BEHAVIOR OF RACCOONS 249 



centage for the six preceding days was 87. Evidently there 

 is some connection between "lever order" and the ability to 

 react correctly. 



2. The normal order of colors and levers gave an accuracy 

 of 87% for 15 trials, which is a normal record. 



3. Mr. Hunter operated in the place of Mr. Gregg, and secured 

 an accuracy of 86% for 7 trials with the positive series and a 

 percentage of 89% for 9 trials with the negative series. 



4. Mr. Gregg operated, the levers were washed, fresh color 

 cards used, and the normal order of presentation was given. 

 Nine trials gave an accuracy of 89%. 



5. The red and white cards were again interchanged in posi- 

 tion, and the two series consisted of R-B-W, 1-2-3, an d R-R-R-, 

 1 -1 -1. The animal was fed whenever it chose to react to either 

 series. Both series stimulated positive responses. Discrimina- 

 tion was absent. There were but two inhibitions in 16 trials 

 and both of these were in response to the R-R-R series. 



6. The normal color and lever order was given and a dis- 

 criminative accuracy of 93% was secured in 15 trials. 



7. Mr. McPheeters operated in the place of Mr. Gregg, as 

 the latter still retained a suspicion that the animal might be 

 reacting to some difference in his method of giving the two 

 series. The normal color and lever order was adhered to. An 

 accuracy of 90% in 10 trials was secured. 



8. The normal lever order of 1-2-3, an d 3-3-3 was given, but 

 the positions of the cards were so shifted that the colors were 

 presented in the order cf R-W-B, and B-B-B. Jill was fed 

 only after response to the series R-W-B. Assuming that the 

 discrimination was based upon differences of "lever position," 

 this test gave an accuracy of discrimination of 90% in 20 trials. 

 Evidently the shift of presentation order of the cards did not 

 disturb the discriminative reactions in the least. 



9. Violet, green and yellow cards were substituted for the 

 white, blue and red cards respectively. The normal lever order 

 was adhered to. This change rendered both series different 

 from the normal in color and brightness. No disturbance in 

 behavior resulted. The two series were discriminated success- 

 fully in 86% of 21 trials. 



10. The cards were now removed and the bare levers were 

 presented in their normal order. The animal successfully dis- 



