404 THE ELECTRIC TELEGRAPH. 



be required to keep his eye upon the galvanometer, which 

 in laying long cables is very fatiguing, we proposed to use 

 in the paying out of one of the Mediterranean cables an 

 arrangement by which, if the insulation fell below a certain 

 value, the needle of the galvanometer would be deflected and 

 make contact with a fixed point, closing the circuit of a 

 delicate relay and small battery. The relay was in turn to 

 close the circuit of an alarm, which was to give instant notice 

 of the occurrence of a fault, and prevent such accidents as 

 allowing one to leave the ship for two or three hours before 

 the electrician is a.ware of it, and when its recovery may be 

 attended with the entire loss of the submerged part of the 

 cable. 



101. Determination of the Position of a Fault in the Insula- 

 tion. When a cable has been sheathed with wet hemp, 

 tested daily during the manufacture under water, and kept 

 wet on board the ship, it is almost an impossibility that a fault 

 can leave the ship before being discovered by the electrical 

 engineer. Formerly, when a cable was made up dry, coiled 

 dry in the hold of the ship, and first wetted in entering the sea 

 on being submerged, it was to be expected .that, at this 

 moment, when too late to repair it with facility, the first in- 

 dications should be given of faulty insulation. In this way, 

 many a cable that appeared moderately well insulated when 

 on board the ship, proved to be very bad on leaving it. 

 Although such occurrences are now impossible, faults still 

 occur in cables by abrasion against the bottom, by the line 

 being torn by anchors, and in other ways, after submersion. 

 In such cases, the determination of the exact position of the 

 fault is not an easy matter, on account of its varying resist- 

 ance and polarisation, and the changing direction of earth 

 currents. 



When the cable is laid, and therefore one end only at the 

 disposition of the operator, a fault may be determined by the 

 separate measurements made at both ends, or by those made 

 at either. It is preferable, however, to have measurements 

 made at each end to compare them, where it is possible, by 

 which the chances of errors are reduced. 



