262 E. C. MACDOWELL AND E. M. VICARI 



classes are 0.615 in width. The x" obtained from the two pairs 

 of curves are 66.49 and 84.78, both of which give values for P of 

 0.000,000. In other words, the odds against curves as different 

 as these occurring as random samples of the same population are 

 at least over 1,000,000 to 1. As far as this statistical test is 

 concerned, the distributions for tests and controls are proved 

 to be different. 



e. Summary of results from the distance data. When the test 

 and control rats are compared on the basis of the distance cov- 

 ered in running their trials on the maze, the tests are found to 

 go further. This has been shown in the summaries of the rats 

 in various groupings according to the sex and strain, when the 

 averages of each rat for six different groups of its trials are used 

 as the basis, and when the averages of the total distances covered 

 on each of the twelve days are used as the basis of comparison. 

 Tested by their probable errors, significant differences are found 

 for each period excepting the second half of training and the first 

 three days of retention. 



Apparently both sets of rats can completely master the maze, 

 but the tests take more trials to cut down the excess distance 

 than do the controls. It is a matter of the number of trials, 

 then, before all the excess distance of both sets of rats is elimi- 

 nated. However, when all the trials of the tests and of the con- 

 trols in the form of ratios to the corresponding points in an 

 assumed normal standard curve are classified in frequency dis- 

 tributions, the X" test indicates that there is a real difference in 

 the distributions of the test and control ratios. All these results 

 are in close agreement with those from the time data. The two 

 sets of data are derived from the same movements of the same 

 rats, but they differ in the mechanics of their origins; time from 

 readings of the stop-watch operated synchronously with the 

 rats' departures and arrivals, and distance from the readings 

 of the chartometer after tracing the pencil lines on the record 

 sheets. 



,/. Correlation between time and distance data. To test mathe- 

 matically the similarity between the time and distance criteria, 

 correlation coefficients have been calculated from the data of 



