TIME INTERVALS IN TELEPHONIC CONVERSATION 



285 



observed without producing excessive amplitudes of the recording 

 strings during the strong parts of speech. These devices, as used, 

 distorted the recorded envelope of the speech sounds considerably, but 

 assisted materially in showing just where speech traces started and 

 stopped. Noise occasioned little difficulty in this study. When ob- 

 servable, the traces of the noise were so characteristic that little con- 

 fusion with speech waves resulted. 



BELL TELEPHONE LABORATORIES, NEW YORK 



NEW YORK 



Fig. 1 — Connections to New York-Chicago tie line to obtain 

 oscillograms of conversations. 



Some observations of response and resumption time were lost when 

 double talking occurred, becau.se echo suppressor operation introduced 

 appreciable loss into one side or the other of the recording circuit.^ 

 High gain in the automatic volume control and high speech volumes, 

 however, permitted many of such instances to be properly recorded. 

 There is little reason to believe that the instances lost were frequent 

 or that they differed in any particular regard from those recorded; 

 such instances occurred only in the event of double talking involving 

 weak speech by the responding talker. 



Instances in which double talking was clearly recorded required a 

 certain amount of arbitrary judgment to decide whether the response 

 was induced by a resumption pause or represented a negative response 

 time. Usually the structure of the conversation was evident, but it is 

 recognized that in some cases there is room for a legitimate difference 

 of opinion. 



^ This loss, while considerable, was not nearly so great as that introduced into the 

 transmission path by the echo suppressor operation. 



