ALCOHOLISM AND BEHAVIOR OF WHITE RATS 229 



the lower end the controls are more numerous, at the upper end, 

 the tests. This is evidence that the means are good indices of 

 these two distributions; that the differences between the two 

 groups of rats, tests and controls, as given by their averages, 

 are real and not due to some peculiarity of the frequency 

 distributions. 



There can be no question but that there was a tendency for 

 the test rats to spend more time in learning the maze than did 

 the controls. That the differences in their averages is not 

 significant in retention and in the second half of training when the 

 'incompletes' and 'failures' are included, in no way weakens 

 this conclusion. The maze was simple enough to be learned 

 by all rats, and the number of trials given was great enough to 

 include overtraining for many of the rats ; that is, many rats were 

 continued after they had stopped improving their average time. 

 Longer training would probably have removed the difference 

 between the averages of the tests and controls. In the number 

 of trials given, the differences between the averages was enough 

 reduced in the last part of the period of training to fall within 

 the range of chance variation. The averages for each rat from 

 which the preceding summaries have been made are given in 

 table A in the appendix. 



h. Time: averages for each day. The next step in the analysis 

 of the time data is the comparison of the averages of the tests 

 and controls on each of the eight days of training and the four 

 days of retention. This series of calculations is based on the total 

 time spent each day by each rat in running its three trials. 

 Instead of using averages based on groups of 12, 24, or 36 trials, 

 the sums of the three trials on each day are now used. The 

 succession of the averages of these sums gives a curve (learning 

 curve) that shows the daily reduction in the time required to 

 make three successive trips through the maze. Such a curve 

 for all the strains and both sexes together is given in figure 3. 

 This curve shows that the tests (broken line) are above the con- 

 trols (solid line), that is, took more time on each day of training 

 and on all but the first day in retention. The light lines show the 

 tests and controls when the 'failure' rats are included. The 



