satisfactory results, owing to the capacity of each conductor as regards the other two not being the same. When the current is used to drive rotatory converters, this is of serious moment, and, to avoid this defect, three-core cables are used. These are similar to those described for distributing mains, consisting of three conductors, separately insulated, and laid up together to form a cable, which is then armoured with steel on the outside. To secure safety, the middle point of the three-phase winding is earthed, as also is the armouring. A section of a three-core or 'clover leaf' paper insulated cable of this kind for transmitting three-phase current at a pressure of 6500 volts is shown in fig. 67.

The most popular method of laying high pressure mains is to armour them and lay them directly in the earth or on the solid system, in the latter case sometimes omitting the armour; the solid system is greatly to be preferred. In the case of extra high pressure three-phase cables, it is well to take the additional precaution of laying the armoured three-core cable in a cast iron trough connected to earth. In order to secure efficient connection between the armour and the casing, the Author uses one gunmetal or malleable iron bridge in each length of casing, secured to it by means of set screws and clamped to the armouring.

High pressure cables should never be drawn into the same pipes as low pressure ones, and should be laid as far as possible from them, and on no account should they be allowed to pass through the same junction box; if this be neglected, there is danger that a man may mistake one cable for another.

Means should be provided for clearly distinguishing high pressure mains from one another, where several run together, at every joint box.

The necessity for varying the pressure on high pressure feeders of different lengths is not so keenly felt as in the case of low pressure feeders, but in certain cases it has to be resorted to. The same devices as were described for low pressure feeders may be employed, and the same objections arise. Of course, the battery method is not applicable to alternating current; when this class of current has to be dealt with, choking coils may be used instead of resistances, and this method is then not so wasteful, though it is still open to objection.

The question of the use of cut-outs on a distributing network has already been discussed; the matter is somewhat different as regards feeders.

In the case of low pressure feeders, it is obvious that if the distributing network has no fuses, and a fault has to be burnt out, the feeders must of necessity carry sufficient current for the purpose, and this practically precludes the use of fuses on them. Moreover, if the insulation of a feeder failed, a fuse at the station end only would not be of any use, since the network would send current back into the fault. The only way to obviate

this is to have a cut-out which will open the circuit when current passes through it in the opposite direction to the normal. Such a cut-out is easy to arrange for continuous currents, but for alternating it is more difficult. A successful device for the purpose, invented by Mr L. Andrews, is described below.

As a matter of fact, it is found that low pressure feeders very rarely give trouble, the insulation of conductors of large size being much less likely to fail than that of small, probably because, when continuously insulated, the covering is considerably thicker, and, in the case of bare copper systems, the weight and rigidity prevent the conductors shifting. The usual practice, therefore, is to run low pressure feeders directly from the station to the network without cut-outs of any kind.

Probably the best practice would be to have a return current cut-out at the end of the feeder nearest the network, and none at the station end. In the event of a short circuit on a feeder, it would be cut off at once from the network, and the ammeter at the station would show which was the faulty main, and it could then be switched out by hand.

High pressure feeders either have a number of isolated transformers connected to them, or they are connected with a sub-station. In the former case, each transformer has its own fuse, so as to cut it out, if it or the circuit supplied by it go wrong, while the feeder itself is controlled at the station by a fuse. When this kind of feeding is adopted, the distributors rarely form one large network, and, therefore, the difficulties described above in relation to low pressure feeders are on a smaller scale, being non-existent in the case of a number of isolated distributors, each fed by its own transformer, since, in this case, current cannot flow back from the distributors into the mains.

By far the most usual course, however, is to have sub-stations, and very frequently each has a separate feeder; while, in the case of a large sub-station, there may be several mains connecting it to the generating station.

In such cases a number of transformers are connected in parallel to one omnibus bar into which all the feeders deliver current, being therefore also in parallel. The ordinary practice is to place fuses at both the generating and sub-station ends of the feeders and on each transformer.

Now, it is obvious that if there be, say, two mains, and one of them becomes short-circuited, current will flow into the faulty one, and blow its fuses at the station end; but in order to blow the one at the sub-station end, the current necessary to do this must pass through the good main, and since the fuses are necessarily all of the same size, it being impossible to foresee which main will fail, the fuses at both ends of the good main will blow. It is assumed that the network, fed by the large sub-station, is an isolated one; if several sub-stations feed into one large network, the effect will not be so marked, but will be of the same order.

The same remedy as in the case of the low pressure feeders must be applied, viz., a cut-out which will open the circuit when the current reverses its direction, and not when it exceeds the normal.

If, instead of the main itself failing, one of a bank of transformers burns out, the fuse connecting it to the omnibus bar will go, and cut off the high pressure connection. The result will be two-fold. In the first place, this transformer ceasing to contribute its quota of the supply, an increased load will be thrown upon the remainder. If their fuses be light, they will blow; to meet this contingency they should be made very heavy and the transformers be constructed to stand a good overload. The second result will be that the current will be reversed in the low pressure winding, and the transformer will short-circuit the low pressure side instead of the high, the result being the same, namely, the blowing of the fuses on both transformers and mains.

