THE LOCALISATION OF BUJiAKS ANJJ, FAULTS. 469 



the reslstauce, and we may then take it that the resistance so 

 added to the battery circuit is such that the current is the 

 same through the fault in both tests of a pair. The value of x, 

 so obtained can then be taken as final. 



If the fault varies much, the resistance to add may work out 

 too high, and the result of having too much resistance is to 

 make x too little or to "bring the fault home." And the more 

 the fault is brought home the higher the resistance to add 

 works out, thus making matters worse, which may result in 

 putting the fault on the testing table unless it undergoes 

 some other change. When a fault is behaving like this there 

 is nothing to guide one but experience. 



Dr. Kennelly also pointed out that, alternatively, the current 

 through the fault could be kept constant by varying the E.M.F. 

 of the testing battery as given in outline in the work on 

 "Testing" by the late Mr. Schwendler, and he suggested 

 increasing the battery power when taking the " earth." A pre- 

 liminary pair is first taken in the ordinary way with a certain 

 number of cells, giving readings of F and E ohms respectively, 

 from which a first approximation of x is obtained in the ordinary 

 way. Keeping this battery for the free reading, the number 

 of cells to be used when taking the " earth " is found as follows. 

 Multiply the number used when taking the " free " by the 

 following quantity : — 



(L-a;) (F + a.)' 



Where a is the resistance of one of the fixed arms of the bridge, 

 an even bridge being used. This assumes the battery resistance 

 negligible in comparison to the other resistances of the circuit. 

 If this comes out a fractional number, the nearest whole number 

 is of course taken as the required number of cells. 



Another pair is then taken with the new battery power for 

 the "earth" and the original battety power for the "free," 

 and the results worked out by the formula. This gives a 

 nearer approximation of x and with the new values of F and x 

 the number of cells for the earth reading is again worked out. 

 If this differs from the former result a fresh pair is taken with 

 the battery power corrected. After a few pairs it will be found 

 that the number of cells needs no further alteration, and the 

 readings may then be taken as final. 



