612 



BELL SYSTEM TECHNICAL JOURNAL 



Readings should be obtained on all the noise peaks while the source 

 is being operated in a normal manner. If the scale of the particular 

 sound level meter used is limited, it may not be possible to read the 

 highest as well as the lowest peaks with a single potentiometer setting. 

 In such cases, the distribution of peaks may be measured in two or 

 more groups. 



Figure 1, Curve A, illustrates the measurement of source peaks in 

 the laboratory and represents a cumulative distribution of the peaks 

 from a typewriter as measured at a horizontal distance of about 2 feet 

 from the type bar guide. The machine was operated by an experienced 

 typist at an average rate. 



When it is not possible to simulate actual conditions of use of a 

 device sufficiently well in the laboratory, measurements on the source 



</i 76 



12 5 10 20 30 40 50 60 70 80 90 95 96 99 



PER CENT OF PEAKS ABOVE LEVEL SHOWN BY ORDINATE 



Fig. 1^ — -Distribution of noise peaks in a typing room. 



will have to be taken in the field. This may involve measuring in the 

 presence of considerable noise from other sources. In general, it is 

 feasible only to measure source peaks which are above the ambient 

 noise level. If, however, an appreciable number of peaks is above this 

 noise level, the rest of the distribution can be estimated and the 

 average value determined. Statistical methods for doing this have 

 been worked out for the case of normal distribution curves.^ Experi- 

 ence has indicated that the distributions of noise in db frequently are 

 approximately normal, so that these methods are applicable. 



Figure 2 is an illustration of a distribution of a group of sources 

 measured under adverse noise conditions. This curve shows the noise 

 which came from the metal trays in a cafeteria. The distribution had 

 to be obtained in the field because it was not feasible to estimate in 



