420 BELL SYSTEM TECHNICAL JOURNAL 



level of the reference tone is L decibels when this condition is reached, 

 the sound is said to have a loudness level of L decibels. When the 

 character of the sound being measured differs only slightly from that 

 of the reference tone, the comparison is easily and quickly made, but 

 for other sounds the numerous factors which enter into a judgment of 

 equality of loudness become important, and an experimental method 

 should be used which will yield results typical of the average normal 

 ear and normal physiological and psychological conditions. 



A variety of methods have been proposed to accomplish this, 

 differing not only in general classification, that is, the method of 

 average error, constant stimuli, etc., but also in important experi- 

 mental details such as the control of noise conditions and fatigue 

 effects. In some instances unique devices have been used to facilitate 

 a ready comparison of sounds. One of these, the alternation 

 phonometer, ^^ introduces into the comparison important factors such 

 as the duration time of the sounds and the effect of transient condi- 

 tions. The merits of a particular method will depend upon the 

 circumstances under which it is to be used. The one to be described 

 here was developed for an extensive series of laboratory tests. 



To determine when two sounds are equally loud it is necessary to 

 rely upon the judgment of an observer, and this involves of course, 

 not only the ear mechanism, but also associated mental processes, and 

 effectively imbeds the problem in a variety of psychological factors. 

 These difficulties are enhanced by the large variations found in the 

 judgments of different observers, necessitating an investigation con- 

 ducted on a statistical basis. The method of constant stimuli, wherein 

 the observer listens to fixed levels of the two sounds and estimates 

 which sound is the louder, seemed best adapted to control the many 

 factors involved, when using several observers simultaneously. By 

 means of this method, an observer's part in the test can be readily 

 limited to an indication of his loudness judgment. This is essential 

 as it w'as found that manipulation of apparatus controls, even though 

 they were not calibrated, or participation in any way other than as a 

 judge of loudness values, introduced undesirable factors which were 

 aggravated by continued use of the same observers over a long period 

 of time. Control of fatigue, memory effects, and the association of an 

 observer's judgments with the results of the tests or with the judgments 

 of other observers could be rigidly maintained with this method, as 

 will be seen from the detailed explanation of the experimental pro- 

 cedure. 



12 D. Mackenzie, " Relative Sensitivity of the Ear at DifiFerent Levels of Loudness," 

 Phys. Rev. 20, 331 (1922). 



