VIII. SUMMARY OF RESULTS 



The evidence furnished by these experiments supports 

 the position that the learning of one or more mazes is 

 usually beneficial with respect to the learning of another 

 following them immediately, as determined by the fewer 

 trials required and lesser time and smaller number of errors, 

 compared with those of the control-group. There were no 

 instances of negative transfer in either errors or time, 

 but there were two instances of negative transfer in trials. 



When two adjacent mazes contained identical parts, 

 the more expeditious learning of the second was not due 

 to the saving of errors in the identical part. Whether the 

 identical part exerted a favorable influence which was 

 effective in the saving of errors elsewhere than in the 

 identical part is a problem that these experiments did not 

 attempt to solve. 



Although the average savings are higher when the trans- 

 fer is from the more difficult to the less difficult maze, 

 there is no ground for the statement that the transfer 

 from the harder to the easier maze is in every instance 

 higher than from the easier to the harder. The transfer 

 from the easier to the harder may be greater. 



The transfer of training was found to be not only positive 

 but persistent, except in two instances of negative transfer 

 in trials, throughout the learning of five series of mazes by 

 a different group in each series. Although it was thus 

 persistent, it was not cumulative; that is to say, it did 

 not increase regularly with the increase in the number 

 of previous maze learnings. There was no case in which 

 the transfer, as determined by any one of the three criteria, 

 increased throughout an entire series; and if there were 

 a number of instances in which there was an increase for 

 three maze-learnings, there were more successive decreases 

 for three mazes than there were successive increases for 

 three. 



63 



