394 BELL SYSTEM TECLINICAL JOURNAL 



doubles the amplitude of the arriving voltage, thus giving larger 

 quantities to work with, and second, that it eliminates the necessity 

 of taking into account the characteristic impedance of the cable and 

 the impedance of the balanced type of sea earth which is usually 

 employed as the earth connection of the amplifier. By means of a 

 string oscillograph in the output of the amplifier, the wave shape of 

 the received voltage and the nature of the extraneous interference can 

 be determined. The amplifier is calibrated by impressing on it a 

 measured voltage of the same frequency as that of the received voltage. 



Knowing the values of received voltage and the corresponding 

 transmitted voltage, the values of attenuation constant can be readily 

 computed. By plotting the values of attenuation constant corre- 

 sponding to various values of frequency and transmitted voltage as 

 functions of the latter quantity and extending these curves to the 

 axis of zero transmitted voltage, the values of attenuation constant 

 corresponding to a very small current in the conductor can be obtained 

 for various frequencies. 



Assuming that all the parameters have been accurately pre- 

 determined, there are three sources of error which might possibly 

 cause a difference between the measured value of attenuation constant 

 and that computed from the average values of the cable parameters 

 by means of formula (2). In the first place, the parameters are not 

 uniform throughout the cable as is assumed in deriving this formula. 

 In particular, the inductance may vary from point to point. At each 

 point where the capacity or inductance changes value reflections of 

 voltage and current will take place and the effect of these reflections 

 should be to increase the attenuation constant of the cable. For 

 variations of the parameters of the order that is to be expected in 

 loaded cables, the increase in attenuation constant is quite small, 

 and the magnitude of this increase can be computed approximately 

 by a method due to Carson.^ Another source of error is the presence 

 of extraneous interference superposed on the received voltage. This 

 factor is usually troublesome only at the highest frequencies and 

 lowest voltages employed, and in this case measurements of the 

 oscillograms of received voltage and of calibrating voltage will give a 

 value of the received voltage independent of interference. The third 

 source of error is due to the presence in the transmitted voltage of 

 harmonics of the fundamental frequency. These harmonics are 

 attenuated in transmission over the cable to a much greater degree 

 than is the fundamental, so that they constitute only a small per- 



6 Electrician, Vol. 86, p. 272, 1921. 



