378 BELL SYSTEM TECHNICAL JOURNAL 



circuits was referred to as a "program weighting network," while 

 that used with the message circuit was the ordinary "message weight- 

 ing network." The noise and crosstalk volume was then recorded in 

 db referring to reference noise with either program weighting or 

 message weighting. Reference noise is that amount of interference 

 which will produce the same meter reading as 10~^^ watt of 1,000-cycle 

 power, which is 90 db below 1 milliwatt. 



The results of this survey indicated that in consideration of the 

 layout and levels of the existing message circuits and of the noise 

 existent on these circuits and on the program circuits, the value for 

 maximum program volume, should, under normal conditions, be + 3 

 referred to reference volume; that is, at this value the best balance 

 between program to message crosstalk and program circuit noise 

 would result. It was also determined that on very long sections, or 

 on sections where all circuits were subjected to severe noise exposure, 

 the maximum volume on the program circuits could be increased 3 db 

 to improve the signal-to-noise ratio on the program circuits. This 

 higher volume could be permitted in these cases since on the longer 

 sections the message circuits also usually operate at higher levels, 

 and on the especially noisy short sections the increased crosstalk to 

 the message circuits will ordinarily be masked by the greater noise. 



The average noise measured at San Francisco or Chicago at the 

 circuit terminals at the reference volume point was 49 db above 

 reference noise "program weighting" when the restoring network 

 was included at the receiving terminal. The noise averaged 5 db 

 higher than this with the restoring network removed. This value of 

 noise is about 43 db below the maximum power of the program 

 measured at the same point with the same measuring instrument. 

 This, therefore, establishes a signal-to-noise ratio of about 43 db, 

 thus permitting a volume range of approximately 40 db. 



The various tests referred to gave statistical data concerning the 

 transmission performance of the circuits from which it could readily 

 be predicted that the circuits would transmit programs with very little 

 impairment to quality. To substantiate this, very critical listening 

 tests were made, comparing the quality of a program after it had been 

 transmitted over various length circuits with the same program trans- 

 mitted over a reference circuit which was distortionless over the 

 frequency range for which the circuits were designed, namely, to 8,000 

 cycles. Figure 25 shows schematically the terminal arrangements 

 employed at San Francisco for these listening, or, as they are more 

 commonly called, comparison tests. 



Various types of programs were used, such as speech, vocal and 



