Having obtained the two deflections in this manner, we next proceed to determine what is technically known as the "insulation constant," which consists in multiplying the deflection d obtained through the standard in the first operation by the value of the standard R in megohms and fractions of a megohm. This proceeding gives us a number, the magnitude of which will, to a certain extent, determine the accuracy to which it is possible to work in obtaining results; the higher the value of the constant the greater the ultimate accuracy. With a delicate galvanometer and a battery of some 400 Leclanché elements, it is quite possible to obtain a constant having a value of one million, but lower values will, of course, suffice for all ordinary purposes.

Having obtained our constant, all we have to do is to divide it by d1, the deflection obtained through the object under test.

The insulation of a cable is usually stated in megohms per mile or some other unit of length; thus, if the cable under test be two miles long, and we obtain a deflection of 200 deg., the true deflection for one mile of the cable will be 200 deg. divided by 2, or 100 deg.

The electrification may advantageously be plotted in the form of a curve, the degrees and time being registered on the vertical and horizontal ordinates respectively. In this manner any unsteadiness is at once detected by a fluctuation in the curve.

In practice, the insulation tests on a cable are usually taken with the zinc pole of the battery on the cable and the carbon or copper, as the case may be, to earth, but, if a check be needed, and more especially in the case of submarine cables, readings are taken with both poles of the battery consecutively, and should correspond with one another if due care be taken to discharge the cab'e thoroughly between the respective applications of the current.

A further check is provided by what are known as earth readings." If we disconnect the battery from the cable at the end of, say, five minutes, and connect it (the cable) instead to earth through the galvanometer, we shall obtain a series of gradually decreasing readings which should fall practically to zero at the end of another five minutes from the time at which the bat

tery was disconnected.

These readings should be carefully noted at similar intervals to those of the electrification readings. If all is as it should be-i.e., if the insulation of the cable under test be perfect in every respect, the earth reading at the end of the first minute, added to the last electrification reading, should be equal to the electrification reading observed at the end of the first minute, or dl.

In order to arrange for these earth readings in practice, we may introduce a well-insulated plug switch of two segments between the two spring terminals of Kl in Fig. 39; then, when K1 has been finally released at the end of the last electrification reading, the plug can be inserted, thus short-circuiting these two terminals, and connecting the cable directly to earth through the galvanometer G. Care must be taken to remove the plug, however, at the end of the discharge, as otherwise, when K1 is again depressed, the battery E will be shortcircuited.

Now as regards the question of efficient earths. In testing a circuit which has been already wired, we have, of course, to depend upon the best earth we can get, which will usually be a water pipe; water pipes are preferable to gas pipes, as the red lead, etc., employed in jointing the latter tends to introduce a high resistance into the circuit. A small surface should be filed clean and bright on the periphery of the pipe, and the earth connection bound tightly round it, and, if circumstances permit, a soldered connection should be made, as unsteadiness is often caused in otherwise careful insulation tests by the bad contact at an earth connection.

In cases where cables are tested at a factory before being installed, much better circumstances exist for the procuration of a good and efficient earth. The drums containing the cable are immersed bodily in large tanks of water, connection with which is obtained by means of a metal plate or coil of bare wire also immersed. The two ends of the cable are, of course, kept clear of the water, and are prepared and waxed in the usual manner. The cables should be in the water at least 24 hours before the test is taken on them in order to allow the water to percolate thoroughly to all points, and also that the cable may have time to attain the same tem

perature as the water, the temperature of which should. be taken at the time of making the test, for dielectric resistance, like that of copper and other metallic conductors, varies with a variation in temperature.

Lead-covered and armoured cables are usually tested by making the lead covering or armour an earth, as the case may be, and, in all such cases, the earth connection should be sweated or soldered. Concentric cables and twin conductors are tested for insulation from core to core, as well as to earth. Lead-covered cables possessing a fibrous insulation, or such materials as impregnated paper, etc., as a dielectric, should all be immersed in water for testing purposes, as on the imperviousness of the lead sheathing depends the efficiency of the insulation, for, once moisture has obtained an entry by a pinhole or flaw in the lead, the hygroscopic nature of the material at once provides a direct path for it to the conductor, and a "fault" is the inevitable result.

It is a well-known fact that the more perfect the earth which surrounds the insulation of a cable or wire, the higher is the apparent insulation resistance of that cable. I say apparent, because the effect is due to a more rapid electrification and consequent smaller deflection at the end of one minute from the application of the current.

