MEASUEEMENT OF NOISE KAYE 177 



polated in the circuit with the object of approximating the results 

 more closely to the average aural interpretation. The weighting 

 curve may conveniently be chosen for a loudness corresponding to 

 that of a 1,000-cycle tone at 30 to 40 decibels above the threshold of 

 audibility. (Free, Journ. Acoust. Soc. Amer., July, 1930.) 



If it is desired to analyze the noise, this may be effected by in- 

 corporating electrical tuning or band-pass filtering devices Into the 

 circuit and so determining in turn the amount of energy associated 

 with individual components or bands of frequency. It is not always 

 possible, however, by such means to get sharp selectiveness, but, in 

 any event, we have to recognize that as yet our knowledge is not 

 sufficiently general to enable us to correlate exactly the overall loud- 

 ness of a noise with the energy or loudness of its constituents. 



It is possible to construct microphone and amplifier units which 

 are reasonably portable, such as that developed by Davis at the 

 National Physical Laboratory. 



SEARCH-TONE METHODS OF NOISE ANALYSIS 



The question of the practical analysis of noises which are reason- 

 ably periodic has been much facilitated by the introduction of search- 

 tone methods, which in general enable sound to be more conveniently 

 analyzed and with higher selectivity over a wide frequency range 

 than is possible by the method of tuned circuits. 



When search-tone methods are employed the noise to be analyzed 

 is received in a microphone, the current of which is amplified and 

 " mixed " in a valve-rectifier or modulator with that of a pure search 

 tone of constant intensity and variable known frequency from a 

 heterodyne oscillator. As a consequence, the modulated current con- 

 tains not only the search tone but also the summation and difl^erence 

 tones formed from the search tone and the various individual con- 

 stituents of the noise. For example, if the frequency of the search 

 tone is S and that of a particular constituent of the sound is C, the 

 frequencies of the summation and difference tones so formed will be 

 (S + C) and (S - C), respectively. 



One way of revealing the existence of either of these tones is to 

 apply the modulated current to a highly selective mechanical reso- 

 nator, such as a steel bar capable of vibrating longitudinally. Then, 

 as the search frequency is varied continuously, the bar will begin 

 to resonate whenever either (S + C)or(S — C) becomes equal to 

 the natural frequency of the bar. As we know S, we can evaluate C, 

 and, further, the degreee of response of the bar, which is observed by 

 suitable means, will give us a measure of the energy in the constitu- 

 ent in question. In practice S may range from, say, 11,000 to 16,000 

 cycles per second, while the natural frequency of the bar may well 



