170 THE SOCIAL LIFE OF ANIMALS 



treatment was given the fish in the aquaria with 

 opaque partitions and those with clear glass. The 

 trained fish were then removed and those from the 

 small side chamber were gently transferred to the 

 larger side. An hour later they were given an ordi- 

 nary test such as had been given to the trained fish. 

 As is clearly shown by the graphs in Figure 38, the 

 fish which had been able to watch the others react 

 behaved decidedly more like trained fish than those 

 which had not been able to see their fellows perform. 



As a final check, the whole test was repeated, ex- 

 cept that no fish were placed in the larger side of 

 the aquarium. Fifteen times each aquarium was 

 lighted up, the door opened, and the experimenter 

 stood ready to feed any imaginary fish that might 

 come through. Then when those in the side passages 

 were transferred, there was no essential difference in 

 the behavior of the fish from the two types of 

 aquaria, and the experimenter was free from any sug- 

 gestion that he might have been signaling the fish. 



The results of these experiments suggest that there 

 is such a thing as imitation among goldfish. Whether 

 there is or not depends, as Dr. Welty rightly says, 

 largely upon the definition given to the word imita- 

 tion. These fish probably do imitate each other on 

 a relatively simple instinctive level. The untrained 

 fish that watched the reaction of their trained fel- 



