150 HARVEY CARR 



nation, viz., that this order is a function of the strength of the 

 entrance attraction characteristic of the various cul de sacs. 

 It is admitted that other selective factors may also be effica- 

 cious; otherwise higher correlation values should have been 

 obtained. 



The validity of the above principle of explanation is sup- 

 ported by the data of table III, giving the correlation values 

 between the number of errors for the various cul de sacs and 

 their spatial order in the maze. A positive correlation obtains 

 for the six mazes I-a, I-b, I-c, Il-a, III, and IV. In these cases 

 the animals entered the various cul de sacs with a frequency 

 roughly proportionate to their proximity to the point of entrance, 

 and it was for these six mazes that a positive correlation was 

 found between order of elimination and nearness to the food 

 box (table I). In other words the first cul de sacs were entered 

 the most frequently and were the last to be eliminated. In 

 mazes I-d, I-e, and Il-b, on the other hand, the animals for some 

 reason entered the last cul de sacs more frequently than the 

 initial ones (negative correlation between number of errors and 

 nearness to entrance), and in these cases the final errors were 

 the last to be eliminated (table I). 



TABLE III 



Correlations Between Nearness to Entrance and Number of 

 Errors at Different Stages of Learning 



Trials I-a I-b I-c I-d I-e Il-a Il-b III IV 



1 322 .929 .822 -.964 -.107 .000 -.607 .846 .750 



2-3 250 .679 .036 -.928 -.107 -.392 -.214 .709 .666 



4-5 232 .143 .215 -.321 -.750 .143 -.892 .373 .800 



1-5 322 .715 .679 -.928 -.214 -.107 -.607 .846 .700 



6-10 286 .786 .643 -.250 -.500 .465 -.107 .573 .733 



11-until learned.. .215 .250 .179 -.285 -.071 .536 .393 .500 .417 



It is thus the phenomenon of error distribution in the maze 

 that demands explanation in order to comprehend the order of 

 elimination. Our results enable us to offer but few explanatory 

 suggestions in regard to error distribution. 



1. The persistent tendency for rats to keep returning to the 

 point of entrance after exploratory excursions operates to in- 

 crease the number of entrances into the initial cul de sacs. This 

 returning tendency is well known, and it is at once obvious that 

 these returns must result in repeated explorations of the initial 



