We have now to consider the second method in which the strips are permanently strained in the direction of their length. There is only one system of this kind, viz., Crompton's system, and in this all the drawbacks. to the other methods are completely eliminated.

The strips are placed flat on the insulators, and are gripped at the two ends between heavy gun-metal 'bridges' with phosphor bronze set screws. These are ordinarily carried on cast iron straining bars fixed across the ends of the culvert which may be of any length up to 100 yards. They are insulated from the bar by means of porcelain insulators of the form shown in fig. 74, rubber pads being placed in all cases between the metal and the porcelain. The whole straining bar is shown complete in fig. 75. Where the number of conductors is large, and where the section is very heavy, the frame of the straiuing bar is made of two steel girders of railway rail section: a seven-way straining bar of this type is shown in fig. 76.

FIG. 77.-Double shed insulators.

At intervals of 15 yards, 'supporting bars' of cast iron, enamelled all over in their latest form, are fixed across the culvert about 1 inches from the bottom. They are of T-section, and from the upper surface project cast iron stalks, also enamelled, one for each conductor, on which are slipped. double shed insulators of the form shown in fig. 77, a rubber cap being placed over the top of the cast iron to act as a buffer. The supporting bar itself is shown in fig. 78. On these insulators is placed a gun-metal cap, on

ЛЛЛ

FIG. 78. Cast iron supporting bar.

which rests the conductor.

Owing to the strips being laid flat, and to their being under considerable tension, the conductors are very rigid and no jockeys are necessary.

The straining bars are placed in manholes, and a small box cover may be placed over each supporting bar, or they may be flagged over; the former is probably the better course. In either case, the culvert can be completed, and the conductors drawn in afterwards, openings being required only at the supporting bars. This is a great advantage, and the section of the

conductors can be increased at any time by drawing in additional strips, if the T-joints be taken off.

The small number of supports, and the excellent form of the insulators, renders the insulation extremely high, and this kind of distributing main is, in the Author's opinion, by far the most satisfactory that has yet been devised for continuous current distribution. It is not, of course, adapted to alternating current. The materials of which it is composed are concrete, copper, cast iron, brass, and porcelain, all practically imperishable. For very large conductors it is cheaper than other systems. Services can be jointed on at any point by simply cutting a hole in the side of the culvert for their admission, and clamping them on to the strips; the cost of making service connections is thus much less than with any other system. The security of the system is great, for, the conductors being rigidly fixed a good distance apart and the insulation being absolutely incombustible, there is far less danger of a short circuit than with any continuously insulated system; there is, however, the danger that with very heavy currents, such as those flowing into a short circuit, the electro-magnetic action may cause the strips to swing into contact, or even to leave the supports, but this is rare.

The chief drawback to the system is its bulkiness, it being impossible, in many instances, to find space for it. Another supposed drawback is the danger of flooding: it is difficult to see how the water could get in, but, even if it did, it is unlikely it would reach the copper, as the culverts have so pronounced a fall and are connected to boxes of considerable capacity at both ends which are drained to the sewers. Further than this, even if the copper were immersed, supposing the water to be pure, it is known to be quite feasible to continue running on the main until the culvert can be emptied. The accumulation of coal gas would be a source of danger; but this can be avoided by proper ventilation, which should always be provided.

CHAPTER XX.

DISTRIBUTING MAINS-INSULATION RESISTANCE AND

COST.

HAVING now examined in some detail the various systems of distributing mains, it may be well to compare the insulation afforded by them. In the first place, it is necessary to point out that the insulation of a main which has been in use for some time will be very different according as the current supplied through it is alternating or continuous. In the former case, there is no appreciable electrolytic action, but, with continuous current, there is, and the result is a most marked difference in the insulation of the positive and negative conductors. The immediate cause is electric osmosis: this is a phenomenon whereby moisture is attracted to a negatively electrified conductor and repelled from one positively electrified. Thus, if two conductors be connected by means of saturated threads with a vessel of water, and the potential of one be maintained above that of the water and that of the other below, it will be found that moisture will travel along the threads up from the water to the negative and down away from the positive.

The result of this action is that the insulators supporting the negative. conductors, or, if continuously insulated, the ends of the cables, become damp, and the insulation resistance in consequence falls. In cases where decomposition of a dielectric takes place, the same phenomenon causes a supply of moisture to be kept up and so facilitates the chemical action. In cases where only two conductors are used, and one may be earthed, as in tramways and high-tension continuous-current feeders, the positive should, of course, always be the pole insulated, and the drying action on this pole is then very valuable.

