THE ALTERNATION PROBLEM 



379 



given. As a consequence the percentage values for the inter- 

 vals of 50 to 95 seconds are based upon a total of 48 trials each. 

 The following results are apparent from an inspection of the 

 graph. 1. A gradual increase of the interval from 16.5 to 50 

 sec. exerts but little effect upon the accuracy of the act. The 

 lowest record of correct choices for any animal for two succes- 

 sive days' trials was 82%. Six of the animals were able to make 

 a record of 100% for a similar number of trials. 2. An increase 

 of the interval up to 44 sec. did not disturb the accuracy of the 

 act for normal conditions. A test for the normal time interval 

 was interpolated after the group was given the 44 sec. interval. 

 A group record of 96.5% was secured for a total of 288 trials. 



95 120 150 



Figure 8.— Graph 1, percentage of correct choices for group with increasing time 

 intervals. Graph 2, percentage of correct choices for group for large time 

 intervals and the introduction of new conditions during the delay. 



This value is not represented in the curve. 3. The number of 

 correct choices suffers after a period of one minute is reached. 

 This drop in the percentage values for the longer time intervals 

 is not due to a diminished hunger motive as the number of trials 

 per day was decreased from fourteen to six. The introduction 

 of the longer intervals decreased the percentage values for the 

 group about 10%. The lowest individual percentage record for 

 the eighteen trials devoted to the three large intervals was 

 80, while the highest was 100. The decrease in the values was 

 limited to five of the eight rats. 



The experiment was continued with somewhat different con- 

 ditions. After each trial the rats were allowed a few bites of 

 food and then were placed upon an adjacent table. At the 



