64 RUTLEDGE T. WILTBANK 



Looking at the matter from the point of view of the 

 various mazes, rather than from that of the various groups, 

 it was found that as far as the trial-criterion is concerned 

 increase in the number of previous maze-learnings did not 

 always result in an increase of positive transfer. According 

 to the error-criterion, in the case of but two mazes did the 

 favorable effect of prior learnings increase regularly. 

 There was no maze in which the favorable effect according 

 to the time-criterion increased regularly with the increase 

 in the number of prior learnings. 



It was found that a positive correlation existed between 

 the savings in trials and errors, in trials and time, and in 

 errors and time, though a series of mazes, estimated on 

 the basis of the average number of trials, the average 

 total errors and the average total time per rat; and that 

 a positive correlation existed among the savings when 

 they were estimated on the basis of the average errors 

 and average time per trial, except in three instances. 



The transfer-effect as between two mazes when the first 

 is only partially learned did not become positive according 

 to all three criteria until the transfer following the sixteenth 

 trial upon the former maze, transfers having been effected 

 after the second, fourth, eighth and sixteenth trials. The 

 transfer according to the time criterion was positive after 

 the change of mazes following two trials, and remained 

 positive throughout the series, and nearly uniform when 

 computed on the basis of the average time per trial. But 

 when it was computed on the basis of the average surplus 

 time per rat, it was both positive and cumulative. 



When a transfer was made from a maze completely 

 learned to one already partially learned, and the learning 

 of the latter then completed, the latter maze was learned 

 with a saving in trials when the partial learning consisted 

 of two, four or eight trials, but the transfer was negative 

 when the partial learning consisted of sixteen trials. The 

 transfer of training, as shown in the average total number 

 of errors per rat, was positive when the partial learning 

 consisted of two or four trials in D, and negative when 

 it consisted of eight or sixteen. It was positive when 

 it consisted of two or eight trials in E, and negative when 



