ROOM NOISES IN TELEPHONE LOCA TIONS 655 



canal is always open to the air of the room regardless of how firmly the 

 receiver is pressed against the ear. The tone is generated by means of 

 a phonograph record and a magnetic phonograph record pick-up, and 

 is a variable-frequency tone, the pitch of which varies between certain 

 limits several times per second. An attenuator is placed between the 

 magnetic pick-up and the receiver. The observer sets the attenuator 

 at a point where he can barely recognize the sound of the warble in the 

 presence of the room noise. He also obtains the setting at which he 

 can barely hear the warble in a perfectly quiet room. The difference 

 between these two settings is a measure of the masking effect of the 

 noise in this room upon the warbler tone, for this particular observer. 



An idea of the frequency composition of a given room noise may be 

 obtained by using several different warbler tones, each covering a differ- 

 ent portion of the voice-frequency range. This is based on the fact 

 that, in general, a tone of a given frequency masks to a greater extent 

 tones that are near it in frequency than tones that are far removed 

 from it in frequency. 



The phonograph records used in the present room noise survey were 

 three-band records, i.e., three warbler tones were cut on each record, 

 each tone occupying about one-third of the available space. The fre- 

 quencies included in the various bands were as follows: high band, 1500- 

 5600 cycles per second; middle band, 750-1500 cycles per second; low 

 band, 250-750 cycles per second. In each band the frequency varied 

 continuously from the lower to the upper limit and back to the lower 

 limit, the period of such a complete "warble" being about one-sixth of 

 a second. 



Electrical Method — Room Noise Meter. — There is, of course, a number 

 of different electrical methods which might be employed for measuring 

 room noise, ranging from a single over-all measurement to a complete 

 wave shape or frequency analysis. The complete analysis or the meas- 

 urement of energy present in a considerable number of narrow fre- 

 quency bands is subject to the disadvantages, for such a survey as this, 

 of slowness of measurement and bulkiness of testing equipment. 



The method which was adopted was one based on the use of a fre- 

 quency weighting simulating the sensitivity of the ear. This frequency 

 characteristic is shown on Fig. 1. It is an equal loudness weighting; 

 that is, the room noise meter was so constructed that different single- 

 frequency noises of equal loudness would give approximately the same 

 meter readings. The shape of an equal-loudness curve is somewhat 

 flatter for high levels of loudness than for low ones. The weighting 

 curve chosen for the meter was for a loudness corresponding to that of 

 a 1000-cycle tone 30 or 40 db above the threshold of audibility. This 



