2U BELL SYSTEM TECHNICAL JOURNAL 



disturbance which corresponds to a narrow segment of the spectrum persists 

 for a considerable period so that there is much overlapping of the contribu- 

 tions of neighboring segments. It is therefore difficult to deduce the nature 

 of the disturbance at any particular time from any narrow region of its 

 spectrum. For the same reason it is difficult to associate the delay ex- 

 perienced by an aperiodic signal with the steady state characteristic of a net- 

 work when the attenuation of the latter is changing rapidly with frequency. 

 The net result of our study then is that steady state phase delay has no 

 direct relation to the particular types of delay of an aperiodic signal which 

 we have chosen to investigate. When the amplitude does not change 

 rapidly with frequency, envelope delay is identical with the delay produced 

 in the maximum value of the envelope of a disturbance corresponding to 

 that part of the signal spectrum which is in the immediate neighborhood 

 of the frequency in question. The envelope delay, together with the phase 

 shift, determines the delay in the maximum absolute value of this dis- 

 turbance, subject to an uncertainty of half a period. This uncertainty 

 depends on the particular combination of signal spectrum and system 

 characteristic. When the amplitude does change rapidly with frequency, 

 the envelope delay still gives the delay in the maximum value of the envelope. 

 However, this maximum is so fiat that the interpretation of the results is 

 very difficult. 



