120 'Journal of Comparative Neurology and Psychology. 



size but painted a light red, of the same color-tone as the previous 

 red — so far as our discrimination went — but considerably brighter 

 to ordinary vision than the green, which was, it will be remembered 

 itself a little brighter than the red hitherto used. In all the 

 unbaited test experiments made on and later than the afternoon 

 of August 4, the following procedure was adopted: While the 

 dark red sticks were still used on the baited forceps in the experi- 

 ments where the fish was fed, when it was tested without bait we 

 removed the dark red sticks and substituted for them the green 

 sticks on the same forceps, using the same rubber bands that had 

 previously fastened the dark red ones. Thus all the apparatus that 

 could have the odor of food about it was now attached to the green 

 sticks. On the forceps that had previously carried the green sticks 

 were fastened, with the rubber bands that had been used for the 

 green, the pair of light red sticks. When the forceps thus arranged 

 were put into position, the fish had the choice between two unbaited 

 forks, one, the green, having about it whatever food odor was 

 present, the other, the light red, having in common with the pair 

 from which it was usually fed only shape, size and color, not 

 smell or brightness. If, then, the creature persisted in biting first 

 at the red pair, it would show that the impulse to bite was "asso- 

 ciated" with the color red, not with smell or brightness. It will 

 be seen from the table that in the twenty-five tests made under 

 these conditions the subject never once failed to bite first at the 

 red. 



One last, very remote possibility of error lay in the chance that 

 the green and red paints might have had different odors. Inas- 

 much as the sticks were all covered with the same varnish, the 

 chance was slight, but it was guarded against by putting, where 

 they would not shovv, daubs of green paint on the red sticks and 

 daubs of red paint on the green sticks. The results were wholly 

 unaffected by this precaution. 



In the work from August lo to August 12 inclusive, comprising 

 forty experiments in all, we substituted for the green sticks on 

 the empty forceps a pair painted a light blue, lighter than the 

 green and approximately equal in brightness to the hght red. 

 The last-named were still used and the forceps exchanged as 

 before in the unbaited test experiments, and food was placed in 

 the forceps carrying the dark red sticks in the ordinary feeding 

 experiments. The fish made not a single error in distinguishing 



