246 'Journal of Comparative Neurology and Psychology. 



his position in the box, he will never vary from it. If not, he will 

 employ your act when his position makes it convenient and he is 

 looking at the latch you began with. He will also vary from it 

 very often but not a whit more often than a raccoon not put through 

 will vary from the act he seems to establish in his early trials. 

 Moreover, an animal may begin a new way sometimes after a hun- 

 dred or more trials, for example. No. 4 combined acts I and 6 in 

 Box 13 after several hundred trials; No. 3 combined them much 

 earher; No. 2 after mounting the platform in Box 13 many times 

 took to lifting it with both paws. When it was dropped the jerk 

 in addition to the weight of the platform would raise the bolt. 

 This was an awkward method and, while it occurred almost con- 

 secutively during three days' work and now and then for some 

 time longer, it was gradually relinquished. 



It would seem that enough experimental evidence has been 

 presented to show that the raccoons do learn without innervating 

 their own muscles. But the opposite condition as found by 

 Thorndike in cats, namely, that they learn by "trial and error" 

 only, has been made to support so important conclusions concern- 

 ing the mental life of animals, that I shall risk taxing the reader's 

 patience with a further recital of experimental tests. 



X N, 



a 



Fig. 2. 



In these experiments I used the card-showing device already 

 described, but I placed a lever holding a color on the front side 

 of the apparatus so that the animal might learn to lift it himself. 

 This could be done either by the nose or the paws. It was easiest 

 at the beginning of the ascent to raise the lever with the nose but 

 hard to elevate it thus completely. It required a vigorous toss 

 of the head to make the lever reach the point where it would not 

 fall back. On the other hand, while difficult to start with the 



