154 BELL SYSTEM TECHNICAL JOURNAL 



In addition to the requirements for weighting and rule of combina- 

 tion, it was thought desirable to employ an indicating instrument in 

 which the change of reading was about as rapid as the change in 

 appreciation of loudness in human hearing. From published results 

 and confirming tests it was determined that on the average, the 

 indicating instrument should reach a full deflection for sounds lasting 

 .2 second or longer. 



Under these general specifications, several models of circuit noise 

 meters were built and two series of tests were made to determine their 

 adequacy for measuring circuit noise. These tests were made under 

 the auspices of the Joint Subcommittee on Development and Research 

 of the Edison Electric Institute and the Bell System. The first was a 

 rather extensive series of articulation tests on open-wire toll circuit 

 noise. Since none of the toll circuit noises tested contained compo- 

 nents of importance above 2,000 cycles, a series of judgment tests was 

 carried out on representative noise of the type arising by induction in 

 telephone circuits exposed to a-c. lines supplying rectifiers and on vari- 

 ous high-frequency noises derived therefrom. 



The articulation tests showed that when toll circuit noises of various 

 types produced equal losses in articulation under the given set of 

 telephone conditions, they were measured as substantially equal by 

 both the objective and subjective methods of measuring. The ob- 

 jective method gave a slightly better correlation than did the subjective 

 method even though the average of 18 individual observers was used 

 in the latter. While the correlations were not as close with the high- 

 frequency noises as in the case of the more common types of toll 

 circuit noise, on the average the noise meter rated the rectifier noises 

 at least as well as did the ear balance method, the latter using 10 

 observers. 



A device called the "Telephone Interference Factor Meter" for 

 measuring or rating the wave shape of power system currents and 

 voltages in terms of their influence on exposed telephone circuits was 

 described in the Osborne paper of 1919, referred to above. With this 

 instrument, an indication was obtained of the total harmonic content 

 of a given voltage wave, the individual components present being 

 weighted approximately in proportion to their relative interfering 

 efi"ects. 



The data obtained from the more recent studies of relative inter- 

 fering eff"ects described above have made possible a revision of the 

 method of measuring T.I.F., in which the basic principle has been 

 retained but in which the frequency weighting characteristic has been 

 revised somewhat and its range extended to about 5,000 cycles. In 



1 



