MEASUREMENT OF TELEGRAPH TRANSMISSION 153 



nism controlled by the start pulse (not shown), it travels in the direction 

 of the arrow and the condenser charges vary as indicated by graphs 

 B and C. It will be seen on graph B that condenser CI is charged 

 during intervals between grounded segments and mark-to-space 

 transitions. Graph C shows that condenser C2 is charged between 

 the grounded segments and space-to-mark transitions. If there is 

 no transition while the brush is traveling between two adjacent 

 grounded segments the condenser continues to be charged until the 

 brush touches the second grounded segment at which time it is com- 

 pletely discharged. Therefore, the useful charging interval is that be- 

 tween a grounded segment and a transition occurring before the next 

 grounded segment is traversed by the brush. These intervals are 

 indicated by the heavy-lined portions of the graphs. They amount 

 to 37.5 per cent of a unit pulse as a maximum, i.e., the maximum 

 distortion which can be measured is about 37.5 per cent. The 

 currents flowing as a result of the comparison of the condenser voltages 

 with the reference voltage are indicated on a "Total-Distortion 

 Meter" TD which is, in reality, a peak-indicating voltmeter, and on 

 a "Bias Meter," B which is sufficiently sluggish to give an indication 

 corresponding to the average distortion. These meters are calibrated 

 to indicate directly the percentage distortion with miscellaneous 

 teletypewriter characters. 



Good accuracy is obtained with these sets; when measuring distor- 

 tions of small or moderate amounts with a well-adjusted set, the 

 indication is accurate to within about 2 per cent distortion at 60 words 

 per minute. For occasional large distortions or for higher speeds, 

 the accuracy is not quite as good, and there are certain possible mutila- 

 tions of signals, such as the dropping out of pulses, which would not be 

 readily detected. 



In addition to measuring miscellaneous teletypewriter characters 

 these sets may be used with recurring test signals in which the spacings 

 of the transitions are such that the maximum characteristic effects 

 will be obtained, and with signals which experience mostly bias and 

 fortuitous effects. In this way a measure of the components of 

 distortion may be obtained. Such tests are commonly made in 

 adjusting variable networks to minimize characteristic distortion, i.e., 

 making the equalization tests, referred to above, with selected tele- 

 typewriter signals. 



In special testing where it is desired to separate the total distortion 

 into its components, this may be done by measuring the systematic 

 distortion with the first 6 signals of Fig. 1 and then measuring the 

 bias and total distortion using the / character of Fig. 1 (which is sub- 



