OPEN-WIRE CROSSTALK 43 



positions, it is necessary to know both the phase and magnitude of 

 the crosstalk coefficients. For practical purposes, however, the 

 coefficients may, in most cases, be regarded as algebraic quantities 

 having sign but not angle. 



The direct component of the total crosstalk coefficient may be 

 readily computed as discussed in Appendix A. If more than a very 

 few wires are involved, an exact calculation of the indirect component 

 is impracticable but a fair approximation may be obtained by the 

 method discussed in Appendix A. This method is used when a wire 

 configuration is under consideration but is not available for measure- 

 ment. 



As pointed out in 1907 by Dr. G. A. Campbell, an accurate calcu- 

 lation of the total crosstalk coefficient would involve determination 

 of the "direct capacitances" between wires of the test pairs. Since 

 these capacitances are functions of the distances between all combina- 

 tions of wires on the lead and between wires and ground, their calcu- 

 lation is usually impracticable. In the past, the crosstalk coefficients 

 were computed by a method proposed by Dr. Campbell which involved 

 measurement of the direct capacitances.^ 



The part of the coefficient due to the electric field was computed 

 from the "direct capacitance unbalance." The part due to the mag- 

 netic field was computed as discussed in Appendix A. When loaded 

 open-wire circuits were in vogue it was necessary to be able to separate 

 the electric and magnetic components of the coefficients. After 

 loading was abandoned this separation was unnecessary and it was 

 found more convenient to measure the total coefficients than to 

 measure the direct capacitances or dilTerences between pairs of these 

 capacitances. 



As previously discussed, in designing transpositions it is necessary 

 to compute the interaction type of crosstalk indicated by Fig. 2C, and 

 it is, therefore, necessary to have some coupling factor for use in this 

 computation. Such a coupling factor could, theoretically, be deter- 

 mined as indicated schematically by Fig. 12. The interaction crosstalk 

 between two short lengths of line would be measured by transmitting 

 on one pair and receiving on the other pair at the junction of the two 

 short lengths as indicated by the figure. 



If there were but a single tertiary circuit such as c of the figure, 

 the crosstalk measured would be that due to the compound crosstalk 

 path fiacncb- In this product, Uac is the near-end crosstalk between a 

 and c in the right-hand short length d and rich is the near-end crosstalk 

 between c and h in the left-hand short length. Since nac and rich when 



^See papers by Dr. Campbell and Dr. Osborne listed under "Bibliography." 



