SUBMARINE TELEGRAPHY 203 



tical improvement of the material. Resistance coils would now 

 be considered veiy bad if their normal values were inaccurate 

 to the extent of one part in a thousand ; they may be procured 

 ranging from one unit to 100,000. The standards issued by 

 the Committee above named profess to be identical in their 

 resistance, without a greater error than one part in ten thousand. 

 Still greater accuracy could be obtained if required, but the 

 precautions necessary are then very numerous, as may be seen 

 on consulting the various papers by various members of the 

 Committee on Electrical Standards, published in the British 

 Association Eeports from 1862 to 1865. 



A very wide gulf separates the present practice from the old 

 plan of simply ascertaining the continuity of the conductor. 

 Every hank of copper wire is tested for resistance even before 

 it is spun into a strand. The resistance of the strand is measured 

 by the engineers when covered with gutta-percha, and before 

 being admitted to form part of the cable * for twenty-four hours 

 previous to this test it is kept at a stated temperature. The 

 conductor of the manufactured cable is also daily measured, less 

 for the purpose of ascertaining its electrical properties than to 

 ascertain its temperature from its observed electrical resistance, 

 and also to check the length supposed to be in circuit when 

 other tests are made. These tests are interfered with by variations 

 of temperature, by slightly imperfect connections, by the in- 

 duction of the wire upon itself, and, after the cable is laid, by 

 earth-currents. But the precautions thus rendered necessary 

 are well understood, and carefully observed in the case of all 

 important lines. The quality of the copper enters into the engi- 

 neer's specification with precisely the same numerical accuracy 

 as its weight : it is referred to definite units ; and no more fre- 

 quent disputes arise between the contractor and engineer as to 

 these measurements than as to the weights of material supplied. 



A further use of these measurements will be spoken of when 

 treating of repairs ; but for the present let us leave the tests of 

 the conductor to consider those of the insulator. The conductor 

 may have more or less resistance, and work worse or better in 

 consequence, but if the insulation be defective, the cable may 

 not work at all, and the tests of insulation are therefore the 

 most important of all. The old rough test was defective in 



