658 BELL S YS TE M TE CHNI CA L JO UR NA L 



An over-all calibration of the meter, as a function of frequency, is 

 given on this figure. To obtain this, separate determinations were 

 made of the volts generated by the condenser transmitter per unit of 

 pressure, and the meter reading per volt generated by the transmitter, 

 as a function of frequency; and the results were combined to give the 

 values shown. Harmonics in the testing waves were reduced to such a 

 point that they did not affect the results. After a substantial part of 

 the survey had been completed, a check was made of the electrical por- 

 tion of the calibration, and the changes found were quite negligible. 



Accuracy of the Meter. — The precision of the apparatus is substan- 

 tially greater than the precision with which ordinary varying noises can 

 be measured. The readings obtained for steady inputs are propor- 

 tional to the input, with an error of less than 1/2 db, over the entire 

 range of noise amplitudes found in the survey. The apparatus is 

 shielded electrically. In only one case did electrical fields produce any 

 observed errors in the readings ; this was when an attempt was made to 

 measure the room noise near a rotary converter in a power station. 

 The vacuum tubes are mounted in such a way that the effects of ordin- 

 ary mechanical vibration on the readings are negligible. 



Comparison of the Two Methods. — In general, the meter method gives 

 results in physical terms while the masking method gives them in terms 

 of effects on the ear; consequently, the choice of the method to be em- 

 ployed in any particular case depends somewhat on the use to which 

 data will be put. It is true that the meter includes a network to simu- 

 late the sensitivity of the ear for various frequencies; it does not, how- 

 ever, simulate other properties of the ear, such as the departures from 

 linearity in response by which subjective tones are produced by the ear 

 mechanism, and the complicated way in which one sound masks an- 

 other.*^ 



The meter method, unlike the masking method, avoids any errors 

 due to variations in human ears. This advantage is offset to some ex- 

 tent by the fluctuations of the meter needle, which make it difficult to 

 obtain the mean reading if the noise is unsteady as is the case with 

 most room noises. 



In the case of noises of a distinctly intermittent, staccato character, 

 the warbler tone can be heard and recognized in the brief intervals when 

 the noise is a minimum. A preliminary investigation showed that, for 

 a noise of this sort, the relation between readings obtained by the mask- 

 ing method and by the meter method was different from the relation 

 obtained for a steady noise, the warbler readings being relatively lower 

 in the case of intermittent noise. 



* R. L. Wegel and C. E. Lane, "Auditory Masking and Dynamics of the Inner 

 Ear," Physical Rev., Feb. 1924. 