It is not easy to obtain results of tests made on mains that have been in use for some time, since it is usually impracticable to discontinue the supply through them, but the following is an instance bearing out the above statement as to the effect of the polarity: In a particular case, the insulation resistance of the positive conductor was found to be 161.3 megohms, while that of the negative was only 24.5 megohms. This test was made after the

main had been in continuous use for about five months. The difference of potential maintained between the conductors was 400 volts continuous.

In order to compare the relative insulation afforded by the various systems, it will be necessary to compare that of mains measured before they have been brought into use. In order to show this comparison, the accompanying table has been drawn up. Column 1 shows the description of cable, column 2 the method of laying, column 3 the insulation resistance per mile of the cables when tested under water at the maker's works, column 4 the insulation resistance per mile of the cable after laying in the manner shown in column 2. In each case, the conductor is half a square inch in section, except in the case of vulcanised bitumen drawn into cast iron pipes, the figures for which apply to quarter square inch cables.

The cases cited are not exactly means of tests, but are fair average samples taken from actual tests as representative. In the case of the paperinsulated cable, there were a number of straight-through joints on the cable, which accounts in large measure for the low insulation resistance after laying.

It must be observed that the state of the ends of the cables has an enormous effect upon the measured insulation resistance, on account of the surface leakage taking place over them. In the cases given, the ends were always freshly pared and dried, because it was desired to measure the insulation resistance of the dielectric, not the surface leakage; but, even so, in some states of the weather, it is most difficult to obtain the same result as in a factory, where the atmosphere is fairly dry. For this reason the insulation measured after laying is often lower than before, although the cable has not actually deteriorated. In the case cited for the diatrine cable, the figures are reversed, probably because the test was made under exceptionally favourable conditions in the street, or else because sufficient attention was not paid to the ends at the maker's works.

If the ends are left for a few hours, or even minutes, exposed to the atmosphere, the insulation will drop to one-hundredth part of its former value or less. Hence it is seen that the apparently high insulation resistances attained are quite fictitious, since the conditions necessary to yield them cannot be fulfilled in practice. Therefore, although the insulation resistance of bare copper systems is apparently much lower than that of cable systems, it is measured under working conditions, and the difference is thus much less than it seems to be. Experience shows that the insulation of bare copper, properly laid, is far more permanent than that of most cable systems, and it is far better to maintain a moderately high standard than to have an excessively high one, liable to break down with lapse of time.

The cost of any system of mains will necessarily vary according to local conditions, and for each system it will consist of a considerable number of different items. In examining the cost, we will assume that the main to be laid comprises three conductors of square inch each.

In what follows the figures given are approximate only, but will serve as a fairly close guide. The basis is given, as far as practicable, so that altered conditions can be allowed for.

The chief items of cost in every system of mains are: (1) Excavation of trench, including carting and tipping of spoil, and removing obstacles, such as pipes. (2) Providing and fixing conduit. (3) Cable, including conductor and insulation and cost of laying in position. (4) Filling in ground. (5) Making good the surface of the roadway. (6) Junction boxes. No allowance is here made for service boxes. (7) Sundries. These comprise watching at night during progress of work, lighting of cutting at night, and numerous small items.

It is not possible to consider each item separately, so three representative examples will be given of the cost of the complete main.

(1) Vulcanised rubber cable laid in 6-inch earthenware pipe. The pipe is assumed to be laid at a depth of 1 foot 6 inches to the top of the pipe. This will be below most of the gas services and above most of the water services. The cost of the best quality tested earthenware pipe, fitted with Doulton patent joints, is about 1s. 11d. per yard, delivered. The cost of excavating will be about 8d. per yard run, and the cost of tipping about 6d. per load. If the tip is situated at a moderate distance, the carts can make, say, five journeys per day each. The cost of horse, cart, and driver, if hired, is about 9s. per day. The cable, if of 2500 megohm vulcanised rubber, is about £700 per mile delivered, or, for the three conductors, £2100 per mile run. The cost of reinstating is, for York flags, 1s. per square yard; for granite sets, 2s. 6d. per square yard. Each junction box, with 3-wire Pillar Distributor fixed complete, is about £14. Based on actual experience, the cost of all the items, including sundries, works out to about £2700 per mile.