146 BELL SYSTEM TECHNICAL JOURNAL 



It is of interest to consider the case of bias alone. With uniform 

 bias the displacement of mark-to-space transitions is in one direction 

 and that of the space-to-mark transitions is in the other direction with 

 respect to their positions in undistorted signals. Since the start 

 transition is mark-to-space the result is that the effective displacement 

 of subsequent mark-to-space transitions is zero and the efTective dis- 

 placement of space-to-mark transitions is numerically equal to the 

 bias. In practice bias is seldom uniform and may vary with the 

 signal combinations. In these cases there is an effective displacement 

 of mark-to-space as well as space-to-mark transitions. 



From the foregoing, it will be seen that it is of considerable practical 

 value to be able to measure teletypewriter circuits on a start-stop 

 basis in terms of displacement of transitions with respect to the start 

 transition. (For a more complete explanation of the effect of dis- 

 tortion on teletypewriter operation, reference should be made to 

 published discussions.** ^' ^) In testing with miscellaneous signals, 

 bias may for convenience be taken as the average effective displace- 

 ment of space-to-mark transitions relative to mark-to-space transitions; 

 characteristic distortion will have the appearance of a combination 

 of fortuitous and bias effects; and the maximum total distortion will 

 be the sum of the average effect and the variation therefrom which 

 causes the greatest displacement. 



Other Changes in Measuring Technique 

 In measuring with normal and inverted signals ^ on circuits of the 

 types commonly employed, the result obtained for the bias varies 

 somewhat from pulse to pulse of the test signal. A case in which this 

 variation is appreciable is that of carrier telegraph having level 

 compensators (automatic devices which correct for changes in the 

 magnitude of the received current). With these compensators, the 

 response is fairly rapid, the result being that the bias is to some extent 

 a function of the signal combinations of the transmitted material. 

 This bias variation is also noticeable with open-and-close d-c. telegraph 

 circuits having large bridged capacitance or series inductance. 



On account of this bias variation, it is desirable to measure the 

 algebraic average of distortion of the individual pulses of miscellaneous 

 signals and take this as the bias. This may be conveniently done by 

 measuring on the start-stop basis mentioned above. In such measure- 

 ments the differences between the distortions of the individual pulses 

 and the average distortion may be considered as due to the combination 

 of characteristic and fortuitous effects; further measurement is 

 necessary in order to separate these effects. A measure of character- 



