128 BELL SYSTEM TECHNICAL JOURNAL 



of the system. At high frequencies the transmitter should reflect 

 nearly as much alone as when part of a large wall, and the readings 

 with and without the wall should be nearly equal. Plotting ratio of 

 response without, to response wuth the wall was expected to yield a 

 curve which could be used to make the final reduction of electrical 

 output to incident sound energy, and so permit a more accurate de- 

 termination of the spectrum of sound energy of the voice. 



No consistent results werre obtained after several trials and the ex- 

 periment was abandoned. The failure is doubtless to be ascribed to 

 standing waves, the character of which is very sensitive to the location 

 in the room of the transmitter and the wall. This experiment is to be 

 repeated under more favorable conditions when standing waves can 

 be eliminated. 



Thus, the curves finally obtained show no more than the frequency 

 distribution of energy in speech in terms of the mechanical energy of a 

 more or less ideal transmitter diaphragm. However, this information 

 has its value because in any given configuration of transmitter, 

 speaker, and room, there is a definite correspondence between the sound 

 energy of the voice and the force acting on the diaphragm on which it 

 falls, and in telephony at any rate it is this action on the diaphragm 

 with which we are immediately concerned. 



In conclusion we may give a determination of the total energy rate 

 of speech, obtained as a by-product of the preceding investigation. 

 Knowing the calibration of the system in absolute units, it is possible 

 to determine the alternating pressure on the condenser transmitter 

 diaphragm exposed to continuous speech from the normally modulated 

 voice under the conditions of the experiment. Using the mean of the 

 values obtained with 9 observers we find for the alternating pressure 

 11.3 dynes per sq. cm. (r.m.s.) for a distance of 2.5 cm. from mouth to 

 diaphragm. This corresponds to an energy flow of 3.2 ergs per sq. 

 cm. per second. Assuming that this energy flow is distributed uni- 

 formly over a hemisphere of 2.5 cm. radius, we may take 125 ergs 

 per second as the total sound energy flow from the lips with the 

 normally modulated voice. 



