280 E. C. MACDOWELL AND E. M. VICARI 



DISCUSSION 



Although no general discussion of the literature upon experi- 

 mental alcoholism is presented in this paper, reference must be 

 made to one paper on account of the very close similarity of the 

 present subject with some of the work therein described. Arlitt, 

 '19 (pp. 41-44), has presented the results of training in a maze, 

 the grandchildren of alcoholized white rats. The alcohol was 

 administered with the food; doses of different sizes were given 

 for different periods to various sets of animals obtained from 

 a dealer. In order to eliminate strain differences, these sets were 

 chosen at random; animals from the same source were used as 

 controls. Three sets of rats are given in the third generation 

 (the grandchildren of alcoholics b}^ alcoholics) that are parallel 

 to the ones we have described in this paper. Four of the nine 

 grandchildren of rats that received 0.5 cc. of 90 per cent alcohol 

 per day for five months did not succeed in making a single trip 

 through the maze, so they are not included in the averages; 

 the other five in this group were inferior to normals by each of 

 the three criteria: total time spent, total number of errors made, 

 and number of trials required to attain a certain degree of per- 

 fection in running the maze. The eleven grandchildren of rats 

 that were given 0.25 cc. per day for thirty-nine days were inferior 

 to normals in the time spent, and practically equal to normals 

 in the number of errors and the number of trials required to 

 master the maze. The eleven grandchildren of rats that were 

 given 0.25 cc. per day for four months spent more time than the 

 normal, but made fewer errors and learned the maze in fewer 

 trials than the normals. These results are based on comparisons 

 with five normal rats by means of group averages unchecked by 

 probable errors of the averages or of the differences. 



The irregularity in the results leads us to suspect two things: 

 1) that the differences between the averages do not have signifi- 

 cance directly in proportion to their size, and that some of the 

 differences observed would not be significant at all if compared 

 with their probable errors; 2) that, although random selection 

 of the rats to be treated and to be used as controls maA^ have 

 eliminated strain differences in the generation treated with 