Thus a length of cable immersed in water which provides the earth during the test will show an apparently higher insulation resistance than the same cable when lead-covered and tested with its lead sheathing as an earth. In the same manner, a cable on which the lead sheathing is tight will yield better results than one to which it has been loosely applied. This fact is well worth specially noting, as it is liable to give rise to mistaken impressions regarding the results of certain insulation tests.

Electrification is seldom if ever experienced in testing circuits where the cable has been already installed, as the earth is in such cases very imperfect, and the deflection through the cable or wire will usually be found to settle down at once to a permanent value.

Another matter which should be noted in testing electric light and similar circuits, including a number of fittings, is this-that, although the cable or wire used

may originally have yielded under test an insulation resistance of some hundreds of megohms per mile, it will not indicate anything like this result when wired in position, owing to many reasons, chief amongst which may be mentioned imperfect earth, surface leakage at fittings, etc., etc.

As regards the temperature co-efficient for dielectrics, this varies considerably with the nature of the material, no two samples being exactly alike; it is, in consequence, impossible to lay down a hard and fast rule to suit all insulating media, but, in exceptional cases, an approximate co-efficient can generally be obtained from the manufacturers, who determine it by dint of experiment. As a matter of fact, in cable factories, the testing tanks are usually provided with the necessary steam fittings for raising the temperature of the water to any required. degree, and the tests are taken at that temperature, usually 60 deg. Fahr. for india-rubber and 75 deg. Fahr. for gutta-percha.

Government specifications commonly provide a table of temperature co-efficients of dielectrics for the testing electricians of the manufacturers to work to, but it is very questionable whether such tables are ever correct for the particular sample under test. Red-tape, however, prevails, as usual, even in such details.

The "straining," "stressing," or, as our American cousins term it, "puncture" test of insulated cables and wires intended for use in high tension circuits takes the form of a practical application of an alternating current of given voltage for a definite period, rather than the galvanometer test to text-book rule, described in the preceding paragraphs, although this test should also be instituted in every case, both before and after the stressing test, as I shall proceed to explain.

All insulated cables and wires intended for use on circuits at a voltage of 500 and upwards should, after having been subjected to the usual insulation test by the direct deflection method previously alluded to, be subjected to an alternating voltage at least twice that at which they are ultimately intended to work, whilst immersed in a tank of water at zero potential, or, in other words, connected to earth, or the remaining terminal of the generator. If lead-sheathed or metal-armoured, the sheathing or ar

mouring, of course, takes the place .of the water except in the case of cables insulated with a dielectric of fibrous disposition, in which case, the imperviousness of the lead sheathing governs the value of the insulating medium, and should be tested whilst submerged.

The testing current is usually obtained from an alternator of suitable output, generating at, say, 200 volts, and is raised to the required value by a step-up transformer, or bank of transformers, the primaries (low tension) being connected in parallel, and the secondaries (high tension) in series. The high tension portion of the plant requires to be well insulated, and, to this end, the transformers are usually removed from their metal cases, and mounted on insulating stands, care being taken to connect the secondaries in the right direction, so that they assist rather than oppose one another. A well-insulated lead is taken from one terminal of the secondary circuit to one extremity of the cable under test, the opposite extremity of the latter being left free as in ordinary insulation testing. It is necessary to state here that the extremities of the cable require an equally careful preparation for the prevention of surface leakage in this case as in that previously described under the heading of "Insulation Resistance Measurement." The remaining_terminal of the secondary circuit is connected to earth as represented by the water in the tank, or the metallic sheathing of the cable; if the latter, the connection should be soldered.

The connections as described above are represented diagrammatically in Fig. 40, in which A is the alternator, T1, T2, and T3 a series of three step-up transformers, B an ammeter in the primary circuit, C the submerged cable, Vh and V7 high and low tension voltmeters respectively in the secondary and primary circuits, Vh being of the electrostatic variety. The one Vl serves as a check upon the other Vh, if the multiplying power of the transformers be known. Sh and Sl are high and low tension switches respectively; these two switches should at least be introduced. From the point of view of safety it is better to multiply than reduce the number, as, for instance, by introducing a sub-section switch, s1, s2, and 83, in the primary circuit of each transformer.

The method of conducting the test is as follows: All