EXPERIMENTAL EVOLUTION. 121 



whatever differences previous^ existed between the tests and controls. More 

 test rats failed to show signs of learning, but those that did tend to profit by 

 exi:>erience appeared to have no handicap as compared with the controls. 



"The curves for the third criterion, the numbers of wrong choices made, 

 offer the following observations: The test and the control rats continuously- 

 reduced the numbers of wrong choices in the right-hand problem, but the 

 controls made fewer wrong choices. Considering the stronger right-hand 

 tendencj' with which the controls started out, this superiorit}^ can not be held 

 to prove greater ability in learning or in adaptability. The rate of elimination 

 of wrong compartments is much the same for the tests and controls. 



"The tests and controls both made more errors in the first 20 trials of the 

 left-hand training than they did in the beginning of the first problem; this is 

 evidence that there was real learning in the first problem which interfered with 

 the learning of the opposite problem. The second set of 20 trials in the left- 

 hand problem shows an abrupt reduction in the number of errors, but 

 in the following sets of trials errors are eUminated much more slowly. 

 During this time the controls have made more progress than the tests. In 

 this problem the fewer errors made by the controls can not be explained by 

 their original tendency, nor can it be due to any differential effect of the right- 

 hand training, since both tests and controls were doing about the same at 

 the end of that training. 



"Although the numbers of right and left hand end-compartments that were 

 chosen first in the left-hand problem do not indicate any improvement after 

 the first set of 20 trials, the number of wi'ong choices does seem to be reduced. 

 There are fewer correct choices, but also fewer wrong choices. This is mainly 

 due to the eHmination of repeated choices of the same door that had been a 

 correct door in the right-hand problem. 



"For the general failure of the training in the second problem, there does 

 not appear to be any obvious explanation. Learning of this iy^pe is possible; 

 it is demonstrated in the right-hand problem. The failure is not due to the 

 chance effect of combining rats which individually show very different results, 

 for only 4 rats showed anything that might be considered signs of learning 

 the left-hand problem. 



"The observations just made may be summarized as follows: Considering 

 the end-compartments that were chosen first, it appeared that (1) both tests 

 and controls increased the numbers of right-hand choices and decreased the 

 numbers of left-hand choices when food w^as given only in the right end- 

 compartment, whichever one that happened to be; (2) that both tests and con- 

 trols reversed the preponderance of their choices from right end-doors to left 

 end-doors in the first 20 trials after food was given only in the left end-com- 

 partment; but after the first 20 trials there was no further improvement in the 

 direction of training, but rather a tendency for the numbers of left-hand 

 choices to fall off and the numbers of right-hand choices to remain the 

 same; (3) that in the first problem the test may have a little advantage, as far 

 as there is any difference at all; in the second problem the controls appear to 

 show a greater degree of adaptability. Comparing the preliminary training 

 with the retention tests, the controls appear to have been more lastingly 

 modified than the tests by the training; but when certain rats that did not 

 appear to show any signs of learning at all were eliminated (5 controls and_8 

 tests), the advantage in fiivor of the controls in their greater adaptability in 

 the left-hand problem no longer is found. 



"There remains one valuable source of information that has hardly been 

 touched, namely, the study of the methods employed by incli\'idual rats in 

 meeting the situations presented, quite apart from their success as tested by 



