224 



THE AMERICAN NATURALIST 



[Vol.L 



varied day by day. In case the way through the maze was 

 not found in 360 sec. the animal was removed and tested 

 again the following day. 360 sec. is thus the maximum 

 record for a single trial. 



Seventeen trials were made with each individual. This 

 was a desirable number for two reasons : first, because this 

 number was sufficient for the average mouse to learn the 

 maze, and secondly, because the seventeen trials could be 

 divided into three somewhat homogeneous groups. The 

 first two trials are largely affected by chance, so, although 

 given here for completeness, they are not averaged in the 

 final ratings for each individual. The second group of 

 five trials represents the period of more rapid learning, 

 the third group of ten trials the results when the learning 

 is slow or completed. In this paper the averages of the 

 last fifteen trials are used as the index of performance. 

 For some purposes the last ten or the last five trials might 

 be preferable. The rate of learning as determined by the 

 relation between the first and last groups may also prove 

 to be of value. 



In addition to the time, the number of errors, i. e., the 

 number of cases in which the mouse tried to go through 

 the locked gate, is given in the tables, as this is a measure 

 of the activity of the animal. With only a few exceptions, 

 however, the error and time curves correspond. Con- 

 sideration of the correlation between error and time, and 

 between performances at the beginning and the close of 

 the trials is postponed until data can be given for a larger 

 number of individuals and the records for other kinds of 

 behavior. 



In Tables I to V are given the complete records of the 

 mice tested, grouped in families as described below. 

 The average time is 44 sec. per trial for the last fifteen 

 trials. The distribution of the individuals is shown in 

 Fig. 2. In 41 cases the time was under 20 sec, in 19 cases 

 between 20 and 40 sec, in the remaining 30 cases between 

 40 and 200 sec. None of the mice failed to learn the maze. 



When the experiment started, several colored mice- 

 chocolate, agouti, gray, black and yellow-were tested, 



