least twice their diameter to avoid stripping the thread; in short, considerable strength is required to hold the conductors firmly in place. Some device should be provided to hold the switch open and prevent accidental closing. This can usually be accomplished by simply placing the switch so that the arm hangs downward by gravity when it is open. The stupid mistake is often made of reversing this arrangement, so that the handle tends to fall and close the switch; which danger should be avoided, even if the switch is provided with a catch to hold it open. The handle of any switch, even for low-tension currents, must always be perfectly insulated. Special forms of switches for arc, alternating, and other circuits will be described under Electrical Distribution in Volume II.

Safety Fuses. Almost all electrical circuits, except those for constant-current arc-lighting, are protected from abnormal increase of current by safety fuses. These consist of wires or strips of metal introduced into the circuit, and so designed in cross-section and resistance that they will melt and open the circuit in case of excessive current, before the rest of the circuit becomes unduly heated. These devices are extremely simple, and would seem to be very satisfactory for the purpose; nevertheless, much uncertainty exists in regard to the theory and practice of safety fuses. The requirements for an effective safety fuse may be stated as follows:

1. They should melt with a definite current.

2. They should not change in this respect by the effect of time, heating, or other action of the current, or in fact under any reasonable conditions.

3. They should act promptly.

4. They should give a firm and lasting contact with the terminals to which they are attached.

The ordinary practice is to manufacture spools or coils of wire, composed of some easily fusible alloy; or, for larger currents, flat strips of fusible metal provided with copper terminals are used. The idea is that each size of fuse will carry a certain normal current, but will melt and open the circuit with a certain excess of current, which is usually put at 25 per cent or 33 per cent increase. But, as a matter of fact, the fusing point is consider

ably modified by many conditions, such as temperature of the surrounding air, position of the fuse, whether it be open or enclosed, vertical or horizontal, and whether it be on the floor, wall, or ceiling. The length of the fuse has also a great effect, the heat being rapidly absorbed from the ends of a short fuse by the blocks to which it is connected; and, furthermore, the size of these pieces would vary the amount of heat which they are capable of taking up.

It is evident, therefore, that the exact conditions under which a given fuse is to be used should be specified, in order to secure uniform results. In fact, fuse blocks or boxes should be standardized, in which cases they ought to be sufficiently reliable in their action, since the physical principles upon which they depend are simple and definite. Their construction is also so very simple and cheap that it would seem to be a mistake to give them up for more complicated devices, except in certain cases. The objection is often urged against fuses that they have a certain capacity for heat which allows the current to rise considerably above the normal point if it increases very suddenly. This is actually an advantage, because the conductors and apparatus which the fuse protects have a still greater heat capacity; consequently a momentary excess of current which does not melt the fuse cannot cause any injury in the rest of the circuit. This avoids the interruption of the supply and the renewal of the fuse every time there happens to be a rush of current for an instant.

A paper and discussion on "The Rating and Behavior of Fuse Wires," by Professor W. M. Stine and others, contains valuable information upon this subject.*

A report by Mr. W. McDevitt to the Philadelphia Board of Fire Underwriters on "The Gross Untrustworthiness of Fuse Metals and Appliances as a Means of Protection for Electric Circuits — The Remedy,"† gives the results of many tests on fuses. It is a question, however, whether the difficulties are not largely or entirely due to improper and variable conditions, that can be avoided as already pointed out.

Electromagnetic Circuit-Breakers, cut-outs, or limit-switches are used in place of fuses to protect electrical circuits from exces

* Trans. Amer. Inst. Elec. Eng., October, 1895.

↑ Electrical Engineer (N.Y.), Feb. 5, 1896.

sive current.

The circuit leads through a helix, as represented in Fig. 138, the electromagnetic action of which automatically opens a switch when the current reaches a certain strength. The final break occurs on the carbon strips on each side, thus saving the metallic switch-contacts from the destructive flashing which takes place when heavy currents are interrupted. Circuit-breakers possess the advantages over fuses that they can be instantly opened by hand if desired, they are easily closed and put in condition to

act again, and they can be set to open with a definite current. As already stated, however, fuses could probably be standardized to melt at a sufficiently exact limit.

In general it may be said that circuit-breakers are suitable for main circuits, or for those which are frequently overloaded, as in electric railway practice, and that fuses are applicable to branch circuits; but the latter are by no means confined to this use, since they are employed almost exclusively in the majority of incandescent-lighting plants. Circuit-breakers are usually preferable on switchboards.

CHAPTER XXIII.

ELECTRICAL MEASURING INSTRUMENTS.

THE subject of electrical measurements is one which should be treated in elementary and general works, or in those devoted to it. But, like the principles of electricity, it usually forms a large part of almost every special treatise as well; although, as already stated, it would seem that these subjects are preparatory to, and not part of, the study of any particular branch. It will therefore be sufficient in the present case to briefly consider the various classes of instruments and their applicability to electric-lighting purposes.

Classification. The most important types of electrical measuring instruments may be subdivided as follows:

1. Ordinary galvanometers. - Coil acting upon permanent magnet.

2. Electromagnet devices. - Coil or electromagnet acting upon soft iron.

3. Electrodynamometers. Two coils acting upon each other. 4. D'Arsonval instruments. Action between coil and mag

netic field.

5. Electrothermal devices. Heat generated by current produces expansion or other effect.

6. Electrostatic instruments. — Mutual action of two charged bodies.

7. Electrochemical devices. — Electrolytic effect of current.

8. Miscellaneous instruments. - Repulsion and other peculiar effects of alternating currents.

Galvanometers. Until within about ten years, practically the only instruments used to measure electric current were simple galvanometers, consisting of a permanently magnetized needle, caused to deflect by passing the current through a coil placed around the needle.

The well-known Thomson reflecting galvanometer still remains the most sensitive means of detecting very feeble currents; but for almost all other uses, galvanometers have been superseded by voltand amperemeters, and other more practical forms of instrument.

The extreme sensitiveness of the Thomson galvanometer is due to five facts: First, the coils usually contain a great many turns of wire; second, they are close to the needle; third, the latter is hung on a single fiber of silk; fourth, the deflection is enormously magnified by employing a beam of light reflected from a mirror; and fifth, the directive force of the earth's magnetism, which opposes the deflection, is neutralized by the use of astatic needles, by a magnet mounted over the galvanometer, or by both. Its principal applications in electrical engineering are measuring the resistance of the insulation of circuits and apparatus, and locating faults in the same. Ordinarily the resistance of insulation has a value of many megohms, consequently it requires a very sensitive instrument to detect the minute current which would flow through such a high resistance. On the other hand, this delicacy is unnecessary, and, in fact, very undesirable, if stronger currents are to be measured.

Furthermore, these galvanometers are disturbed by the slightest vibration, or by any variation in magnetic conditions, as, for example, when a dynamo is started or stopped, even if it be a hundred feet away; the usual result being that the zero point, as well as the deflection corresponding to a given current, are continually changing. Vibration may be taken up by supporting the instrument on springs, or preferably on a heavy metal or stone slab resting upon a cushion of rubber, feathers, or other soft material. The magnetic disturbances may be kept away from a galvanometer by surrounding it with a vertical iron cylinder, the ends of which are covered with iron plates or not, depending upon the intensity of the local magnetic field. But, in any case, holes may be made in the cylinder for observation, etc. These precautions are particularly necessary in an electric-light plant, where vibrations and stray lines of force naturally predominate, and where, in fact, it is almost impossible to entirely eliminate the difficulties which they cause.

The D'Arsonval galvanometer differs from the preceding in the fact that the deflecting part consists of a coil of wire, through