254 report — 1859. 



and which give actual measure (sine or tangent galvanometers giving absolute 

 measure). 



1st. Let A put on a current through his galvanometer to the cable, 

 and let B connect the cabk through his galvanometer to the 

 earth. 



2nd. Note the current e entering at A =e 



and the current e' received at B = e' 



3rd. Put on a battery at B and take away that from A, and now 



note the current/ entering at B =/ 



and that received at A=/' =/' 



4th. Call the resistance of the conductor from A to the fault . . =x 



5th. From the fault to the end of the cable =y 



6th. Call the difference between e—e' =D 



7th. Call the difference between/—/' =d 



8th. Call the resistance of A's galvanometer =G 



9th. Call the resistance of B's galvanometer =g 



ailcltlieu x+G = de' 



y+g D/'' 



In both of these cases, when the resistance of the fault is considerable, it is often 

 difficult to obtain accurate results, as the fault's resistance varies considerably at 

 times, especially if the current used be positive ( + ). 



But when there are two or more wires between the stations in question, the fol- 

 lowing method removes all difficulty, and gives very accurate results. 



Case 3. — At the distant station B have the defective wire connected to a good 

 one, forming a loop from A to B and back again to A. Connect now the positive 

 pole of a battery to the earth, and the negative pole to the differential galvanometer. 

 Connect the one wire of the differential galvanometer to the good conductor, and 

 the other wire through the resistance coils (rheostat) to the defective wire. The 

 current from the battery will now split one portion of the current going through the 

 good wire to the fault, the other portion going through the resistance coils to the 

 faulty cable, and then to the fault where current escapes to the earth. Introduce 

 now so much resistance as shall make the two channels equal. Call this resistance 

 R, and then 



x+y = S, (1) 



ar+R=y+S, (2) 



whence R 



a;=S- J- 



In this way a defect in one of the wires in the Mismeer and Zandvoort cable was 

 tested, for the fault was 54^ knots from the English coast ; and when the fault was 

 cut out, the error was less than the one-third of a mile. The cable being 115 nau- 

 tical miles in length, the error was less than 0'3 per cent. 



The author mentioned a case where the conductor was 120 miles in length, and 

 the defect offered a resistance of from 1000 to 2000 miles, varying continually in 

 amount. Plans Nos. 1 and 2 were tried, as also several others, but the results were 

 very uncertain, and would not indicate the locality nearer than within 30 miles of the 

 true position. This led him to invent plan No. 3, which left a possible error of only 

 2 or 3 miles. In this case the leakage due to all gutta percha, and which is very 

 small, would have produced an error of 4 miles had it not been allowed for. Thus 

 far conductors which are continuous, but whose insulation is defective, have alone 

 been spoken of. 



When the cable or conductor is broken asunder, one of the following plans will 

 indicate approximately the amount of resistance due to the fault itself. 



Case 4. — A cable broken asunder, if possible measure the resistance from each 

 end ; and if the exposed end of the broken cable offer only a very little or no ap- 

 preciable resistance, the two amounts added together will be equal to that of one 

 perfect wire ; i. e. calling x one portion of the broken cable, and z the resistance of 



