106 



BELL SYSTEM TECHNICAL JOURNAL 



are shown in Fig. 5. For convenience and ease of interpretation, these 

 curves have been plotted on "probability" rather than rectangular 

 coordinates, as probability coordinates have the property of making 

 data whose distribution follows a normal law * form a straight line. 

 It will be noted that the experimentally determined points actually 

 fall so nearly on straight lines, that it is reasonable to assume straight 

 lines to represent them. It is likely that with a greater volume of 

 data, still greater conformity to the straight lines drawn, would be 

 obtained. 



In order to superpose the curves for the two volume indicators, the 

 levels are plotted in decibels with respect to the average overload level 



< 





-8-7-6-5-4-3-2-1 1 2 3 4 5 6 7 8 



VOLUME LEVEL IN DECIBELS (0 = MEAN OVERLOAD POINT) 



Fig. 4 — Distribution of overload points. 



determined from the tests. When calibrated to read alike on the same 

 sine-wave power, the experimental peak-reading instrument (with the 

 adjustments described above) reads on the average 7.4 decibels higher 

 on actual programs than the r-m-s instrument used in the tests. 



Now let it be imagined that the test amplifier is the one critical link 

 in a broadcast network and that an operator is given the duty of 

 satisfactorily adjusting the volume levels through the amplifier using 

 either of the two volume indicators tested. If he lets the louder 

 portions of the programs just reach the volume level marked "0 db" 

 on the curves, it will make no difference which volume indicator he 



^ The "normal" law has the form y = Ae~'"''^. 



