258 BELL SYSTEM TECHNICAL JOURNAL 



accomplished by connecting small condensers made up of twisted 

 pairs, between wires of different cable circuits and/or by coupling wires 

 of different circuits together through small air-core transformers. 

 Each unit was individually adjusted after measurement of the cross- 

 talk between the various combinations. 



Maintaining Stability of Transmission 



Referring to the problem of stability, the importance of this will 

 be appreciated from the fact that the average attenuation at the 

 carrier frequencies employed in the 850-mile circuit as set up at 

 Morristown w^as about 1300 db. A circuit was actually set up and 

 tested consisting of nine of the carrier links in tandem, giving 7650 

 miles of two-way telephone circuit whose total attenuation without 

 amplifiers was about 12,000 db. This attenuation, on an energy basis, 

 amounts to 10^^°?. This ratio, representing the amplification necessary, 

 quite transcends ratios such as the size of the total universe to the 

 size of the smallest known particle of matter. 



Balancing this huge amplification against the correspondingly huge 

 loss, to the required precision, one or two db, is a difficult problem. 

 Fortunately, a new form of amplifier employing the principle of 

 negative feedback has been invented by Mr. H. S. Black of the Bell 

 Telephone Laboratories and may be described later in an Institute 

 paper. By making use of this negative feedback principle, amplifiers 

 were produced for this job giving an amplification of 50-60 db and 

 this amplification did not change more than .01 db with normal battery 

 and tube variations. This is ample stability even when it is con- 

 sidered that, with amplifiers spaced 25 miles apart, there would be 160 

 of these in tandem on a circuit 4000 miles long. 



As is well known, the losses introduced by cable circuits do not 

 remain constant even though the circuits are kept dry by means of 

 the airtight lead cable sheaths. Variation in temperature is principally 

 responsible for the variation in efficiency of the circuits. The change 

 in temperature, of course, alters the resistance of the wires and to a 

 lesser extent changes the other primary constants, particularly the 

 dielectric conductance. Fig. 6 shows the transmission loss plotted 

 against frequency of a 25-mile length of 16-gauge cable pair at average 

 temperature (taken as 55° F.) and also the effect of changing this 

 temperature ±18° F. which is about the variation experienced in 

 underground cable in this section of the country. For a circuit 1000 

 miles long the yearly variation amounts to about 100 db. 



The transmission loss at any frequency is a simple function of the 

 d-c. resistance. Consequently, measurement of the d-c. resistance of a 



