346 JOSEPH PETERSON 



are small. Near the food end of the maze the entrances to 

 cul de sacs are fewer in number because (1) the probability is 

 small that the animals will get to this part of the trial without 

 making returns, and (2) there are no returns from the extreme 

 end of the maze due to the fact that the food box holds all rats 

 that reach this end. Every blind alley must be passed in the 

 forward direction at least as many times as the one nearest the 

 food box — usually most of them are passed many more times than 

 it is, — while this last one is never passed in the return direction 

 and hence is never entered from that direction. The other 

 several blind alleys are passed generally in an increasing number 

 of times as their distance from the entrance place in the maze 

 decreases. Exceptions to this statement will, of course, occur; 

 for example, Table I shows only one entrance to cul de sac 8 

 and two to 10. The law is a statistical law, dealing with prob- 

 abilities, and the results of sufficiently numerous cases should 

 conform to it. The result of this general law is, then, that 

 the act of running past the blind alleys becomes more prac- 

 ticed — is more frequently performed — than that of entering 

 them, and, to an even greater degree, than that of returning; 

 and that the proportion of this excess practice in following the 

 true path is greatest nearest the food box. 8 



This explanation, then, suggested by Carr 7 and later by 

 Watson, 8 but here for the first time worked out more fully and 

 applied also to the regressive elimination of blind alleys, seems 

 to explain admirably on the basis of frequency effects, taken 

 in a general way, (1) why entrances to the blind alleys nearest 

 to the food box should be eliminated first, and why, in general, 

 all cul de sacs should be eliminated in the order of their distance 

 from the food box; (2) why the number of entrances to the 

 several blind alleys should increase in the same order. 9 The 

 explanation would hold, of course, as would all the above reason- 

 ing, only on the condition that all other factors influencing the 

 choices at the various bifurcations of the maze — such as the 

 length of the cul de sacs and their general direction with rela- 

 tion to the true path — remained equal at the different points. 



6 This agrees with actual results obtained as published in the monograph re- 

 ferred to in the previous note. 



7 Carr, H. A. Psychol. Rev., 1914, 21, 161 f. 



8 Watson, J. B. Behavior, 1914, Ch. 7. 

 •To be considered on subsequent pages. 



