246 E. C. MACDOWELL AND E. M. VICARI 



summarizing the tests and controls have been compared. Stand- 

 ard deviations and probable errors have been calculated for all 

 groupings that included large enough numbers; a special set of 

 ratios to an assumed standard has been obtained for the purpose 

 of applying the x^ test to the frequency distributions of the 

 test and control data. Every method of treating the time 

 data leads to the same conclusion, namely, that the test rats as 

 a group differ from the controls; this difference is more apparent 

 in the first half of training while the learning was rapid, but even 

 in the last half of training, after there was very little general 

 improvement a similar though smaller difference is found. 

 Tested by probable errors, significant differences are found 

 between the averages involving the first half of training and the 

 fourth day of retention; the distributions of all the data for tests 

 and controls, compared by means of the x^ test, are found to 

 differ more widely than could be explained by random sampling. 



3. Comparison of the test and control rats on the basis of the distance 

 covered in running each trial 



a. Distance: averages for different groups of trials. As explained 

 in the first section, the data on distance were obtained from 

 camera-lucida drawings of the course of each trial of each rat. 

 As the rat went through the alleys of the maze, its image reflected 

 upon the record sheet was followed with a pencil, thus making 

 an accurate and lasting record of just where the rat w^nt. These 

 pencil lines were measured by means of a map measurer, or 

 chartometer. Figure 7 reproduces a record sheet; it bears an 

 inverted plan of the maze. These sheets are about l/13th 

 the size of the maze itself. In the early trials, when the rats 

 were going long distances, several sheets were used to record one 

 run. The data presented are in terms of the actual lengths in 

 centimeters of the pencil lines that appear on the record sheets; 

 accordingly, all the distances are l/13th of the actual distances 

 covered by the rats, but the relative difference between the 

 tests and controls is the same as though the actual distances 

 had been used. Leaving the data in the reduced form has saved 

 the labor of transforming each record into the actual distance, 

 as well as of handling much larger numbers in the summarizing. 



