Appendix IV. 143 



APPENDIX IV. 



OBSERVATIONS ON THE ASSOCIATION PULSE DATA OF DODGE AND BENEDICT. 



The measurements of the association pulse by Dodge and Benedict were 

 divided into three divisions — pre-stimulus, stimulus to reaction, and post- 

 reaction. The latter is sometimes mentioned as post-stimulation. (See 

 their page 198.) In their figure 31 they have embodied diagrammatically the 

 records of Subject VII. In these curves it may be clearly seen that an increase 

 in pulse rate follows the stimulation of the association words. One point, 

 however, deserved consideration in reference to these curves. It appears that 

 the right end of the curve almost invariably fails to reach the level of the left 

 end. The large fall which frequently occurred between the pre-stimulus and 

 post-reaction ends of the curves in figure 31, taken in connection with the 

 statement on page 198 of Dodge and Benedict's report that all the pulse waves 

 were read and the data are capable of any other arrangement that statistical 

 interest might demand, led us to reread the record which forms the basis for 

 Dodge and Benedict's table 34, page 197. 



It will be observed from their figure 30, page 195, that there are unbroken 

 horizontal lines which separate the reactions into groups of five. This was 

 produced by breaking the circuit of the pulse-recorder so that the marker 

 point gave a straight line. Sometimes this break in the circuit came quite 

 soon after the fifth reaction, hence there are fewer post-stimulus cycles after 

 the fifth reaction, and furthermore, the last figure in a fifth reaction was not 

 joined directly to the first measured value of the next reaction. In Dodge 

 and Benedict's table 34, page 197, it will be seen that the fifth reaction rarely 

 has more than four pulse cycles in the post-reaction phase. In three cases, the 

 fifth or last reaction in a set contained only three pulse cycles in its post- 

 reaction phase. The two readings of the record checked remarkably well. 

 In the final tables 35 and 36, from which figure 31 was drawn, many measure- 

 ments are dropped out by limiting the number of columns. For example, the 

 number of columns from stimulus to reaction is given in the average tables as 

 2, while in table 34 there were many times when three or more pulse cycles 

 came within that period. The same may be said for the pre-stimulation and 

 the post-reaction divisions of the tables. The averages at the foot of table 34 

 show that in 26 cases pulse cycles were omitted from the final averages, and 

 that in 21 cases pulse cycles were omitted from the pre-stimulus section. Four 

 of these cycles were of longer duration than the last which was included in 

 their table; 12 were shorter and 5 of the same length. It is therefore cer- 

 tain that if we plotted table 34 with the additional data, curve 2 for the first 

 day in figure 31 (page 201) would connect at its ends on the same level. The 

 discrepancies between the level of the ends of these curves and also in the data 

 in table 36 are largely to be explained in this way. 



One point in the technique of Dodge and Benedict intensified this tendency 

 to drop out data at the extreme ends of the table. This was the fact that there 

 was no experimental break between the post-reaction and the pre-stimulation 

 pulse cycles, since the records were taken on a kymograph, although an appa- 

 rent break was produced when the record was cut to be removed from the 

 drum. Since this apparent break was not always the same in its time rela- 

 tion to the three phases of the record, the post-reaction phase was occasionally 

 shortened and, in consequence, the pre-stimulation phase was lengthened, and 

 vice versa. When the final tables were made up these extra long ends were not 

 included in the averages, thus omitting the essential transitions between the 

 two ends of the curves under discussion. 



