 

 then making two tests with both forks empty. If the fish's 

 appetite was good, we would give it two more tests with the 

 baited fork and finish with two "unbaited" tests, again. In this 

 way, without greatly weakening the association, we accumulated 

 forty-four experiments where both forks were empty. In two 

 only of these did our subject bite at the green. These two occurred 

 in the first four days of experimenting, and in the second case the 

 fish merely touched the green, then swam to the red and bit 

 vigorously. The results thus show that the sight of the red for- 

 ceps came to be connected with the impulse to bite, quite inde- 

 pendently of the sight of the food. As a matter of fact, the fish's 

 behavior throughout indicated that the sight of the food played 

 little part in setting ofi^ the biting impulse. It seldom bit directly 

 at the food, but nearly always at the ends or sides of the sticks, 

 and if by accident the food became detached and floated in the 

 water half an inch or so away, the chub still ignored the morsel 

 and bit persistently at the stick. 



The possibility that the smell of the food might have guided the 

 subject remained uneliminated by merely testing the animal with 

 unbaited forceps; for, since food was so often in the red pair and 

 never in the green, the odor of the food might be supposed to 

 linger about the former. Further, we still had the brightness error 

 to deal with. It was possible that the fish distinguished between 

 the red and the green forceps not as difi^erent in color but as dif- 

 ferent in brightness. To avoid this, we took a suggestion from 

 Graber, and prepared a pair of sticks exactly like the others in 



