346 On the Use of Price s Guard- Wire in Insulation Tests. 



appear less than its true value. If, however, the two tests 

 give about the same result (care having been taken to pro- 

 periv discharge the cable between them) we have proof that 

 the effect of surface leakage is inappreciable. It is frequently 

 more expeditious to make the two tests above mentioned than 

 to clean and coat the end? with paraffin-wax in such a 

 way as to make one feel confident that no surface leakage 

 exists. 



In making the guard-wire measurements it is desirable to 

 put the guard as near the braiding or sheathing as practi- 

 cable, consistent with keeping the resistance between the 

 guard- wire and the sheathing high compared with that of 

 the batterv. Bv so doing any error due to the guard- wire 

 is lessened, for the path over which surface leakage can 

 occur is made as long as possible and nearly equal to that 

 over which leakage takes place when no guard-wire is used. 

 An approximate correction is then possible, even when the 

 two tests with and without the guard-wire do not give the 

 same result. 



Suppose, for example, the swing of the galvanometer- 

 needle on the first charge is s, divisions, and s 2 , s 3 were the 

 swings on " recharge" after the same period of " insulation w 

 with and without guard-wire respectively, then the value of the 

 recharge swing, if no surface leakage existed, would be given 

 by the expression 



approximately. 



!— $2 



2*i 



The preceding formula is generally sufficiently accurate 

 for practical purposes, but it is possible to obtain absolute 

 accuracy in the use of the guard-wire with the (i loss of 

 charge " insulation test in the following way. Instead of 

 maintaining the potential of the guard-wire constant, while 

 that of the insulated conductor is steadily falling, cause the 

 potential of the guard-wire to fall at the same rate as that of 

 the insulated conductor. Then no electricity can pass between 

 the conductor and the guard-wire, and surface leakage will 

 be absolutely prevented. 



A potentiometer and an electrostatic voltmeter can be con- 

 veniently employed in carrying out this experiment, and it is 

 interesting to remember that the Thomson and Varley poten- 

 tiometer was originally devised for the special object of 

 enabling the potential of one conductor to be made to 

 diminish as the potential of another conductor steadily fell 



