TRANSMISSION FEATURES OF NEW TELEPHONE SETS 363 



not unknown, the importance of such control was not fully appreciated. 

 Little choice was available in the quality of reproduction provided by 

 transmitter and receiver, because of meager design knowledge. 



Relatively recent and quite rapid developments in knowledge of the 

 problems involved, in materials, methods, and measuring facilities, 

 have now presented the necessity for a solution in essentially a three- 

 dimensional form. These three dimensions may be described as 

 volume, noise and quality. The solution of the problem on this 

 basis is obviously more difficult, and has required the development 

 of methods for quantitatively evaluating and rating their character- 

 istics in terms of some common yardstick. 



Methods of Rating Transmission 



For reasons already suggested, such a yardstick must be based on 

 service performance — on the results obtained by actual users in the 

 course of day-to-day telephone service. 



Extensive investigation has indicated that the best comparative 

 measure of this transmission performance in local exchange service is 

 to be found in the time rate of the occurrence of repetitions required 

 by subscribers for understanding telephone conversations.^ Or, more 

 explicitly, when two transmission conditions have the same repetition 

 rates, all other service factors being equal, these conditions are taken 

 to be equal with respect to transmission performance. Where two 

 conditions are not alike it is usually possible to evaluate the difference 

 in the repetition rates for the same users by inserting distortionless 

 loss in the better condition until both have equal repetition rates. 



Thus, by taking as a reference a typical telephone circuit of specified 

 make-up, the effects of various factors such as distortion, noise, 

 attenuation, sidetone, or type of instrument, may all be expressed in 

 the common terms of the reference circuit trunk which will give the 

 same repetition rate. 



Instead of making this adjustment in every case for the purpose 

 of evaluating the relative performance of different test conditions for 

 the same users, the evaluation may be made rather closely over a 

 limited range by the following typical relation derived from repetition 

 observations on circuits containing trunks, the losses of which were 

 varied over a range of values. 



db = 50 logio Ri/R2,^ 



where Ri, and R2 are the repetition rates of two conditions under com- 

 parison, and the db figure is the change in the reference trunk which 

 has the same relative effect on the repetition rate. 



