598 



BELL SYSTEM TECHNICAL JOURNAL 



Consider now the second requirement which the recording system 

 had to meet: namely, that the output of the system should be a linear 

 function of the input within the working energy range at each fre- 

 quency. Thorough investigation of the condenser transmitter had 

 shown that this instrument met this second requirement very well; it 

 was only necessary to test the remainder of the system. Fig. 7 gives 



Freqiunry 



Kig. 7 — .■\miilitU(Je frequency characteristics of circuit-oscillograph al ililTerent 

 energy- levels 



the results of these tests, the voltages introduced in series with the 

 transmitter at the input being maintained at different constant levels, 

 while the frequency was \'aried. An inspection of the data shows 

 that this requirement was \ery accurately fulfilled, by the whole 

 electrical system. 



Returning now to the owrall characteristics of llie apjiaratus, it 

 was thought advisable to test the calibrations in am])litude and phase 

 lag by comparing the computed and the observed distortion when a 

 square-topped acoustic wave was impressed on the apparatus. The 

 steep sides and the flat tops of these waves can be reproduced with- 

 out distortion only if the apparatus possesses first class characteristics, 

 both in amplitude and phase lag, and the test was a severe one. As 

 would be expected from the calibration curves of Fig. 6 there was a 

 certain amount of distortion in recording this wave, and the square- 



