EXPERIMENTAL EVOLUTION. 



115 



"The speed of running (centimeters per second) does not give any clear 

 difference between the tests and controls as to the general rate of movement. 

 The males and females in two strains show the speed of the tests less than 

 that of the controls, but in the third strain the speed of the tests is greater 

 than that of the controls. None of these differences is so great that it may 

 not reasonably be due to chance; in no case is the difference as great as 

 three times its probable error. The conclusions to be drawn, then, are that 

 (1) there is no general difference in the nature of the two series of rats as 

 expressed in the rate of their movements; (2) there is a difference that retards 

 the learning of the test rats. 



"Perfect trials, or those made without wrong turns, afford a clean-cut 

 method of comparing the tests and controls. When the number of perfect 

 trials made in the total number of trials by each rat was considered, the 

 following results were found (table 2) : 



Table 2. — Numbers of perfect trials. 



" It is plain that the tests made fewer perfect trials in the given number and 

 that the difference is great enough to be significant. The second use of the 

 perfect trials is to compare the number of trials before the first perfect one was 

 made. The averages obtained were: Tests, 19.48; controls, 13.82; difference, 

 5.66 ± 1.66; D/P.E., 3.41. 



"Here again the tests are inferior, taking more trials before learning to 

 make the first perfect one. In this case also the difference is too great to be 

 due to chance alone. The third criterion based on perfect trials is the time 

 spent in running them; the final averages (in sees.) are as follows: Tests, 8.48; 

 controls, 7.52; difference, -0.95 ±0.25; D/P.E., 3.84. 



"Although there does not seem to be a general difference in the speed of the 

 tests and controls, there does appear to be a significant difference in the time 

 that the two sets required in running their perfect trials, the controls going 

 faster. 



"We believe that the above points show that the tests and controls differ 

 as groups in their behavior in the maze. From the standpoint of learning 

 their way to the center and going there for food, the tests are less successful 

 than the controls. The alcohohc treatment of the grandparents is the only 

 basis upon which the rats have been divided into the groups of tests and 

 controls; therefore the alcohohc treatment of the grandparents seems to be 

 responsible for the inferiority of the tests in running the maze." 



"Maze-behavior of the children of alcoholized rats. — Since the extensive study 

 of the generation described above has shown that the various criteria for the 

 comparison of the tests and controls give closely similar results, it has seemed 

 necessary to use only part of the available criteria for the other rats. There- 

 fore, for the other generations we decided to employ only time and the three 

 criteria dependent on perfect trials, omitting distance, speed, and errors. 



"During the summer Miss Charlotte Oilman has summarized the data on 

 the maze-behavior of the children of the alcoholized rats. The parents of the 

 rats considered above were of this generation. Although these results are 

 still subject to certain further checkings, they are presented at this time for 

 comparison with the other generation. It may be stated at once that the com- 



