FREE ASSOCIATION. Ill 



was consequently necessary to correlate the processes by some scheme 

 that would identify each phase of the records, as well as to unite the 

 various records into one whole. 



The signal for giving each stimulus word was transmitted to the 

 operator (Wells) at each revolution of the kymograph drum by an 

 automatic break in the 2- volt incandescent signal-lamp circuit. Since 

 the kymograph was regulated to make 1 revolution in 10 seconds, these 

 signals placed the stimuli 10 seconds apart. At the moment of actually 

 giving the stimulus word, the operator simultaneously pressed a tele- 

 graph key that registered the event on the kymograph record by a 

 characteristic break in the curve. On the continuous spiral record cor- 

 responding to 50 experiments, these breaks come at approximately the 

 same moment of each revolution, and make a more or less approximately 

 straight line. When the subject responded to the stimulus, the operator 

 signaled the moment of response by releasing his pressure on the tele- 

 graph key, and the recording curve correspondingly returned to its 

 pre-stimulation base-line. The latent time of each response thus 

 appeared on the records as a plateau, whose rise corresponded with 

 the moment of stimulation and whose fall corresponded with the opera- 

 tor's reaction to the response of the subject. 



A constant error in the association time as thus recorded is involved 

 in the fact that the stimulation signal is given synchronously with the 

 stimulus word, while the recorded moment of reaction must include the 

 personal equation of the operator, who can give the signal only after he 

 hears the subject speak. While it makes all our values somewhat too 

 large, in the comparison of one series of performances with another, 

 this constant error is negligible. 



Aside from this constant and negligible error, the probability that 

 any measured association time corresponds with the real association time 

 is dependent on the variability of the personal equation of the operator. 

 Our records are protected in this respect by the fact that Wells is an 

 unusually practiced reactor, with a small mean variation. Moreover, 

 we did not aim at an accuracy greater than is implied in the rather 

 large unit of measurement of 0.01". 



We shall probably be criticized for not using some more mechanical 

 form of stimulus and reaction key. The answer to all such criticism 

 must be to emphasize the main purpose of the free-association experi- 

 ments. Their main value lies in the character of the response. Any- 

 thing that tends to disturb that phase of the experiment is unpardon- 

 able. Other phases are only of relative importance. For example, it 

 would have been easy to give the stimulus word optically, with all the 

 accuracy that characterizes the word-reaction experiment. But the 

 optical word is a stimulus for a very different mental operation from the 

 auditory. The inevitable associate for the optical word is its auditory- 

 motor associate. We depended on that regular connection in the word- 



