FREE ASSOCIATION. 113 



not a mere respiration rhythm, but is superposed on the latter in a 

 definite manner. 



A time-line was also introduced into the records to control the accu- 

 racy of the kymograph. The pendulum of an accurately running clock 

 was made to break the electric circuit of a time-marker, which was 

 thus permitted to vibrate against the drum for a moment every 2 

 seconds. This intermittent time-record is so delicate that it can not 

 interfere in the least with the other lines, while it serves as an absolute 

 guarantee of the speed of the kymograph. 



STIMULUS WORDS. 



The series of stimulus words was that given in the Appendix of the 

 monograph by Woodworth and Wells. 1 The Kent-Rosanoff 2 words 

 were eliminated from it, and made into two series, D and D', as here- 

 after described. The entire series was divided into 20 lists of 50 words 

 each. One list formed the material for a single experimental period. 

 Six lists were given on each experimental day, regularly alternating 

 with the Faradic threshold experiments. On a few occasions diffi- 

 culties of technique caused a delay which necessitated the omission of 

 the threshold experiment, but the interval between the association 

 experiments approximated 12 minutes in each case. The instructions 

 to the subject were verbal, in a form that frequent repetition has re- 

 duced to practical uniformity. On the first day, conventional examples 

 of stimulus and response were given to the subject, who also reacted 

 correctly to preliminary stimulus words before the experiments were 

 begun. In this manner all difficulties in understanding the nature of 

 the test were avoided during the experiment. If a stimulus word was 

 misunderstood, it was taken in the sense in which it was understood; 

 if the response were doubtfully understood, the subject was requested 

 to spell it, or was asked about ambiguities. 



There are three more or less standard ways of dealing with the data 

 of the association experiment. These are: (1) according to the reaction 

 time of the response; (2) according to certain quasilogical relations of 

 the response and the stimulus word; (3) by the statistical frequency 

 of the responses within the range of the material where this has been 

 determined. In addition, the present experiments record the pulse- 

 reactions of the subjects and, in certain cases, also "psycho-galvanic" 

 reactions. These phases of the experiment are first described in order, 

 after which some questions of correlation are dealt with. 



Woodworth and Wells. Psychological Monographs, 1911, 13, No. 57. 

 2 Kent and Rosanoff, Am. Journ. Insanity, 1910, 67, pp. 37 and 317. 



