58 THE SCIENTIFIC PAPERS OF 



where X and t represent specific conductivities and times occupied 

 in both experiments. 



This result is independent of any eccentricity of the wire in its 

 insulating covering. The method is therefore well adapted for 

 determining the specific resistance of materials, but as it is neces- 

 sary to ascertain whether the wire is throughout the cable, con- 

 centric with the insulator, this method cannot be exclusively used. 



Besides this process requires considerable time in testing well 

 insulated cables. Again, another objection to its exclusive use 

 arises from the possibility of slight faults in long cables passing 

 unappreciated, as the loss of tension through such faults will be 

 exceedingly small as compared with the whole charge. 



We therefore prefer to determine the loss of tension not by an 

 electrometer, but by measuring the charge a, and after the lapse of 

 one minute the discharge b by the galvanometer needle. We then 

 have the loss of quantity or tension during the minute 



J.-l-l . . . (VI.) 

 In order to associate this formula with the system previously 



t) 7/ 



developed, it is only necessary to remark that is equal to ^ 



tt jii. 



The cable having been tested from the earliest stage of its 

 manufacture (in lengths of one knot) subsequently during the 

 joining and covering of the cable, and finally during the paying 

 out, these tests must strictly control each other, and must con- 

 sequently be recorded systematically. The chief care during the 

 submersion of the cable should be to detect at once the slightest 

 change in its insulation, in order that the paying-out machinery 

 may be stopped instantly. It sometimes happens, however, that 

 a fault does not appear immediately on submersion. It is there- 

 fore necessary, if a fault appears, to calculate its exact place before 

 taking any other steps to remove it. In order to do this effectually, 

 it is necessary to test the cable from both ends, i.e., from the ship 

 and from the land station, as the determination from one side gives 

 only the maximum distance. 



In paying out submarine cables, we pursue the following plan 

 of testings : 



