272 BELL SYSTEM TECHNICAL JOURNAL 



the condenser d, at the beginning of a particular signal element, was 

 determined largely by the length of the mark or space immediately 

 preceding this element and to some extent by the ones preceding that. 

 As a result, this mark or space was lengthened an amount depending 

 upon its position in the signal combination, thereby simulating the 

 effect of positive characteristic distortion. It was necessary to intro- 

 duce this type of distortion in two sections when values of distortion 

 greater than about 35 per cent were desired. This was accomplished 

 by connecting winding W^ of relay A in a holding circuit similar to 

 that of winding W^ of relay Dz. 



Fortuitous distortion was introduced by means of a telegraph cross- 

 fire arrangement. It consisted of the neutral relay D^ whose windings 

 were connected into a local circuit and operated by means of a mechan- 

 ical Morse code transmitter. The contacts of the relay had positive 

 and negative batteries connected to them and the armature was con- 

 nected to the telegraph circuit by means of condenser d and series 

 resistance Ri. The magnitude of the maximum fortuitous distortion 

 was determined by the magnitude of the crossfire impulses, this being 

 controlled by means of the condenser and resistance. Since the com- 

 bination of the crossfire impulses with the telegraph signals was 

 erratic, it is obvious that positive and negative distortions ranging 

 from zero up to the maximum value were obtained at one time or 

 another. 



The methods described above for distorting the signals simulated 

 closely, except possibly in the case of positive characteristic distortion, 

 the conditions which occur on actual telegraph circuits. For this 

 reason, the results obtained in the tests may be taken as reasonably 

 representative of results which would be obtained on actual circuits. 



4. Method of Making Operating Tests 



Two series of tests were made. The first series covered all types of 

 distortion individually and in combinations while the second series 

 covered only those individual types of distortion the efi^ects of which, 

 as shown by the first series of tests, appeared questionable and for 

 which check tests were desired. 



Testing was done on alternate days between the hours of 9 a.m. and 

 4 p.m. Six operators were available, two for sending and the remain- 

 ing four for receiving. These were divided into two groups and ap- 

 peared alternately for the tests. 



The method of testing was briefly as follows : the sending operator 

 was provided with a vibroplex key and with copies of messages to be 

 transmitted. These messages were taken chiefly from Hungarian 



