184 BELL SYSTEM TECHNICAL JOURNAL 



From the above illustration it is evident that if we know how the 

 noise-to-signal ratio varies on a given circuit with a certain terminal 

 arrangement, it is possible to determine the percentage lost circuit 

 time for an improved or degraded system, the relative effectiveness of 

 which can be expressed in terms of db above or below the initial 

 arrangement. 



Transatlantic Noise-to-Signal Data 



As a part of the regular operating routine on the various trans- 

 oceanic short-wave radio telephone circuits associated with the Bell 

 System, measurements of noise at the receiver output are made at ap- 

 proximately half-hourly intervals. These measurements are made at a 

 point in the voice-frequency wire circuits where the speech volume is 

 normally held constant. They are therefore effectively measurements 

 of noise-to-signal ratio although not expressed in such terms. The 

 instrument used is known as the Western Electric 6-A Transmission 

 Measuring set.^ In the following discussion measurements made 

 with this instrument are referred to as "6-A Noise." 



In Fig. 2 (a) the upper curve shows the percentage distribution of 

 6-A noise values measured at New York during 1930 on an 18-mc. 

 London-New York circuit. These and the curves to follow are plotted 

 to an arithmetical probability scale. The year 1930 was severely 

 disturbed and from the radio transmission standpoint is perhaps 

 representative of the peak of the well-known eleven-year magnetic 

 disturbance cycle. Since the performance of a two-way telephone 

 circuit depends upon transmission conditions in the two directions the 

 upper curve of Fig. 2 (a) does not accurately portray the full effect of 

 the noise factor upon the circuit. The lower curve in this figure 

 represents the distribution of the higher of simultaneous ^ 6-A noise 

 values measured at New York and London. The small difference 

 between the two distributions is evidence that the most important 

 influence — that of magnetic disturbance — affects the transmission in 

 both directions coincidentally. 



It will be noted that these 6-A noise curves of Fig. 2 (a) and of the 

 following figures bend downward in the region of low 6-A noise and 

 upward where the 6-A noise becomes high. There is reason to believe 

 that these bends are introduced by the terminal equipment and that 

 the actual noise distribution of interest here approaches a straight 

 line on the probability scale used, or in other words is a fortuituous 



' L. Espenschied, "Methods for Measuring Interfering Noises," Proc. I.R.E., Vol. 

 19, p. 1951, November, 1931. 



^ Measurements less than seven minutes apart at the two ends of the circuit were 

 treated as "simultaneous" in this analysis. 



