the Use of Price's Guard- Wire in Insulation Tests. 345 



the two tanks need not be removed. An insulation test applied 

 under the above conditions gives the resistance of the part of 

 the cable in the earthed tank, and a comparison of this with 

 the insulation resistance of the whole will show whether one 

 half is worse than the other half. By repeating this test with 

 say one quarter of the cable in the earthed tank, and the 

 remainder in the other tank, the question of local v. general 

 leakage may be decided. 



Instead of two tanks, two drums, one of which can be 

 fairly well insulated, might be employ ed, or one drum and 

 one tank used, for we find that after the braiding of a cable has 

 been thoroughly wetted it acts as a fair conductor for several 

 hours, even on a dry day. It is therefore not essential to 

 have the part tested completely immersed, although of course 

 immersion is conducive to constancy of temperature. 

 Sufficient connexion with the wet braiding can be made 

 by wrapping bare wire round it. If the cable be long this 

 may be done at several places along its length. 



By employing the two drums above mentioned faults in 

 braided and other unsheathed cables can be readily localized, 

 without in any way injuring the cable. After connecting 

 the guard-wire with the insulated drum the wetted cable is 

 wound from the earthed drum to the insulated one, the galva- 

 nometer being connected with the inner conductor during 

 the process. When the fault passes to the insulated drum 

 a sudden diminution of deflexion occurs. 



In some cases we have only used one tank (the earthed 

 one in fig. 2) in testing part of a coil of cable, the remainder 

 being first wetted and hung up on a wooden beam supported 

 between two tables. By inserting metal sheets connected 

 with the guard-wire between the beam and the tables errors 

 which might otherwise arise from surface leakage are elimi- 

 nated. 



The guard-wire properly applied gives complete protection 

 against surface-leakage errors when the tests are made by 

 the " direct deflexion " method, but in the " loss of charge " 

 test it is desirable to make measurements both with and 

 without a guard-wire. Unless this be done one cannot be 

 sure that surface leakage is not influencing the results 

 obtained. 



A little consideration will show that if any surface leakage 

 exists a guard-wire will tend to keep up the potential of the 

 inner conductor during the period of u insulation/ 1 and there- 

 fore cause the insulation resistance to appear greater than its 

 true value ; whereas if no guard- wire be used surface leakage 

 will facilitate discharge and cause the insulation resistance to 

 Phil Mag. S. 5. Vol. 49. No. 299. April 1900. 2 B 



