270 
THE TELEGRAPH. 
positive current, because the signals are better, the currents 
are more intense and less liable to change from the effects of 
the fault. This practice is, however, reprehensible, for a 
defective cable should be used with negative currents only. A 
defective line may be used for an indefinite time, so long as 
the conductor remains continuous. 
Sometimes faults of this kind are caused by the presence of 
foreign bodies in the insulating material. If a fragment of 
zinc or a piece of wire should establish a communication 
between the conducting core and the external covering of the 
cable, a fault would be produced having a resistance almost 
nil, and the nature of the fault would readily be recognized by 
the absence of polarization. 
A fault of the second kind (solution of continuity) may 
easily be combined with one of the first kind ; for the line 
may not only be broken, but also be in more or less perfect con¬ 
tact with the earth at the place of rupture. In this case it is 
impossible to operate simultaneously at the two extremities. 
We can only measure the resistance of each section of the 
cable, and estimate, according to the polarization, what 
fraction of the total observed resistance must be attributed to 
the fault itself. In this case, we can with certainty determine 
the maximum distance within which the fault must be situated. 
With the minimum of polarization, the denuded copper wire 
of a ruptured cable usually gives a resistance equal to several 
miles of the conducting wire. 
Faults of this kind also often present themselves as a solu¬ 
tion of continuity of the conductor, while the insulation at the 
point of rupture continues good. In a submarine cable, the 
distances of such faults may be very accurately measured by 
finding the capacity of the portion of the cable between the 
coast and the rupture. The capacity per unit of length 
being known, the testing shows the distance with great 
precision. It is scarcely possible to apply this method to 
overhead lines, as in them the insulation is rarely sufficiently 
good to give exact results. 
Faults of the third kind (contact of two adjoining con¬ 
ductors) are easily discovered if the contact is local and has but 
a small resistance. In this case, the resistance of the loop 
produced by the contact is measured, and half of this will 
plainly correspond with the distance of the fault. 
