40 BELL SYSTEM TECHNICAL JOURNAL 



far-end crosstalk) this total transverse coupling varies about directly 

 with the frequency and the crosstalk coefficient commonly used is the 

 crosstalk per mile per kilocycle. 



If many wires are involved, it is impracticable to determine these 

 coefficients with good accuracy by computation and they are, therefore, 

 derived from measurements. Examples of near-end and far-end 

 coefficients, plotted against frequency, are shown in Fig. 11. The 

 coefficients are for pairs designated 1-2 and 3-4 on the pole head 

 diagram shown on the figure. These coefficients were derived from 

 measurements of the near-end and far-end crosstalk over a range of 

 frequencies. The length of line was about .2 mile and, for the range 

 of frequencies covered, this length is sufficiently short so that inter- 

 action crosstalk is negligible and the transverse crosstalk is directly 

 proportional to the length. The coefficients plotted are, therefore, 

 nearly equal to the measured values of crosstalk divided by the length 

 and by the frequency. (A small correction was made at the higher 

 frequencies to allow for deviation of near-end crosstalk from simple 

 proportionality to length and the curves were "smoothed" through 

 the actual points calculated from the measurements.) 



In order to obtain the crosstalk coefficients applicable to a short 

 part of a long line, all the wires on the line were terminated in such a 

 manner as to roughly simulate their extension for long distances in 

 both directions, but without crosstalk coupling between the test pairs 

 in such extensions. This is done by terminating each pair at each 

 end with a resistance approximating its characteristic impedance and 

 connecting the midpoint of each resistance to ground through a second 

 resistance. These latter resistances terminate any phantom of two 

 pairs as indicated on Fig. 11 for pairs 1-2 and 3-4. Any circuit with 

 ground return is also terminated by these resistances. 



Both of the test pairs are transposed at the midpoint of the line 

 during the measurement. This minimizes the currents reaching the 

 ends of the tertiary circuits and makes even the above approximate 

 termination of the tertiary circuits of little importance. 



Figure 11 shows near-end and far-end crosstalk coefficients for three 

 conditions, A , B, and C. The two curves marked A show the measured 

 values with all wires terminated and the test pairs transposed as 

 described above. 



For curves B, only the transposed test pairs were terminated as 

 described above and the other wires were opened at the middle, at the 

 quarter points and at both ends. Since no section of any of these 

 wires connected points of substantially different potential in the field 

 of the disturbing circuit there were practically no currents or charges 



