342 BELL SYSTEM TECHNICAL JOURNAL 



spending element in the working reference system, that is, when it 

 is part of a typical telephone system. This method of determining 

 losses was adopted because the effects of distortion, noise and side- 

 tone in any one element depend to a greater or less extent on all the 

 characteristics of the remainder of the circuit. It is therefore essential 

 that each element of the reference system be fairly representative of 

 the corresponding component of the telephone plant, if the ratings 

 are to be approximately additive. This has been taken into account 

 in the choice of the working reference system. Certain approxima- 

 tions are involved in the system of data outlined, but they are minor 

 and are justified in the interest of simplification of the method. 



Preparation of Effective Data 

 Data for preparing effective loss ratings have been obtained pri- 

 marily from repetition counts made during a series of special trans- 

 mission observations on calls between telephone employees in the 

 regular course of their business. These calls were made over special 

 facilities which permitted the variation of the circuit constants over 

 a wide range and the rating of the various conditions relative to each 

 other in terms of repetitions. During these tests, different types of 

 instruments were used and changes were made in the sidetone charac- 

 teristics of the sets, and the attenuation and type of trunk. Each 

 type of change covered rather completely the whole range found in 

 the present telephone plant and to some extent that expected in the 

 future plant, but it has been practicable to cover only a small portion 

 of the combinations of instruments and circuits which might be used 

 together in commercial service. 



In the preparation of the necessarily large quantities of effective 

 transmission data from the transmission observations, the principal 

 problem is to determine the rating of any complete circuit from the 

 limited number of complete circuits which have been rated directly. 

 The determination of these additional ratings is relatively easy if the 

 circuits can be described in terms of a few simple characteristics which 

 will serve as a basis for interpolating between the ratings obtained 

 directly from observations. The physical measurements which can 

 readily be made in the required quantity describe a circuit in a com- 

 plex manner, namely, in terms of the efficiency, at each frequency in 

 the voice range, of the several speech and noise transmission paths of 

 the complete circuit. These data must, therefore, be combined in 

 some way to give a relatively small number of parameters for describing 

 the circuit. 



The definitions of these parameters may be more or less arbitrary, 



