Yoakum, Behavior of Squirrels. 553 



the outside latch box, but did carry over scratching and gnawing 

 movements acquired in the sawdust box. Both animals apparently 

 found the task of breaking the habits acquired in the one or the other 

 of the earlier experiments almost insuperable. The slightest noise 

 or movement in the room, or the failure on their part to raise the 

 latch at the first pull on the string, would invariably drive them 

 to use some one or more of these previously acquired movements. 



The work of the untrained squirrels, ISTos. 1 and 4, was of a 

 much different type. '^Useless" movements were as much in evi- 

 dence, and their activity was at all times as great as that of !N^os, 

 2 and 3. However, the movements of 1 and 4 were random move- 

 ments, "useless" in the sense of not getting the animal nearer its 

 food, and further "useless" since they did not carry out any one par- 

 ticular line of attack, as did the movements of squirrels Nos. 2 and 

 3. Pulling the string, biting at the wire, running over the box, 

 etc., were random activities directed toward food. The scratching 

 and steady work at the door by the trained animals constituted 

 an activity which the most casual observer would judge to be directed 

 toward effecting a definite mode of entrance to the food box. 



An interesting point in learning to attend to the string came out 

 in the work of all four animals. At first, the string went unnoticed 

 or was subjected to the least possible scrutiny and the first pulling 

 was entirely accidental. At the seventh, eighth and ninth trials, the 

 string itself was singled out and became the point of the most eager 

 attack. When found, it was subjected to the severest kind of strains. 

 Hemp string no longer withstood their attacks and had to be replaced 

 by electric light cord, and finally by flexible steel wire. The associa- 

 tion between the string and the food supply had become definite, but 

 the movements used in pulling the string had not yet become auto- 

 matic. After the ninth and tenth trials, the unnecessary time and 

 energy spent on the string were gradually eliminated, until finally a 

 single pull on the string became the cue for a rush to the door. Fre- 

 quently in later trials, this pull was made so hastily that it failed 

 to open the door. 



(c) The Maze. — The method of conducting the experiments on 

 the maze is practically a repetition of that adopted in work upon the 



