May 2, 1 901] 



NA TURE 



be actuated by the fundamental lower harmonics, and 

 the sound it gives out will, in consequence, be different in 

 character from the sound originally made at the trans- 

 mitting end. The effect will show itself, therefore, in 

 defective articulation, or distortion of the sounds arising 

 out of the distortion of the telephone currents. 



It has been shown by Mr. Oliver Heaviside that there 

 are ways in which this distortion may be prevented and 

 a "distortionless circuit " constructed. Without entering 

 too deeply into the subject we may point out briefly the 

 methods by which this may be effected. Since the cable 

 possesses capacity, the first effect of sending current into 

 it is to charge it, and no signal can be received at the far 

 end until the cable is partly charged, and no further 

 signal until the charge has had time to get out. Now if 

 the insulation resistance of the cable be diminished, the 

 charges will more readily leak out and thus it would be 

 possible to expedite signalling ; but at the same time the 

 attenuation is increased, for more of the current will 

 leak out of the cable ; the remedy is, therefore, only a 

 partial one, for though the speed of signalling may be 

 increased, so much current will leak out on the way that 

 the amount arriving at the far end may be too small to 

 work the receiving instruments. Instead of simply 

 diminishing the insulation resistance or of distiibuting 

 artificial non-inductive leaks along the cable, inductive 

 leaks may be placed at definite points along the cable ; 

 this method was proposed by Prof S. P. Thompson in a 

 paper read at the International Congress at Chicago in 

 1893.' A diagram of the cable construction suggested 

 by Prof Thompson is shown jn Fig. 7 ; the capacity is 

 represented as though it were not evenly distributed but 



consisted of a number of condensers, c,c, connected as 

 shunts to the cable ; the inductive leaks are represented 

 by the coils l,l. The capacity and self-induction are 

 therefore combined in parallel, and it is well known that 

 they can be combined in this way so as to behave, for a 

 definite frequency, exactly as an ohmic resistance. The 

 capacity of a submarine cable may be partially neutralised 

 in this way, but the remedy is only a partial one for three 

 reasons. Firstly, the inductive leaks, to correctly neu- 

 tralise the capacity, should, like the capacity itself, be 

 evenly distributed along the cable and not distributed in 

 jerks ; secondly, the correction will only be exact for a 

 particular frequency ; lastly, the leakage is increased and 

 the same defect consequently occurs as in the case con^ 

 sidered above in which the distortion was corrected by 

 diminishing the insulation resistance. Theoretically, 

 therefore, the system proposed by Prof. Thompson does 

 not offer a perfect solution or give a truly distortionless 

 circuit ; but it would greatly diminish the distortion, 

 though at the same time increasing the attenuation, and 

 might therefore give a practical means of increasing the 

 speed of signalling or even obtaining telephonic com- 

 munication over the cable. 



As Mr. Heaviside has shown, the only true way of 

 obtaining a distortionless circuit — of obtaining the dis- 

 tortionless circuit, as he calls it — is to balance the effect 

 of capacity by self-induction distribuled aloiii; the cable 

 in series with it and not as a leak to it. The four 

 quantities which control the propagation of disturbances 

 or signals along the line are its resistance, R, its external 

 conductance, or conductivity of the insulation, K, its self- 

 induction, L, and its capacity or " permittance," S, and 

 the signals will be propagated without distortion if 



1 See the Electrician, August iSq3, p. 439. 



NO. 1644, VOL. 64] 



l/r = s/k. The equality of these two ratios may be^ 

 obtained by altering any of the four variables, but 

 practically we may consider r and s as fixed. Inordinary 

 cables the value of the ratio i, r is very small, and that of 

 s,K comparatively large. In order to make the two 

 equal we may increase k, that is to say diminish the 

 insulation resistance, but this, as we have seen, leads to 

 excessive leakage and is not, therefore, desirable. The 

 method suggested by Prof Thompson amounts practically 

 to converting the capacity, s, partly or wholly into insula- 

 tion conductivity, K, and thus diminishing s K until it is 

 as small as LR. The self-induction coils added in this 

 system must not be confounded with the self-induction of 

 the cable L, for they are added as shunts to the cable. 

 The ratio l/R may also be made equal to S; R by adding 

 self-induction coils in series with the cable, thus increasing 

 the value of L ; this is the solution adopted by Dr. Pupin. 

 Here again the ideal solution is only obtained when the 

 self induction is evenly distributed, but a practical solutioR 

 can be obtained by placing coils at intervals along the 

 cable. 



Dr. Pupin, besides repeating a good deal of Mr. 

 Heaviside's theoretical investigations, worked out the 

 necessary values of the self-induction of the coils and the 

 maximum distance apart at which they can be placed in 

 order to imitate sufficiently well an evenly distributed 

 self induction. He then proceeded to build some coils 

 and to experiment with them on an artificial cable. The 

 results of some of these experiments are interesting, as 

 they point to the great improvement the addition of the 



inductance produced. An artificial cable was built up 

 with condensers in the usual way in 250 sections, each 

 section representing a mile of cable ; between each section 

 were placed induction coils which could be short-circuited 

 by plugs. A diagram of this cable is given in Fig. S ; as 

 before, the capacity is represented as if it consisted of 

 condensers, C,C ; the induction coils are shown at L,L ; 

 these coils are short-circuited by inserting the plugs at 

 the contacts p,p. When all the coils were in circuit 

 telephonic communication could be carried on with 

 perfect ease over the whole length, 250 miles, of the 

 cable ; when, however, the coils were short-circuited 

 conversation was good up to 50 miles only, fair up to 75, 

 impracticable at 100 and impossible beyond 112. It 

 must be remembered in considering these results that 

 the cable was an artificial one and that possibilities of 

 error are consequently great, so that the results must not 

 be transferred v/ith too much confidence to the case of 

 an actual cable. 



Apart from this, however, the results are extremely good, 

 and Dr. Pupin is to be congratulated on having obtained 

 experimentally a practically distortionless circuit. It is 

 perfectly true, no doubt, that Mr. Heaviside had obtained 

 the solution already theoretically ; but the engineers 

 generally require to have their attention attracted by 

 actual experiment and are not too prone to make changes 

 on a theoretical basis only, however sound. Whether a 

 cable can be commercially constructed on the lines of 

 Dr. Pupin's artificial cable is a question for the practicians ; 

 we have no doubt that, now its advantages have been 

 demonstrated, they will be able to find a way. The 

 enormous advantage of Transatlantic telephony can never 



