122 ANIMAL LIFE AND SOCIAL GROWTH 



tional tanks of the same size. In all cases where 

 this experiment has been tried, the isolated fishes 

 learn to come through the opening more slowly 

 on the average than do the grouped fishes; the twos 

 react more slowly than do the fours but there is no 

 essential difference between the larger groups 

 tested. In all cases, once the fishes have learned 

 this simple maze, they react more steadily if a 

 small group is present than if the individuals are 

 isolated. These are, so far as I know, the first 

 experiments to be reported upon the effect of 

 class size upon the rate of learning in a school of 

 fishes and they show quite clearly that in non- 

 aggregating goldfishes, up to a given number at 

 least, the group learns more rapidly than does a 

 single animal. 



While this result is uniformly obtained with a 

 set-up such as has just been described, it does not 

 follow that the same results are to be expected 

 from all sorts of experiments upon the effect of 

 numbers of fishes present upon the rate of learn- 

 ing. In our earlier experiments on this subject, 

 mud minnows were trained to jump out of water 

 about a half inch in order to obtain a bit of worm 

 held there under a red light. If four fishes were 

 present the number of responses was much re- 

 duced, in fact, at times, no jumping would occur. 

 The isolated fishes readily learned this problem 



