TRAINING IN WHITE RATS UPON VARIOUS SERIES OF MAZES 51 



and brought back to E after having learned D was for 

 the two trials 13.32 minutes, and for the remaining trials 

 necessary to learn E but .28 of a minute; but the average 

 number of trials necessary to complete the learning of 

 E was but 6.43. 



The negative transfer in errors after sixteen trials would 

 seem to be due to the fact that a nearly perfect habit is 

 interrupted in order to learn another maze; and the dis- 

 concerting nature of such a proceeding is denoted by the 

 increased number of trials necessary to learn the maze, 

 as compared with the control-group and by the increased 

 number of errors per trial. It would seem that this prin- 

 ciple is to be invoked to explain the negative transfer 

 also when the partial learning consisted of eight trials. 



The only criterion logically applicable here, when the 

 two entire group-performances are compared with each 

 other, is the trial-criterion, for as has been just indicated 

 the favorable effect of practise as it is seen in the error and 

 time-records of the group in the trials following its return 

 to the maze whose learning has been interrupted is swamped 

 by the high average made during the trials preceding the 

 interruption. Accordingly, when we compare the records 

 of the control-group with regard to errors and time with 

 the records of the group whose learning is interrupted, we 

 are not bringing together two learning processes which are 

 analogous in these respects. The transfer in errors and 

 in time, however, may be estimated by bringing into com- 

 parison the performances of the rats in the trial-group 

 upon their return to the first maze after having learned 

 the second with the performances of the rats in the control- 

 group for the first maze, beginning with that trial of the 

 control-group which corresponded with the first trial made 

 by the trial-group upon its return. Thus in the case of 

 the interruption after four trials, the first trial of the re- 

 turning group after having learned the intervening maze 

 would be its fifth trial upon the former maze; and the 

 transfer which it has effected from the maze learned during 

 the interruption may be estimated with reference to the 

 record of the control-group beginning with its fifth trial 

 and continuing to the end of the learning. In table 46 



