EFFECT OF SIGNAL DISTORTION IN MORSE TELEGRAPHY 269 



change in length in the signal elements which will cause errors in re- 

 ception can generally be determined from the construction of the 

 machine. It has a fairly definite value for a particular type of machine 

 and is largely independent of the type of distortion which produces it. 



In the case of Morse telegraphy where the signals are received by 

 listening to the clicks of a sounder, the effect of distortion is more 

 complicated. Morse operators do not interpret signals entirely by 

 the length of individual signal elements as does a telegraph printer. 

 They interpret them by the general sound of the clicks due to the 

 succession of dots and dashes making up a particular letter or word of 

 the code. Consequently, it is to be expected that for the same total 

 change in length of signal elements caused by one or more of the three 

 types of distortion, the effect upon operators will differ, depending 

 upon the type of distortion and upon the particular operators who are 

 receiving. It is this phase of Morse telegraph operation with which 

 the present paper deals. 



To carry out the investigation, namely, to obtain data on the effects 

 of various kinds and combinations of distortion on the accuracy of 

 reception by telegraphers and their opinions as to the satisfactoriness 

 of the received signals, several methods of procedure were suggested. 

 These were carefully considered and the following appeared to be the 

 most suitable. 



A circuit, simulating a commercial telegraph circuit, was to be con- 

 structed and arrangements devised for impressing any desired amount 

 of distortion or combination of distortion upon the signals transmitted 

 over this circuit. The manner of introducing the distortions was to 

 simulate as nearly as practicable the manner in which they would 

 occur on real lines and the methods employed for sending and receiving 

 the signals were to simulate closely actual operating conditions. 



As regards the type of test messages to be used by the operators, it 

 was thought that neither plain English nor unpronounceable code 

 would be satisfactory, the former because distorted letters and words 

 could be supplied by the operators from the context and the latter 

 because it would be unnecessarily difficult to send and receive. Con- 

 sequently, messages intermediate between these extremes were decided 

 upon and were obtained by using text from a foreign language with 

 which the operators were unfamiliar. 



For sending the messages, a semi-automatic key (vibroplex), care- 

 fully adjusted to be unbiased and to operate at a speed of 13.5 d.p.s. 

 (dots per second) was provided. Although this speed of operation 

 corresponds to fairly rapid Morse sending, the rate of transmission of 

 words during the tests was fairly low, being about 25 five-letter words 



