RELATION OF PRACTICE TO RATE OF LEARNING 141 



One hundred seventy-five pairs of successive trials made in 

 different periods of practice with an equal number of pairs made 

 during a single day's practice were obtained. The method gives 

 essentially a random sample of the two kinds of pairs of succes- 

 sive trials, with an equal number of errors and an equal distribu- 

 tion of errors between the members of the pairs of each series. 

 The role of chance in the production of duplicate errors may 

 therefore be disregarded, since it should be the same for both 

 kinds of pairs, and the relative proportion of duplicate to dis- 

 tinct errors in the two kinds of pairs may be considered the 

 result of the time intervening between the successive trials. In 

 computing the errors wrong turns only in the maze were con- 

 sidered and only one error was counted at each turn where 

 errors were made. The results of the analysis are given in the 

 following table. The totals of all cases are given and the figures 

 for the numbers of diverse and duplicate errors are directly 

 comparable, being based upon the same number of errors with 

 like distribution between the pairs of trials. 



Successive Practice-periods 



Total number of errors in last trials of first practice-periods 463 



Total number of errors in first trials of succeeding practice- 

 periods 524 



Total number of pairs of identical errors 247 



Total number of diverse errors 493 



Single Practice-periods 



Total number of errors in first trials of pairs corresponding to 



those above 463 



Total number of errors in succeeding trials of pairs corresponding 



to those above 524 



Total number of pairs of identical errors 274 



Total number of diverse errors 439 



In both kinds of pairs the number of duplicate errors is slightly 

 in excess of the number of diverse ones. The maze, as arranged, 

 offers the possibility of 12 different errors and the average num- 

 ber of errors per trial was 2.82. The chances, then, that the 

 number of duplicate errors would equal the number of diverse 

 errors was only about one to eight. The fact that the numbers 

 were equal when the data included so many cases seems to prove 

 that there is some predetermining factor which causes errors 

 once made to be repeated both in single practice periods and 

 from one practice-period to the next. 



