along the exterior of the cable in the expected vicinity of the fault, when the warmth of the cable at that point will afford sufficient evidence of its immediate locality.

A total or partial disconnection in an insulated cable or wire, where the broken extremities are maintained in imperfect contact by the surrounding insulation, may be localised in like manner by connecting the two extremities of the faulty section through a suitable resistance or cut-out to a generator, when the arc set up at the fault, or heating due to the high resistance offered to the passage of the current at that point, will give sensible evidence of the position of the disconnection.

In all cases cited above, if the resistance of the fault be sufficiently great to oppose the passage of an ordinary low tension current, a higher voltage, varying from one to five thousand volts, according to circumstances, will frequently overcome this obstacle, and similarly assist in the localisation.

I wish it to be distinctly understood that in mentioning the above instances for emergency fault localisation in electric circuits, I do not recommend their general adoption, but have rather included them in this series to illustrate what can be done, and what to a limited extent is done, especially in American and Continental practice, in order to get over the difficulty caused by a temporary breakdown in mains and circuits. In all the cases referred to in the foregoing paragraphs thoroughly reliable safety precautions should be adopted; the tests should only be applied to circuits which are available for immediate inspection; and, above all, they should be carried out under the immediate supervision of experi enced men.

The fault or faults having been duly located and repaired, the next proceeding is obviously to test them (the repairs or "joints," as they are commonly termed) in order to ascertain if they have been satisfactorily carried out, and this we will now proceed to discuss under the heading of Joint Testing.

Clark's Accumulation Method is illustrated in Fig. 85, the apparatus required for its conduct being a well-insulated trough T, the insulation of which may be effected by suspending it from a convenient support through the medium of ebonite rods; a sensitive high resistance gal

vanometer G, a testing battery E, capable of yielding from 200 to 300 volts at least, a condenser C, and a Webb's discharge key K. The perfect insulation of the trough must be first ascertained, and, to this end, the pole of the battery which, in the figure, is connected to the extremity of the core containing the joint, is connected to the plate P which is immersed in the water in the trough, and the key K is depressed, thus charging the condenser C. The battery is then disconnected, and,

after an interval of, say, one minute, K is released, and the discharge deflection noted; it should be practically equivalent to the deflection obtained when C is instantaneously discharged at the completion of the charging period.

The insulation of T being thus assured, the joint is immersed, care being taken to clean and dry the insulation on either side for the space of a few inches to prevent surface leakage, and connected with the remainder of the apparatus as indicated in Fig. 85. All being in readiness, the short circuit plug of the condenser C is inserted, and the key K depressed; still keeping K down, the short circuit plug is withdrawn, and the condenser immediately receives a charge from E through the joint. At the end of a stated period of, say, one minute, the key K is released and the discharge deflection noted. This should amount only to one or two degrees, and is compared with that obtained when a perfect piece of core is substituted for that containing the joint. The deflection obtained from the former should not be greater

than that resulting from the latter; if it be so, the joint is defective.

The Discharge Method is primarily the opposite of the foregoing, and consists in fully charging the condenser, which is then allowed to discharge itself for a stated period through the joint, its subsequent discharge deflection at the end of that period being duly ascertained and compared with the instantaneous discharge; if the joint be good, there should be very little difference between the two deflections.

Raymond-Barker's Accumulation Null Method of Joint Testing. Referring to the two joint tests described a few paragraphs back, it will be seen that they consist in comparing the leakage of current through the dielectric of the joint, either into or out of a condenser, and, as this leakage in the case of a perfect joint must of neces sity be extremely small, it follows that any extraneous leakage, however slight, such as that caused by the imperfect insulation of the trough T, Fig. 85, will exercise an appreciable effect upon the test, especially under varying atmospheric conditions, i.e., when the reading from the joint proper may be taken under different conditions to that from the comparative sample of perfect

core.

It follows, therefore, that a test which allows of simultaneous observations on both joint and core, is much less liable to error than one in which the observations are made separately, and the following method provides a means of attaining the required result. It is represented in Fig. 86, where T is a well-insulated ebonite trough divided into two portions, also well insulated from one another by a central partition. p, pl are contact plates immersed in water in the troughs which contain the joint under test, and the comparative length of perfect core respectively and C1 are two equal condensers, E a battery of high E.M.F., and G a reflecting galvanometer. Ki, K2, and K3, shown respectively as two Webb's discharge keys, and a simple circuit key, are combined in Price's mixing key, whilst K4 is a second simple circuit key. The modus operandi is as follows:-Kl and K2 are depressed simultaneously (one movement of the combination key effects this), thus charging C and C1 with electricities of opposite sign through the joint and core

respectively. K1 and K2 are then released, and allow the charges to mix with a tendency to neutralisation. K3 and K4 are then closed, and any preponderance of one charge over the other produces a deflection on G, which will be to one side of the scale zero or the other, according to which possesses the higher insulation, the joint or the core. The direction of indication may readily be determined by experiment with two core samples of known inequality of insulation.

If, in the aforementioned joint tests, the core or joint form part of a circuit of appreciable magnitude, the short circuit plugs of C and C1 must, in the first instance, be inserted, and withdrawn after depressing K1 and K2.

Having thus far dealt with the ordinary routine of testing, we will now proceed to discuss certain matters which, although beyond the pale of testing pure and simple, are nevertheless connected with it, and in many cases have an important bearing thereon. The major portion of the following matter has been culled from recent publications, periodical and otherwise, and, this being the case, it will tend by its presence to bring the foregoing information up to date. We will deal with them severally under the convenient heading of

MISCELLANEOUS.

Price's Guard Wire.-The merits of this device were

recently discussed in a paper read before the Institution of Electrical Engineers, and it merits special mention in that it is a successful application for the elimination

of that bugbear surface leakage, in insulation testing, etc. The principle, as applied to the extremities of a submerged cable under test for insulation, is represented in Fig. 87, where A and B represent the two extremities of the cable, duly prepared and tapered in the manner described under the heading of Insulation Resistance Measurement, G the galvanometer, and E the testing battery. At the points a and b a fine copper wire is wound tightly for some three or four turns around the tapered extremity of the insulation, over which the surface leakage tends to take place, and connected, as shown, between the galvanometer and battery. However great the leakage tendency before this precaution was taken, it will be found to have entirely disappeared when the connections are arranged as in the figure, and only the true deflection due to the cable will be obtained.

The same principle may be applied to prevent surface leakage on the galvanometer by connecting the three levelling screws or case of the latter to the insulated battery terminal, care being taken to ensure the thorough insulation in all other respects of the galvanometer itself, as otherwise a short circuit will exist across the battery terminals.