42 BELL SYSTEM TECHNICAL JOURNAL 



Curves C show the coefficients with the test pairs without transpo- 

 sitions and terminated at both ends as accurately as practicable, but 

 without the midpoints of these terminations connected to ground to 

 terminate the phantom and ghost circuits. These tertiary circuits 

 were, with this arrangement, prevented from connecting points of 

 substantially different potential and the coefficients of curves C, 

 therefore, approach the direct crosstalk coefficients. It is extremely 

 difficult to experimentally determine the direct far-end coefficient. 

 It may be computed, however, and the computed value which assumes 

 perfect terminations and the effect of the phantom completely removed 

 is shown by curve C. 



It may be noted that the near-end crosstalk coefficients are about 

 independent of frequency. This is ordinarily true above a few 

 hundred cycles. The total far-end coefficient (curve A) is about 

 independent of frequency in the important carrier frequency range. 

 The direct far-end coefficient of curve C decreases considerably with 

 frequency for reasons discussed in Appendix A. Since transpositions 

 are ordinarily designed for the condition of a number of wires on a line, 

 the total crosstalk coefficient is the one usually used in practice. 



Curves C of Fig. 11 also indicate that the direct near-end coefficient 

 is much larger than the direct far-end coefficient. This is usually 

 true and, as discussed in detail in Appendix A, the explanation is that 

 the crosstalk currents caused by the electric and magnetic fields add 

 almost directly in the case of direct near-end crosstalk but tend to 

 cancel in the case of direct far-end crosstalk. As discussed in the 

 appendix, the indirect (vector difference of curves A and C) crosstalk 

 in a very short length is due almost entirely to the electric field of the 

 tertiary circuits and is the same for both near-end and far-end crosstalk. 

 In Fig. 11, the total near-end coefficient (curve A) is increased by the 

 indirect crosstalk since curve C is lower than curve A. The reverse 

 is usually true, however. In the case of far-end crosstalk the total 

 coefficient is usually increased by the indirect crosstalk. 



Crosstalk coefficients are vector quantities and may be measured in 

 magnitude and phase. If it is desired to compute the crosstalk 

 between two long pairs of wires which do not change their pin positions, 

 it is only necessary to know the magnitude of the crosstalk coefficient, 

 since the problem is to determine the ratio of the crosstalk for many 

 elementary lengths to the crosstalk for one such length. However, 

 if it is desired to know the crosstalk between long circuits which do 

 change their pin positions, several crosstalk coefficients must be known, 

 one for each combination of pin positions. In order to determine the 

 total crosstalk for several segments of a line involving different pin 



