wire between the subscriber's station and the central office.

Another objection to overhead wires is the annoyance to the tenants of buildings caused by the line-men tramping up and down the stairs, and over the roofs, to construct or repair the wires.

If the wires, instead of being carried over house-tops, are carried through the streets on poles, not only are the poles obstructions to travel, but they and the wires strung upon them are exceedingly unsightly, and it has been repeatedly proved that they are great hindrances to the operations of firemen in case a building, in front of which these wires are strung, is on fire.

But perhaps the greatest objection on the part of the parties operating them is the fact that every two or three years there comes a heavy sleet storm, followed by wind, and the wires not only become entangled, but break, and come tumbling down so as entirely to interrupt communication, and to cause enormous expense for reconstruction.

The annual cost of repairs of overhead wires in cities is not less than thirty per cent of the first cost of construction. These various objections to overhead wires have caused the companies to look about to see what could be done in the way of gathering the scattered wires into cables, and, if possible, to lay such cables under ground.

It is for the interest of the companies, quite as much as for the public, to have this problem solved, and the American Bell Telephone Company has spent large sums in an endeavor to find a practical method of placing the wires of an exchange system under ground.

The difficulties met with have been great, and they have not yet been sufficiently conquered for me to say that it is both technically and economically possible to put all of the wires of any exchange system under ground; but the problem has been so far solved that it may be said to be practicable to run from the central office under ground to a considerable number of points, so located that one or another of them may be reached by a short overhead wire from any subscriber's 'station.

With regard to the burial of telegraph wires: this has been done for years in all the European cities, and can, of course, be done in the same way here.

There is a great difference, technically, between the operation of telegraph and telephone wires. If two or more telegraph wires are

bunched together into a cable of such length as would be used in our largest cities, each wire continues to work practically as well as before, and each wire works entirely independently of the neighboring wires in the same cable. This is true even of the old-fashioned cables in use ten years ago.

When, however, it was first attempted to bunch telephone wires into cables, serious technical difficulties were met with. In the first place, it was found that conversation was very much lowered in intensity when, instead of speaking over an overhead wire say five miles in length it was attempted to talk over a cable-conductor insulated with gutta-percha of the same length. But, worse than this, it was found that conversation carried on over one wire was heard with equal facility on all of the other wires in the same cable.

This decrease of intensity is due to what is known as retardation, by means of which each signal, instead of being sharp and distinct, is partly kept back, so that it overlaps, and mingles with the next. In the case of a telegraph instrument, the signals do not succeed each other with sufficient rapidity for the retardation to be noticeable on lines of such length as would be used in any of our cities.

In the case of the telephone, the electrical undulations in the wire, by means of which speech is transmitted, necessarily succeed each other some three hundred times per second, and in a gutta-percha cable, five miles in length, a considerable retardation and consequent overlapping of the signals, resulting in a diminution of the intensity of conversation, are felt.

The overhearing, or cross-talk, may be due either to a direct leakage between the conductors, or to what is technically known as induction, by means of which signals sent on one wire cause fac-simile signals in all the other wires, even though there is no direct passage of electricity from one wire to the others. Here, too, in the case of telegraph apparatus, the induction is not sufficient to affect even the most delicate apparatus in use. In the case of the telephone, however, the induction is amply sufficient for overhearing in a guttapercha cable five miles in length.

The retardation in any cable is directly dependent on the specific inductive capacity of the material used to insulate each wire from its neighbors; and it is evident that a cable which will present the least

retardation is the one whose insulating material has the lowest specific inductive capacity.

The cross-talk- so far as it is due to leakage - is, of course, prevented by using an insulating material of very high insulating power. So far as it is due to induction, we also want to choose an insulating material of low specific inductive capacity, for the crosstalk is directly dependent upon this quality.

The requisites of a cable, then, in order that it may transmit speech without cross-talk, are good conductivity, high insulation, low specific inductive capacity.

Below is a table showing the specific inductive capacity and insulation of various insulators. The measurements were all made on a wire 0.05 of an inch in diameter, coated with insulation to a thickness of 0.10 of an inch:

Let us take a special case, and compare a gutta-percha cable, having a specific inductive capacity of 4.2 with a Faraday cable of 1.6.

