140 K. S. LASHLEY 



A simple method of recording significant motor activities for 

 a study of this problem is offered by the graphic maze. 3 With 

 it records are obtainable of all the errors made by the animals 

 being trained and it is easy to determine whether or not any 

 of the errors persist through single periods of practice or from 

 day to day. If such persisting errors are the real cause of 

 the results obtained with different distributions of practice, one 

 should expect to find, on the average, fewer errors common to 

 the last trial of one day and the first trial of the succeeding day's 

 practice than to any two successive trials made on the same day. 



Comparison of the numbers of duplicate errors can not be 

 made directly owing to variations in the number of errors made 

 in different trials and the consequent difference in the prob- 

 ability of chance duplication. 4 Thus if nine out of a possible 

 twelve errors are made in each of two successive trials a larger 

 percentage of identical errors is to be expected from pure chance 

 than if only three errors are made in each trial. In making the 

 comparison I have used the records of all the errors made by 

 56 rats in learning the circular maze, when given five trials daily. 

 To avoid the influence of different numbers of errors in the 

 comparison of the number of duplicate errors appearing in single 

 and successive practice-periods the following method of com- 

 puting the results was adopted. 



The records of the last trial in each day's practice and of 

 the first in the succeeding one, where a total of three or more 

 errors appeared in the two trials, were compared and the num- 

 ber of errors in each, together with the number of duplicate and 

 diverse errors, was tabulated. The records of the animal which 

 furnished this data were then examined for a case of two suc- 

 cessive trials on the same day each of which contained a number 

 of errors equal to that in the corresponding trial of the pair 

 made on successive days. The first pair of trials meeting these 

 requirements was taken for comparison and its numbers of dupli- 

 cate and diverse errors were arranged in a separate table. Where 

 no pair could be found which met the requirements the succes- 

 sive trials in different practice-periods were discarded. 



3 Yerkes and Kellogg. A graphic method of recording maze reactions. Jour. 

 Animal Behav., 1914, 4, 50-55. 



Watson. A circular maze with camera lucida attachment. Ibid., 56-59. 



4 The methods of training animals and recording errors in the maze have been 

 described so frequently that they need not be discussed here. The reader who is 

 unfamiliar with the methods is referred to J. B. Watson, " Behavior," 1915. 



