112 



CARNEGIE INSTITUTION OF WASHINGTON. 



Experiment 20. 



Grandchildren of alcoholics — offspring from a mating in experiment 15, third generation; 



5 normal and 5 alcohoUzed rats. 



Table 1. — Results of training. Mutiple-choice problem. 



^ The percentage of correct choices is very slightly higher in case of the normal rats 



"In problem i the right-hand door is the correct door and in problem ii 

 the left-hand door is the correct door. Problem i is immediately followed by 

 problem ii. Both problems are of the same nature and on the same apparatus. 

 In order to form habit ii, habit i must be undone. As the normals learn 

 problem i better, they insist more than the alcoholic offspring on going to the 

 correct door of problem i when the problem is changed. The more fixed the 

 habit the harder it is to undo it, and this is what happens when the rats are 

 switched off from problem i to problem ii, the latter being the undoing of 

 the former. In training a rat from one problem to a distinctly new problem, 

 as from the maze to the trial-and-error problem, the habit is of a different 

 nature and as a rule those that did better in the maze do better in the multiple- 

 choice problem. For this reason the averages of problem ii (omitting the 

 first 3 days of training) are introduced. 



Table 2. — Number of rats trained August 1917 to September 1918. Total. 



First-generation offspring (alcoholized, 17; normal, 17) 34 



Second-generation offspring (alcoholized and from alcoholized parents, 37; from control 



parents, 33) 70 



Third-generation offspring (from alcoholized grandparents, 28; from control parents, 25) ... 53 

 Third-generation offspring (from alcoholized parents and alcoholized grandparents, 9; from 



control parents, 10) 19 



Total "l76 



