RESULTS OF THE WORLD'S FAIR HEARING TESTS 559 



obtained from the results of the Fair tests. The dotted portions 

 represent extrapolations of the distributions beyond the intensity and 

 frequency ranges used in the tests, and are of course speculative in 

 character. The extrapolations to lower frequencies are based on the 

 shape of the contours below 1760 cycles and the high correlation that 

 has been found to exist in individual audiograms for frequencies from 

 64 to 512 cycles. ^^'^^ The extrapolations to large hearing losses for 

 the 3520- and 7040-cycle tones were made by extending the curves 

 of Fig. 5 as suggested by comparison with the results of other surveys. 

 The contours show several interesting things. The range over which 

 hearing acuity varies is quite uniform up to about 2000 cycles, and 90 

 per cent of the group lie within a range of 30 db. Above 2000 cycles 

 the range increases rapidly. Since most of the sounds met with in 

 daily life have intensity levels greater than the 25 per cent contour, 

 fully three-fourths of the people can hear ordinary sounds throughout 

 the frequency range from 100 to 10,000 cycles. 



The Onset of Deafness 



With the increasing attention being given to the prevention of 

 deafness by early detection, it is of considerable practical importance 

 to define the beginning, or onset, of deafness. Perfectly normal ears 

 are not exactly alike; some are more acute and others are less acute 

 than the average. How much less acute than average may an ear 

 become before deafness begins? In a preceding section, hearing loss 

 was evaluated on the basis of the handicap that it would impose. 

 In this section we take a different viewpoint and use the term "be- 

 ginning of deafness" to mean a departure from average hearing acuity 

 sufficient to justify the expectation of associating the departure with a 

 specific cause. 



In Fig. 9 there is shown a typical distribution curve for hearing loss. 

 It shows the relative frequency of occurrence of various degrees of 

 hearing acuity among a given class of people. Such curves can be 

 well described for many practical purposes by two quantities, the 

 average hearing loss and the standard deviation. The latter quantity, 

 designated as cr, is a measure of the spread of the individual values 

 from the average. 



The experience of statisticians with distributions of observations of 

 widely different character indicates that there is little chance of assign- 

 ing specific causes for the deviations of observations which lie closer 



'^ E. G. Witting and W. Hughson, "Inherent Accuracy of Repeated Clinical 

 Audiograms," Laryngoscope, 50: 259 (1940). 



^'^ W. C. Beasley, "Correlation Between Hearing Loss Measurements," Jour, 

 Acous, Soc. Amer., 12: 104-113 (1940), 



