COMPUTATION OF THE COMPOSITE NOISE 611 



the product 



h = eig2 • • • qj • ■ ■ Qn- 



This expression can be simplified, when the number of sources is 

 large and none is particularly outstanding, by considering, instead of 

 the individual sources, an average source having m = N/n peaks. 



The average probability then is 



P 2An 



m N 



300 300« 

 and 



N 



1 - 



300w 

 which leads to the approximation : 



N 



/o = g" = 1 - 



300« 



This expression can be further simplified when the number of sources 

 is large and p — N/300n is small by using the Poisson exponential 

 limit: 



where 



e = 2.718 



so that for this case 



M = (1 - g--'^/3oo)3oo. (4b) 



Measurement of Source Distributions 



The method outlined in this paper for computing the composite 

 noise assumes that information on the noise sources is available. 

 Such data, therefore, must be obtained before the method can be 

 applied. Representative measurements for a particular type of noise 

 source, however, when once obtained, can be used in any future noise 

 computation involving such a source. 



It is necessary to consider carefully the acoustic conditions under 

 which the sources are measured. For greatest accuracy the ambient 

 noise level at the point of measurement should be 20 db or more below 

 the average level of the source. Errors due to reflections can be 

 minimized by making the measurements at a relatively short distance 

 from the source out of doors or in a room that contains a large amount 

 of absorbing material. A distance of 2 feet is a convenient value for 

 most cases, and will be used as a reference value throughout the rest 

 of this paper. 



