FUNCTION OF VIBRISSAE IN BEHAVIOR OF WHITE RAT 17 



when the problem was learned or one period past such time.'^ 

 The numerical results of time and error for each animal were 

 then divided into an equal number of serial groups, say ten, 

 regardless of whether in the learning he had taken twenty or 

 forty trials. In the one case there would be two numbers in 

 each group and in the other four. The averages were then taken 

 by groups and thus there was obtained for each animal a series 

 of ten numbers. If there was an excess it was always distributed 

 among the beginning groups. Suppose instead of twenty 

 there had been twenty-two, and instead of forty, forty-five trials 

 in the illustration used and it was still desired to make ten groups. 

 Then the first two divisions of the first animal's records and the 

 first five of the second animal's records would contain an addi- 

 tional member. The quantities for all the animals were then 

 averaged in serial order and there was obtained in this way 

 numerical data for plotting a curve in which the beginnings of 

 the learning process corresponded as did also the second, third 

 and end stages. In the old curve as usually plotted only the 

 beginning stage was fairly representative of all. 



A curve made in this way should represent the learning pro- 

 cess in its successive stages with the individual variations elimi- 

 nated in proportion to the number of animals used. It admits 

 (jf better comparison between different lines of experimentation 

 and with different species of animals. Direct comparison of 

 these curves will be made in connection with the discussion of 

 the results of the experimentation. 



G. E.xperimentation 



Experiment I. Normal rats on the maze with sides up. For the 

 purpose of comparison we used a set of five males on the maze 

 with the sides to the runways up. This first experiment was 

 made to give us data to compare with that obtained from, suc- 

 ceeding groups where the walls to the ninways were down. The 

 vibrissae brushed the sides as the animals ran but, if they were 

 of service, the exact way in which they were functioning was 

 hidden by the apparatus. The conduct of this group will not 

 be described since it differed in no way from that which has 

 been described for so many similar mazes before. The general 



'^Note: — Except in some cases where the experiment was continued for the 

 sake of comparison with the pubHshed results of others. 



