68 L. W. SACKETT 



30th for the series resulted in the first errorless trip. That the 

 time was not also minimum deserves a word, as there was some 

 hesitation at p but no turning back. The porcupine made the 

 start perfectly on the fourth, seventh, eighth, and tenth trials. 

 He missed blind alley 2 on all except the sixth, ninth, eleventh 

 and twelfth. This showed marked impro\-ement over the fourth 

 day's work. There was but one error made in the second half 

 on the fifth day, that was on the first excursion when the for- 

 mer confusion of / and r recurred. The average time for the 

 day was 2 minutes and 26 seconds, almost one -half the fourth 

 day's average with only four excursions requiring more than this 

 average and with another drop near the middle of the day's 

 curve. This may be thought to indicate that work was being 

 continued too long but the next day's experience does not 

 entirely justify such a conclusion. The average time for»tra vers- 

 ing the last half of the maze was 47 seconds with only two above 

 this average — the first trip requiring one minute and the tenth 

 requiring i minute 15 seconds. 



Sixth day. Experiments t,2> to 48 inclusive. The animal 

 was allowed to go through .the maze 1 6 times on this sixth day. 

 On the seventh trip for the day or the 39th for the series, he 

 again traversed the maze without an error, and in 10 successive 

 trials he made eiTorless trips. This had arbitrarily been classed 

 as the standard of perfection. It is the same standard which 

 Kinnaman used and so might furnish a basis for comparison 

 where comparisons are desirable. 



A glance at the curves in plate II, which are based upon 

 figures in the first two columns of table XIII, shows that the 

 learning curve of porcupine No. 6 with the maze is not unlike 

 the curve of No. 11 with the puzzle-box (plate I) and that both 

 conform closely to the curves of other animals in these and 

 similar tests. It is also noteworthy that the time and error 

 curves are generally parallel from which it may be inferred 

 that either variable is a fair measure of the animal's progress 

 in overcoming difficulties of this kind. This latter observation 

 is all the more important in view of the attack upon the single 

 variable method which Miss Hicks (22) has recently made. To 

 satisfy this new issue which has been raised, errors have been 

 recalculated so as to include not only the criteria mentioned 

 abo\'e, but also each straight small segment of the maze as a 



