CoLH AND Long, Visual Discrunination in Raccoons. 667 



tests of raccoons with open feeding vessels seem to have just this 

 defect, we devised the apparatus which is shown in Fig. 1. 



Fig. 1. Color diseriniinatioii apparatus. 



In this device the feeding vessels were clamped up against the 

 cross-board so that the animal must select the glass at which he pulled 

 by its outside appearance only, and without being able cither to reach 

 into it or to look into it before it was selected. By means of the 

 thumb buttons at the rear of the apparatus every glass, except the 

 one containing food, was locked against the top board. A pull on this 

 glass depressed the short arm of the lever and exposed the top of the 

 glass. A pull, or even the slightest touch, on any other than the 

 food glass was recorded as a wrong choice. M^hen the animal had 

 secured the food, he was removed to the other end of the room, while 

 an assistant placed food in the glass and changed its position in the 

 row of glasses. The changes of position were made at random, ex- 

 cept when the animal had formed a habit of approaching one end 

 of the row. In that case we sometimes avoided putting the food glass 

 at that end, in order that it might not be the first glass approached. 



On the first day that w^e used this apparatus, Raccoon ISTo. 2 was 

 given 171 trials, and he selected the food-glass 122 times, thus giving 

 71 per cent of right choices. The last thirty-seven choices were all 

 correct. The next day, how^ever, this animal made but 35 per cent 

 of correct choices. (Our notes show that he was not hungry, and 

 often touched no-food glasses in passing along the row.) Below is 



