104 BELL SYSTEM TECHNICAL JOURNAL 



listens to the program and adjusts the volume level at the output of 

 the amplifier by means of the coupled attenuators, until he determines 

 the point at which distortion due to overloading is just audible, when 

 the key is operated from the reference to the test condition. This is 

 repeated for a number of different programs and observers until a large 

 number of observations have been obtained. The volume levels indi- 

 cated by the different volume indicators at the amplifier output are 

 determined for each observation. These are found to have a con- 

 siderable spread, due not only to the differences in the nature of the 

 programs but also to differences in the acuity of perception of the 

 distortion by the various observers. The method of analyzing the 

 data is described later. 



In the second technique, the group method, a group of observers 

 simultaneously listens to a program which is repeated with the key 

 operated alternately to the test and reference positions. The two 

 conditions are distinguished to the observers (but not identified as to 

 which is which) by a letter associated with each condition in an 

 illuminated sign. The letters A, B and C are used, two being chosen 

 at random for each test. A vote is taken as to which condition, desig- 

 nated by one of the two letters employed in the particular test, is 

 preferred with respect to freedom from distortion. A number of such 

 tests, covering the range from a level below the point where distortion 

 can be detected by anyone to a level high enough for all to observe 

 distortion, establishes a curve between the per cent of observers 

 correctly choosing the reference condition as having the least distor- 

 tion, and the amplifier output level as read by each volume indicator 

 used in the tests. Similar curves are determined for a number of 

 kinds of program material, and for purposes of comparison the overload 

 point for each program is taken from the point on the curve for each 

 volume indicator, where 80 per cent ^ of the observers voted correctly. 



As noted, judgment tests of this sort require many observations and 

 checks to obtain reliable results. A larger volume of data is available 

 for the individual method, so the results from tests made by that 

 method have been chosen to be reported here. Some tests have also 

 been made with the group method and, while the results are less 

 conclusive, they substantiate those recorded below. 



Tests by the individual method to compare peak-reading and r-m-s 



volume indicators have been carried out a number of times during the 



past two years. In each of these tests a number of observers have 



taken part and a number of samples of program material of a variety of 



' "Audible Frequency Ranges of Music, Speech and Noise," W. B. Snow, Journal 

 of the Acoustical Society of America, July 1931. 



