424 SUBMARINE CABLE LAYING AND REPAIRING. 



carbon key is put down, lifting at the same moment the zinc 

 key with the same hand, while a second or two later the 

 galvanometer key is opened by the other hand. If the spot 

 indicates " less," the galvanometer key is released, carbon 

 key lifted and zinc put on again. After suitably reduc- 

 ing the resistance the same thing is repeated. When the 

 correct balance has been obtained the spot will be seen to move 

 slightly in the direction for " less " and immediately reverse, 

 moving off the scale the other way, this effect being the rise due 

 to sealing up with positive current. Pairs are taken in this 

 way until there is little or no change and the results become 

 uniform. 



The mean of all the zinc readings should be taken, and the 

 mean of all the carbons and the arithmetical mean of the pair 

 so obtained taken as the result, provided there is not much 

 difference between the zinc and carbon means. If the differ- 

 ence is considerable the arithmetical mean does not yield the 

 correct result. In Fisher and Darby's valuable work " The 

 Students' Guide to Submarine Cable Testing" (Ist edition, 

 page 47), it is shown that the correct result in bridge tests 

 with readings widely different is not yielded by either the 

 geometric or harmonic means, but must be found by Kempe's 

 correction, the application of which is clearly explained. The 

 difference in the readings is due to earth current in the cable, 

 the effect of which is to oppose one direction of testing current 

 and assist the other. When the currents are in opposition the 

 line current is less than when the directions coincide, and if a 

 milammeter is available the ratio of the currents to line can 

 be readily seen by connecting it in circuit. A simple correc- 

 tion can then be applied as explained in the note on Reversals, 

 page 505. 



The resistance of the break or fault can be estimated by 

 observation of the rate of polarisation. If the rise in resis- 

 tance due to sealing with positive current is slow as indicated 

 by the rate of movement of the spot, the area of exposure is 

 large and the resistance of the break or fault low, of the order 

 of 50 ohms or so ; but if the rate of polarisation is rapid the 

 resistance of the break may be a few hundreds. It requires 

 considerable experience to estimate the resistance with mini- 

 mum error. The estimated resistance of the break should 



