152 VINNIE C. HICKS 



ing to the above assumptions, this curve should approximate 

 the difference between the time and error curves, or the differ- 

 ence between this curve and the time curve should correspond 

 closely to the error curve. This comparison is represented by 



TABLE I 



Average Times and Errors for Twelve Rats in a Maze with no 



CuL DE Sacs 



Trial Time Error Trial Time Error 



1 6.42 12.60 11 44 .00 



2 4.40 3.25 12 30 .00 



3. 

 4. 

 5. 

 6. 

 7. 

 8. 

 9. 

 10. 



figure 3 in which B is the error curve and A is a curve represent- 

 ing the difference between the time values obtained on mazes 

 with and without cul de sacs. No exact correspondence can be 

 expected in the present case inasmuch as the maze used by 

 the Misses Hybarger and Cowles is relatively simple in type 

 and the error curve represents more than the mere elimination 

 of blind alleys. 



For the above reasons we are forced to conclude that for our 

 conditions time is the best single criterion for an adequate rep- 

 resentation of all features of the learning process. 



In regard to the irregular variations, it is to be noted that 

 it is the distance curve which presents the greatest uniformity 

 and regularity of descent, and that there is very little difference 

 between the time and error curves in this respect. These irreg- 

 ularities probably reflect peculiarities of behavior due to sex, 

 age, individual characteristics, and disturbing conditions which 

 can never be wholly eliminated with the best of technique. 

 The instances of irregularity so often cited in the literature are 

 isolated exceptional cases so pronounced as to attract attention. 

 Certainly the variations exhibited by the curves are not so pro- 

 nounced as one would expect from these instances. The expla- 

 nation is probably to be found in the fact that these exceptional 

 cases tend to be minimized by the law of averages consequent 



