14G4 THE BELL SYSTEM TECHNICAL JOURNAL, NOVEMBER 1957 



ine how they fare in appHcation over untreated Nl carrier. There have 

 been some studies of this point. The general conclusion, to be elaborated 

 below, is that although the frequency subdivision (and also the fre- 

 quency shift) helps against impulsive noise, fewer telegraph channels 

 may be used than over good non-compandored facilities; and at 

 best, the transmission is accompanied by more distortion than expected 

 in a telegraph link of the highest grade. However, a serious possibility 

 for data transmission is indicated. 



2.4.1 Telegraph Tests on Nl Carrier 



Tests on this subject have been carried out by S. I. Cory, J. M. Fraser 

 and others and reported in unpublished memoranda. Before presenting 

 the results of the tests, some background is necessary on the terms in 

 which the results are reported. 



The performance is usually evaluated in tests of this kind in terms of 

 a "maximum checkable" telegraph distortion over a short time (about 

 5 minutes). The "telegraph distortion" represents the displacement 

 from correct timing, of received signal transitions, after the initial 

 mark-to-space transition in the "start" element. These displacements 

 are measured in percentages of the signal element duration. By "maxi- 

 mum checkable" distortion is meant the maximum such displacement 

 that is consistently reproduced in repetitions of the short testing period. 

 This is somewhat larger than the root-mean -square distortion. A larger 

 displacement is, of course, obtainable over a longer testing period. 

 Although this measure of performance has had long use in the telegraph 

 art, other measures are perhaps more readily grasped by and probably 

 of more value to the data transmission engineer. Such a measure, for 

 example, is the error rate. 



The error rate may be estimated through the use of the telegraph 

 transmission coefficient.^ This is a figure which has been designed by 

 telegraph engineers to indicate the performance of a telegraph circuit, 

 particularly when it is made up of several sections. It is more or less 

 proportional to the square of the distortion, and has the property that, 

 when carefully chosen, it can be added for circuits connected in tandem. 

 A small coefficient thus characterizes good transmission, and correspond- 

 ingly, a large coefficient characterizes poor transmission. 



The correlation between peak distortion over 5-minute intervals and 

 error rate, through the telegraph transmission coefficient, is indicated 

 in Reference 8 and Table II. It is to be understood that the correlation 

 is only a rough one. Particularly at the two extreme ends of the scale, 

 the entries in the table can serve only as a general guide to the perform- 

 ance, and the specific numbers are not to be taken too literally. 



