360 BELL SYSTEM TECHNICAL JOURNAL 



visualize clearly the influence of the tertiary circuits in the summation 

 process. To produce such a picture a certain amount of review of the 

 general crosstalk problem will be necessary. This is undertaken in 

 Part I of this paper. 



Part II is devoted mainly to the presentation of test data taken in 

 November and December, 1937, January and February, 1938 on sec- 

 tions of a five-mile length of a twin coaxial cable near Princeton. 

 These data confirm and graphically illustrate certain relationships 

 developed in Part I. In addition they provide information on the 

 tendency of tertiary circuits to complicate the effectiveness of trans- 

 positions and show how interaction crosstalk takes place around 

 repeaters via the tertiary circuits. 



PART I— THEORY 



In any series of crosstalk tests on short lengths of paired or quadded 

 cable where the problem of combining a number of such lengths is 

 concerned it has generally been the practice to terminate both the test 

 circuits and important tertiary circuits in characteristic impedance. 

 Under such a condition the normal influence of all circuits in the pro- 

 duction of crosstalk within each short section is provided for and the 

 summation process, including interaction between successive sections, 

 can be studied under actual line conditions. This is a general method 

 applicable to any type of coupling and was adopted for the Princeton 

 investigation. The effect of discontinuities such as short-circuited 

 tertiaries at the extreme ends of a repeater section can be readily 

 handled mathematically as correction terms due to "end effect." 



To simplify the presentation of the factors involved, the discussion 

 in this section will be confined mainly to the case of far-end crosstalk. 

 In a twin coaxial cable where the transmission in the two units is in 

 opposite directions there actually exists no far-end crosstalk problem 

 since only talker echo, a near-end crosstalk phenomenon, is in- 

 volved.^ In multi-unit cable, however, there will be far-end crosstalk 

 between different systems. Since this type of crosstalk tends to in- 

 crease directly with the number of repeater sections it is important to 

 understand its nature thoroughly. Moreover, in a study of funda- 

 mentals it is possible to avoid certain complications not essential to an 

 understanding of the problem by investigating far-end rather than 

 near-end crosstalk. 



To present a clear picture of the physical meaning of some of the 

 forthcoming mathematical expressions their derivations will be ap- 



' This statement may not hold if the repeater impedances fail to match the line 

 impedance since in that case the far-end crosstalk can be reflected and appear as 

 near-end crosstalk. 



