296 THE AMERICAN NATURALIST [Vol. LVI 



was started; on the first two days the doors were all 

 left open and food was exposed to view in every com- 

 partment; the rats in groups of five or so were left to 

 run at random in the apparatus. On the second two days 

 the front doors were all open as before, but the food was 

 concealed by covers fastened to the back doors, and when 

 a rat entered any compartment the food was revealed by 

 opening the back door; the rats were run singly on these 

 two days and given ten such feedings a day. On the 

 last two days of the preliminary training only the regular 

 series of doors were opened, but the rats were fed on 

 entering any compartment (20 trials). 



Right-hand Problem— In the first problem the rat was 

 fed only when it entered the right-hand compartment of 

 any set-up (those open in any trial) ; after wrong choices 

 the rat was confined in the compartment for half a min- 

 ute, and then, by raising the front door, was permitted 

 to make further choices (10 days, 100 trials); next, the 

 same problem was given with a different series of open 

 doors (2 days, 20 trials). Further training was given 

 in the form of a problem in which the correct door was 

 the open one at the left end of the open series, but the 

 results from this problem are so complicated that they 

 will not be treated at this time. The main reason for 

 this complication is the fact that at the end of the time 

 allotted for the master}^ of the first problem the test and 

 control rats exhibited different degrees of perfection; 

 some had made considerable progress in learning, while 

 others had made very little advance. Accordingly, when 

 the reverse problem was given, those that had learned 

 the most were handicapped by the habit already ac- 

 quired, while those that had not formed the required 

 habit in the first problem were able to progress more 

 rapidly in learning the second problem. 



Results. — From a study of the individual reaction ten- 

 dencies as revealed in the last two days of the prelimi- 

 nary training before the problem was presented, and in 

 the regular training after the presentation of the prob- 



