146 VINNIE C. HICKS 



This equation was performed as follows: Since 131^ seconds 

 represents the average time taken to run the maze correctly 

 (in the experiments described in this paper), all time values 

 above this represent the time consumed in eliminating errors. 

 Thirteen and one-third seconds was subtracted from the time 

 of each run and these remainders were summed, giving the total 

 surplus time eliminated during the experiment. This eliminated 

 time was divided by the total number of errors made with the 

 result that one error was found to be equivalent to thirteen 

 seconds of time. Likewise, 465 inches, the length of the true 

 path, was subtracted from the length of each run and these 

 remainders were simimed, giving a value representing the total 

 amount of unnecessary distance eliminated. This surplus dis- 

 tance was divided by the surplus time, giving three inches of 

 distance as equivalent to one second of time. An arbitrary 

 ordinate value was assigned to each error, and this ordinate 

 unit according to the above calculations is also equivalent to 

 thirteen seconds, and to thirty-nine inches. The three curves 

 constructed on this basis are represented in figure 2. 



The three sets of data can also be equated on a percentage 

 basis. As before, only the surplus or eliminated values are con- 

 sidered. For example, suppose the time of the first run is twenty 

 minutes and this time value is progressively decreased to zero. 

 The percentage method attempts to represent the rapidity of 

 this decrease irrespective of the absolute values. The time 

 values for the various runs are each divided by the time of the 

 first run. This gives a series of percentage values decreasing 

 from 100 to o. The value for any trial represents the percentage 

 of time 3^et to be eliminated, and hence a curve constructed 

 from these decreasing values will represent graphically the 

 rapidity of elimination. Percentage curves are likewise com- 

 puted and constructed from the distance and error data. Any 

 difference between the curves represents a difference in the rela- 

 tive rate of elimination. Our data were computed by such a 

 method, and the results were compared with the curves con- 

 structed by the previous method. According to the percentage 

 method, all curves will begin and end at the same levels, viz., 

 at the 100 and o values respectively. According to the first 

 method, the three curves will begin at different levels, but 

 terminate at the zero point. This fact is irrelevant, however, 

 for comparative purposes. Both methods give identical results. 



