GROUP BEHAVIOR I49 



room performance on the part of the students; and I 

 know of none that does test these points adequately. 



Some of the difficulties which are inherent in ex- 

 perimentation on the effects of class size on the rate 

 of learning in man can be obviated by the use of 

 non-human animals. This procedure does not solve 

 all the requirements for elegant objective experimen- 

 tation, and has the additional real difficulty of elim- 

 inating all possibility of adding subjective impres- 

 sions to objective findings, a point which makes one 

 of the strongest arguments for experimentation on 

 man when feasible. 



In some respects the most completely controlled 

 experiments on the effect of numbers present on 

 the rate of learning are those that Miss Gates and I 

 performed some years ago, using common cock- 

 roaches as experimental animals. (52) Earlier work 

 by two independent investigators had shown that 

 cockroaches can be trained to run a simple maze, 

 and can show improvement from day to day. In our 

 experiments we found that the cockroaches could be 

 trained to run the maze we used by fifteen to twenty- 

 five successive trials on a given day, and showed defi- 

 nite improvement both in time taken to run the 

 maze and in number of errors. However, unlike the 

 experience of our predecessors, these University of 



