ALCOHOLISM AND BEHAVIOR OF WHITE RATS 



273 



The controls eliminated errors of type 1, three trials sooner; 

 they eliminated errors of types 2 and 3a about one trial sooner 

 than the tests. The elimination of the other types of errors 

 is very slightly different for the tests and controls. Comparing 

 the numbers of test and control rats that did not eliminate each 

 type of error in the twenty-four trials in training, the following 

 numbers are found : 



In types 1 and 2, three and four more tests than controls failed 

 to eliminate the respective types of errors. Although the con- 

 trols eliminated errors of type 3a on the average sooner than the 

 tests, one of the control rats did not eliminate this type of error 

 at all. Otherwise, the numbers of rats that did eliminate the 

 respective types of errors and the rapidity in doing so fully agree 

 in favoring the controls. 



The comparisons based on the error data may be summarized 

 as follows: the tests made more errors; this difference was due 

 mainly to errors of types 1 and 2, as approximately equal numbers 

 of errors of the other two types were made by the tests and con- 

 trols; the tests were slower in eliminating errors of types 1 and 2, 

 but very little difference was found in the number of trials 

 required for the elimination of errors of the other types. The 

 variability in the number of errors is significantly greater for 

 the tests for two groups of trials (first half of training and omitting 

 first day) . 



THE JOCRNAL OF EXPERIMENT.\L ZOOLOGY, VOL. 33, NO. 1 



