122 J. C. DeVOSS AND ROSE GANSON 



and exposed the top of the glass so that the cat could reach into 

 it or tip it over and secure the food which it contained. This 

 glass will be referred to as the "food-glass" or the stimulus 

 glass. The others will be called "confusion glasses," or confusion 

 colors. The latter term will often include a gray where the 

 context or the heading of the table has already specified a gray. 

 From the front of the apparatus the colored papers, which lined 

 the glasses made the only visible difference between them. 



As already stated, the Bradley colored papers were used, and 

 one blue cambric. After failing to find any confusions of the 

 stimulus color with Hering grays we tried the cats with any 

 gray we could obtain. The justification for this procedure is 

 that we finally found a gray which two cats confused with each 

 of the food-colors during at least six hundred trials. Only after 

 this had been done was its flicker equivalent determined for the 

 purpose of describing it. 



At the beginning of the tests of each cat each day several 

 glasses were cleaned and fitted with the necessary papers. The 

 experimenter then took a position at one side and in front of the 

 apparatus. Two glasses were displayed on adjacent levers and 

 the cat was allowed to approach the apparatus from in front. 

 If he went directly to the food-glass and drew it down, the choice 

 was recorded as correct. But if he touched the confusion-glass 

 ever so lightly, either with paws or head, the choice was called 

 incorrect. After making his choice, the cat was allowed to 

 discover the food-glass and secure food, but no record was made 

 of this act. The cat was then placed where he could not see 

 the levers while the glasses were being changed about. After 

 this precaution, he was started again from his former position. 



Thirty consecutive choices were considered a "series" and 

 twenty-four or more right choices were required in a series before 

 it was recorded as discrimination. If a cat consistently failed to 

 discriminate two colors for twenty such series (six hundred trials), 

 the result was recorded as complete confusion. In some instances 

 an animal failed to make a discrimination for eight or more 

 series, but showed by the low percentage of errors in each series 

 that it might learn to discriminate and eventually did choose 

 correctly twenty-four times in each of two consecutive series, 

 thus making a discrimination. These were called "difficult 

 discriminations." They doubtless indicate that the cats had to 



