386 BELL SYSTEM TECHNICAL JOURNAL 



Improving the Line 



When the development work for the transcontinental project started it 

 was realized on the basis of the earlier work that impedance irregularities in 

 the existing types of loaded circuits were large enough to set objectionable 

 limitations upon the gains obtainable with two-way telephone repeaters. 

 The recent availability of the Vreeland mercury arc, variable frequency, 

 oscillator had made it possible to get a considerable number of impedance- 

 frequency curves at close frequency intervals, but as yet the specific irregu- 

 larities in the impedance curves had not been correlated with their individual 

 causes. This prior work had been concentrated on loaded cables. 



For a considerable time the new line studies were concentrated on loaded 

 open-wire lines. Since the spacing irregularities were known to be as great 

 or greater than in the cables, and since other sources of irregularity such as 

 intermediate and entrance cables were also present, it was expected that the 

 open-wire impedance-frequency curves would be even more irregular than 

 the cable curves. And such was found to be the case, but in a much greater 

 degree than anticipated. 



Inductance Irregularities: The discussion will initially be directed to the 

 inductance irregularities, since they proved to be the principal problem. 

 In the course of the line measurements it happened that one set of im- 

 pedance curves had a very unusual systematic sequence of ups and downs 

 with rising frequency. This was especially intriguing since the usual 

 curves had non-systematic bumpy characteristics. The cause was found 

 to be an omitted load at a particular load point. This incident resulted in 

 the development of a formula for estimating the position of an impedance 

 irregularity in terms of the frequency spacing of resulting bumps in the im- 

 pedance-frequency curve and the velocity of transmission. 



A comprehensive series of impedance measurements were then started 

 on a long loaded artificial cable at the laborator}', since it was much more 

 simple to measure than to compute. Also, the magnitudes and circuit 

 position of individual and multiple irregularities could easily be controlled. 

 Very valuable data were collected in this manner. 



When the loaded open-wire measurements were resumed, it was noticed 

 that the changes in the impedance-frequency irregularity patterns of 

 particular lines changed substantially from time to time. These changes 

 were found to be due to large inductance changes, up or down, in individual 

 loading coils, and the cause was eventually found to be the magnetizing or 

 demagnetizing action of strong transient line currents induced by lightning 

 discharges. In lines exposed to lightning, sooner or later the inductance of 

 all exposed coils would drop well below the factory value. A coil thus par- 

 tially magnetized by one shock would sooner or later be partly demagne- 

 tized by a subsequent shock and these experiences would be repeated again 