The table predicts that we can talk three times as far with the latter as with the former, and experiment proves it. The cross-talk on the gutta-percha cables ought to greatly exceed that on a Faraday cable; and experiment has shown that, while conversation over a two-mile gutta-percha cable was continually disturbed by existing cross-talk, conversation was carried on over a similarly-constructed Faraday cable, five miles in length, without cross-talk being appreciable.

We have seen, from our table, that india-rubber and kerite, both of which are extensively used for telegraph cables, are equally unfit

with gutta-percha for telephonic work, on account of their specific inductive capacity being nearly as high as that of gutta-percha.

There are two other cables mentioned in the table. The Patterson cable, which has a specific inductive capacity of 3.1 against 4.2 for gutta-percha, and the insulation of which, when new, is 450 megohms per mile, against 190 on the part of gutta-percha. The table predicts that a Patterson cable-which consists of cotton-covered wires soaked in paraffine, and drawn into a lead pipe-ought really to be more suitable for telephonic work than is a gutta-percha cable, and experience bears out this prediction.

There is one fatal objection to the Patterson cable, which has been proved by experience with it in France, Germany, and other places the insulation, although high at first, gradually decreases, and experience has uniformly shown that, after several years of use, this insulation falls so low that cross-talk easily appears, due to direct leakage.

Patterson cables, of course, under different names, were used years ago in France and Germany, and were looked upon with a great deal of favor when first introduced, but the gradual failure of insulation has caused them to be almost entirely abandoned abroad.

Another cable referred to is the Brooks, which consists of coppercovered wires wound with cotton, and drawn into iron pipes, which are then filled with petroleum.

The specific inductive capacity of a Brooks cable is only 2.8 against 4.2 on the part of gutta-percha, and on this account it is suited to telephonic purposes. The Brooks cable, however, like the Patterson, does not retain its insulation, and, indeed, the difficulty of maintaining good insulation in the Brooks cable is far greater than in the Patterson, for it is almost impossible to make the pipes so tight that the petroleum does not leak out, or water leak in.

In the table, I have assigned no value to the insulating power of the Brooks cable, for when the pipes are thoroughly dried, and the cables, after being thoroughly dried, are drawn in, and the pipes are filled with dry oil, the insulation is enormous, and such a cable gives wonderfully good results when used for telephonic purposes, talking excellently well, and being remarkably free from retardation and cross-talk due either to induction or leakage.

The insulation, however, falls continually, and at the end of

three or six months, if it be of any considerable length, will have fallen so low that talking on one conductor can easily be heard on all of the other conductors, because of direct leakage. Because of this difficulty of maintaining insulation, the Brooks cable, after being tried in England, France, Germany, and elsewhere, as well as in the United States, has been generally abandoned.

The makers of the Faraday cable claim that the enormously high insulation, which certainly does exist in it when it is new, can always be maintained, and, of course, there is no reason why its specific inductive capacity, which is the most essential feature, should ever change at all. German counterparts of the Faraday cable have been in use six or seven years, and experience has shown that they have not changed materially in insulating power.

We have seen that, if we were to attempt to construct an underground telephone system, say five miles in length, using gutta-perchacovered wires in cables, we should find difficulty in talking over any one of these conductors, and we should find that conversation on one wire was heard with almost equal facility on the other wires.

If, however, we use, instead of gutta-percha, a Faraday cable, we find in practice that, because of the high insulation and the extremely low specific inductive capacity, we are able to talk over a system of cables five miles in length with perfect facility, and without serious interference from cross-talk between the neighboring wires.

While at Mülheim, in Germany, last summer, I talked over such a system. The makers had constructed a cable five miles in length, and at each end had run the conductors out to a number of stations by means of short overhead wires. The cable was laid back and forth in an open field. Electrically it was exactly the same as if it had been placed under ground. In other words, we had a complete underground telephone exchange-system, with wires running side by side at as great a distance as five miles, and this is as extensive a plant as we shall ever have occasion to use. The statement that the wires in the Paris telephone exchange, which counts over three thousand subscribers, are placed under ground, is true, and it is further true that gutta-percha cables are used. These cables are constructed in a peculiar way. Each circuit, instead of consisting of one wire extending from the central office to the subscriber's station, consists of two insulated wires twisted spirally, the current going over one wire and