248 'Journal of Cojnparative Neurology and Psychology. 



times without error and on July 11 he was still perfect although 

 without practice meanwhile. As No. 2 pursued one method 

 throughout, and No. i and No. 3 the other, it cannot be said that 

 our only crucial tests consist of reactions with the nose which are 

 so forced and unnatural as to be poor evidence. 



As to the establishment of a habit, the records are ambiguous. 

 No. 2 was as perfect after bemg put through one hundred times as 

 he would have been after eight or ten accidental successes. The 

 other two, with more experience, were less perfect. They did not 

 learn the toss I gave the lever, but expected food for raising it 

 only a trifle and letting it drop back. Withholding the food 

 brought the complete reaction. 



I must explain how the raccoons showed that they expected food 

 after the abortive performances. On raising the lever the animal 

 stood with his forepaws on top of the front board to be fed. After 

 every abortive effort he would take this position, then, as food was 

 not forthcoming, he would drop to the floor and dive under the 

 lever again with his nose. All the animals added to this reaction 

 the act of clawing the lever down into the horizontal position so 

 that it might be raised again. The experimenter merely had to 

 feed the animal each time the lever was raised, and the work thus 

 became very rapid. 



Let us summarize this long section: 



(1 ) All the raccoons began, of their own initiative, to run back 

 into boxes into which they had hitherto been lifted. 



(2) All learned to go up to the top of a box and drop through 

 a hole into the box after having been lifted into the box repeatedly. 



(3) All, after having learned to go to the end of a box, up a 

 step and thence to the top of the box, by being lifted through 

 these several stages of the ascent, learned to abbreviate the act by 

 climbingdirectlyupthefront togetto thehole in the top ofthebox. 



(4) No matter how well the animal had learned the through- 

 top reaction, if the front door, out of which he had just come, was 

 not closed behind him he would dodge back through that as the 

 quickest way to re-enter the box. 



(5) All four raccoons learned to undo a fastening by being put 

 through the act. They did not in general duplicate the act they 

 were put through, but neither did they in general duplicate the 

 act of their first success, or of their first three consecutive successes, 

 when these were attained by their own efforts. 