Once more, the remedy is to connect the low pressure side of each transformer to the network by a return current cut-out instead of a fuse.

It is unnecessary to describe the devices for continuous current work, but the one invented by Mr Andrews for alternating currents, already referred to, is interesting for its ingenuity.

The cut-out is actuated by means of a pivoted armature, wound with a fine-wire winding, connected directly as a shunt to the low pressure main to be protected, or through a transformer, if it be a high pressure one. The armature moves in a magnetic field set up by the main current flowing through the feeder.

Now, it is obvious that the current in the shunt winding flows in the same direction, whichever way that through the series winding goes, and that this is true for the instantaneous value of an alternating current ; hence, the armature will move in one or other direction, according to the direction of the current in the main coil.

The armature operates a catch which releases a weight that, on falling, breaks the circuit of the main. The catch is of peculiar construction, and is not affected by vibration, while the current of normal direction tends to lock it in its place. Once the armature begins to move in the opposite direction, however, the catch leaves go immediately, there being practically no friction on it.

The device is arranged not to let go with zero current, a certain amount of current in the opposite direction to the normal being necessary to release it.

CHAPTER XXIII.

TESTING OF MAINS.

Ir is vitally important that mains should be rigorously tested both before. they leave the manufacturer's works and after laying.

It would be out of place here to give detailed specifications for mains. Suffice it to say that every specification should provide for the use of copper of a certain conductivity, which should be measured; for a certain minimum thickness of insulation, or 'wall' as it is called; and for a definite insulation resistance and dielectric strength. The cable, whether it have a waterproof dielectric or be lead covered, should always be tested under water before the outer braidings are put on, and it should be thoroughly dried after immersion; the tests should not be made until it has soaked for at least twentyfour hours. The first test applied should be a high pressure test of, say, 23 to 7 or 8 times its normal working pressure, applied for, say, half an hour. After this, the insulation resistance should be measured after one minute's electrification from a battery giving a pressure of, say, 500 volts.

The insulation resistance to be specified varies greatly with the dielectric. used, and must be altered according to the material; it will depend also upon the size of the cable, the larger the conductor the lower the insulation resistance per mile for the same material.

To take a particular case as a guide to the comparative insulation resistance of different materials, a half square inch cable, if insulated with first-class rubber, should have an insulation resistance of 2500 megohms per mile, if with vulcanised bitumen 250 megohms per mile, if with paper 100 megohms per mile.

In order to further test the quality of the dielectric, a breakdown test should be made; this consists in taking a short piece of the cable and bending it backwards and forwards several times round a drum of small diameter, and then applying a pressure sufficient to pierce the dielectric. This should not be less than four times the pressure applied to the whole cable in the first instance. Every length of cable tested should be marked in such a manner that it can be subsequently identified. With lead-covered cables the lead can be stamped; with other cables, one strand may be hammered out flat and stamped with a very small die.

After laying, each length of cable should be carefully tested for insulation resistance after one minute's electrification with a battery giving not less than 300 volts.

By far the most suitable apparatus for this purpose is the portable testing set supplied by Messrs Siemens. It comprises a battery of 250 Hellesen or Obach cells; a reflecting galvanometer with a flat mirror in which is reflected a fixed scale, the image of which is received by a telescope provided with a prism to enable the readings to be taken from above; an ordinary shuntbox and a Wheatstone bridge containing 10,000 ohms; also a condenser having a capacity of 0.5 microfarad.

The whole apparatus is most conveniently arranged in a light covered truck, which protects it and the operator from the weather. The galvanometer is fixed on a table carried by a tripod, which can be let down so as to rest on the ground independently of the truck. With this apparatus a sensibility of about 100,000 megohms per division of the galvanometer scale is easily attained.

With ordinary care, the apparatus never gets out of order. The Author has had such a set in daily use for over five years, the repairs have been practically nil, and the original cells appear to be as good as ever.

In making insulation tests, whether in the street or at the manufacturer's works, the utmost care must be taken to have the ends of the cable absolutely clean and freshly pared.

After the various lengths are joined up, a high pressure test should be applied; this need not exceed twice the working pressure, but should not be less. The number of lengths that can be tested at once will, of course, depend on the size of the testing plaut.

For this purpose a portable generator must be used. The Author employs an arrangement designed and made to his specification by Messrs. Crompton & Co. It consists of a continuous current motor, taking current at 400 volts, and driving an alternator giving a pressure of about 500 volts. By means of resistances in the fields of both machines, the alternating pressure can be varied at constant periodicity, and both pressure and periodicity by speeding up the motor.

The alternator can be directly connected to the testing lead if only low pressures are required, but normally it is connected to a set of three transformers having their low pressure primary windings in parallel and their secondaries in series. Connections are brought out at five points of each transformer, so that a regular succession of steps of 2000 volts can be obtained from 2000 to 30,000 volts. The intermediate values between two steps are got by varying the fields of the machines.

The transformers are contained in a wrought iron box, with micanite-lined cover. The switches are fixed on top of their respective transformers within the box. The primary current passes through contacts on the lid