374 



BELL SYSTEM TECHNICAL JOURNAL 



Reference to equations (6) and (9) show that this tendency to cancel 

 is to be expected provided the second term of (6) is the controlHng 

 term in Fu Indeed, in lengths where the tertiary is electrically long, 

 equations (8) plus (9) should exactly cancel the second term of (6). 

 In other words, the total interaction crosstalk between two such sections 

 should cancel a portion of the interaction crosstalk within a section. 

 Since the portion which is cancelled is the controlling term the net 

 result is that when two sections are combined the total crosstalk in 

 length 2/ is no more than was measured in length /, as evidenced by the 

 measured curve Fi -\- Fnn + {Fi-\- F//) of Fig. 5. 



This effect persists when two 3000-foot lengths are combined to 

 form a 6000-foot section, as illustrated by the curves of Fig. 6. Here 



105 



115 



120 



M 125 



Z 130 



< 135 



2 140 



40 50 60 



80 100 150 200 300 400 500 600 800 1000 



FREQUENCY IN KILOCYCLES PER SECOND 



Fig. 6 — Crosstalk components in 3000-foot and 6000-foot lengths. 



again {Fi + F//) and Fnn are considerably smaller in magnitude than 

 Fi so that the total crosstalk in 6000 feet cannot differ materially from 

 the value Fi measured in 3000 feet. 



The curves labelled AB and BA were made by using first coaxial A 

 and then coaxial B as the disturbing circuit. The difference between 

 the curves indicates that there is a certain amount of random un- 

 balance within the section. For example, random deviations in the 

 shielding of the two coaxials from a nominal value would result in 

 different values of interaction crosstalk when the disturbed and dis- 



