\\-II.UAM SIEMENS, F.R.S. 47 



case we shall have to measure the insulation resistance, and get 

 our correspondent at the other end to do the same, and communi- 

 cate the result through the cable. We embody these results in 

 two formulas as before, from which we eliminate the resistance of 

 the fault, and establish the numerical relation between x and y, 

 representing the resistances on both sides of the fault, for which 

 we may put lengths of cable, supposing the conductor to be 

 uniform. 



Another description of fault where it is more difficult to operate 

 is, when the cable is ruptured altogether, and you can conse- 

 quently no longer speak through it. In that case we have to deal 

 with the resistance of the cable and the resistance of the fault, or 

 place of contact between the conductor and the sea, which you 

 cannot eliminate. We can, however, approximately determine the 

 resistance due to the fault in observing the amount of polarisa- 

 tion produced by a current of a certain duration. If there is 

 much copper exposed, there will be much polarisation, and, there- 

 fore, little resistance in the fault itself, and vice versd. And thus 

 by careful comparison and analysis we can approximately deter- 

 mine the resistance of the fault and of the remaining resistance 

 which determines the position of the fault. 



Another case arises if a cable is ruptured, but the gutta-percha 

 completely overlaps, and insulates partially the end of the broken 

 wire. In that case we operate by measuring the capacity for 

 charge of the cable, in connecting the cable with a powerful 

 battery for a moment, and suddenly disconnecting it and making 

 connection to earth through a galvanometer, when the needle is 

 bent over to a certain angle ; and from that angle of deflection 

 we can calculate, as before described, the importance or capacity 

 of the Leyden jar formed by the cable to the point of dis- 

 continuity. It is true that there is leakage ; but by measuring 

 the charges, and also the discharges, we find, in the difference, an 

 expression for this leakage, and we can, by taking the mean 

 between the two, determine the exact capacity of the cable 

 remaining, and, therefore, also its length, having determined its 

 specific capacity beforehand. 



Another case, an analogous one, which occurs sometimes, is 

 that the copper wire alone is separated, but the gutta-percha or 

 other insulating covering remains intact. There we have the jar 



