8 RUTLEDGE T. WILTBANK 



of finding out whether there was a greater saving in this 

 identical part of maze B than in the other sections, the 

 distribution of errors within the four sections of maze B 

 made by the control-group and by the trial-group may 

 be compared. The average number of errors per rat made 

 by the control-group, group 2, in the four sections of B, 

 and the average number made by the trial-group, group 1, 

 which had already learned maze A, are given in table 2, 

 together with the percentages of savings made by group 1. 



TABLE 2 



The Average Number of Errors per Rat Made by Groups 2 and 1 in the 

 Sections of Maze B, and the Percentages of Savings 



Sec. 1 



Group 2 16 . 5 



Group 1 1.7 



Savings by Group 1 89 



While group 1 effected a saving in the first section of 

 89 per cent, the savings in the second and third sections 

 were still larger. It cannot be argued on the prima-facie 

 evidence, therefore, that the saving in the identical part 

 was due to the ta... "^^ this identity, when the savings in 

 sections dissimilar from the corresponding sections of maze 

 A were even larger. It might be expected that much con- 

 fusion would arise at the entrance of the second section, 

 inasmuch as the animal had gained undoubtedly as much 

 momentum in the first section of the second maze as in 

 the corresponding section of the first maze, and at the 

 entrance to the second section of the maze it encounters 

 quite a different pattern. Contrary to any preconception, 

 however, the group effected as great a saving in the second 

 section as in the first. In connection with this fact, it 

 may be noted — as has been shown by Webb^ and is shown 

 also in one of the experiments now being reported^ — that 

 an animal brings over from the learning of one maze to 

 the learning of another some factors making for positive 

 transfer and some making for negative. There is, accord- 



2 Webb: op. cit. p. 52. 



3 Infra, p. 18, ff. 



