Page images

placed at various distances from the needle; the coils can also be rotated as a whole about a vertical axis passing through the centre of the needle. The needle consists of a steel ring, suspended with the plane of the ring vertical.

Chrystal gave a very complete explanation of the phenomenon in Phil. Mag. p. 401, vol. ii. 1876; his explanation depends on the assumption-no doubt correct-that the magnetization of a steel needle can be caused to vary by a magnetic field, however small.

Suppose that, through any cause, the needle makes an angle with the plane of the galvanometer coils: owing to the fact that a needle is most readily magnetized lengthwise, that component of a magnetic field parallel to the needle produces a decided change in the magnetization lengthwise, whilst the action of the field in any other direction is relatively small. It is unnecessary to go into details, but it is found, on examination, that the field, with lines at right angles to the plane of the coils, which tends to increase the angular deflexion increases the magnetization of the needle on the other hand, the field which tends to diminish the angular deflexion diminishes the magnetization of the needle. Hence in the case of alternations the currents that tend to increase the deflexion give greater impulses than those in the other direction, and in consequence the angular deflexion of the needle is increased. The same would happen if the needle were not magnetized.

Since the change in the magnetization of the needle will vary approximately as the strength of the alternating current, and the action on the needle varies as the product of the change and the strength of the alternating current, it follows that the action on the needle will vary approximately as the square of the alternating current. This I proved indirectly by using an alternating current of constant strength, and placing the coils at various distances from the needle, having previously determined the relative sensibilities of the galvanometer for the various distances with a steady current.

Still following Chrystal, it is obvious that if a magnetized steel sphere, or any steel body bounded by a surface of revolution whose axis is that of suspension, be suspended, the superimposed magnetization will always be parallel to the lines of force in the field, and hence there will be no turning moment due to this cause.

I suspended a magnetized steel sphere in place of the ordinary needle, and was surprised to find that when the angular displacement of the magnetic axis with respect to the coils was in one direction, an alternating current increased

the deflexion: when the angular displacement was in the other direction, an alternating current diminished the deflexion. A magnetized steel ring behaved in a similar manner; in this case I proved experimentally (indirectly) that the action on the needle varied approximately as the square of the alternating current; also on resuspending the ring so as to make what was previously the lower side the upper one, the direction of the deflexion was reversed. An explanation did not strike me at the time of experimenting, but perhaps the true explanation is that, owing either to some defect in homogeneity or in geometrical shape, the direction most susceptible to magnetization was inclined at an angle to the magnetic axis. Suppose, for example, that the angle were 45°: since the angle which the direction most susceptible to magnetization makes with the plane of the coils is increased by an alternating current, it is obvious that the magnetic axis could be rotated 45° in either direction before there was a change in the direction of the deflexion produced by the alternating current.

Although, undoubtedly, the phenomena described above are almost entirely due to magnetic causes, it was thought advisable to try some experiments with a copper (nonmagnetic) ring.

It must be mentioned that the Edelmann galvanometer was supplied with dampers which consisted of two hollow metal cylinders capable of movement to various distances from the needle. The copper ring was suspended with a diameter vertical and capable of rotation about that diameter; when the dampers were pushed in as far as possible, the angle between the plane of the ring and the plane of the coils was increased by an alternating current; when the dampers were drawn out, the angle was diminished. These repulsions and attractions are no doubt caused by the interactions of the original alternating current in the galvanometer coils, and the induced currents, of various phases, in the copper ring and dampers.

The copper ring was sawn across, so that the circle was no longer complete; the action on the ring of an alternating current in the galvanometer coils was now almost negligible.

A steel ring was sawn across, so that the circle was no longer complete, and then magnetized and suspended with a diameter vertical. An alternating current produced a powerful action on the suspended needle, although no induction-currents of any importance could exist.

In the preceding experiments I was troubled with slight irregularities which I attributed to imperfections in the alter

nating current, and I attributed the imperfections, perhaps erroneously, to slight irregularities in the make and break.

The interruptions in the primary circuit were caused by a platinum point leaving a mercury surface which was covered with methylated spirit. The spark seemed to me to be an erratic factor, and I endeavoured to get rid of it by employing a resistance as a shunt across the spark-gap, by employing a condenser as a shunt, and in particular by a method of Differential Winding, which was only a partial success, described by S. P. Thompson in The Electromagnet and Electromagnet Mechanism.'

The method of Differential Winding diminishes the spark on break, but certainly does not eliminate it; as the action of the arrangement is of some interest, I have given a full explanation of its construction and behaviour in Addendum I.

The device for preventing sparking which acted the best, and which I think is a new one, consists in the use of electrolytic cells or small batteries placed as a shunt across the spark-gap.

I first used electrolytic cells, the platinum plates of which were about 1 in. by ğ in., and dipped in dilute or strong sulphuric acid.

The explanation is that the cells polarize, and on making the gap an E.M.F. is introduced opposed to the E.M.F. of the battery, so that the current rapidly diminishes, decomposing the liquid and doing chemical work.

I may mention, incidentally, that the rapid make and break of current through the electrolytic cells disintegrated the surface of either the positive or negative plates (I did not observe which) and brought platinum into suspension, in a very finely divided state, throughout the liquid.

I tried plates of gas-charcoal in dilute H2SO4; these did not act so well as the platinum.

I tried lead plates in dilute H2SO4; these acted quite as well as the platinum plates, and are of course cheap.

When the above-mentioned device was employed there was no trouble with the mercury; in fact the little sparking there was seemed to purify rather than contaminate the surface.

Having now had some experience with alternating currents, I was recommended to examine some of Lenard's experiments on the electrical resistance of bismuth in a magnetic field with constant and alternating currents.


The telephones and bridge employed were those which are used for the determination of electrolytic resistances by Kohlrausch's method; the bridge-wire is wound on a cylinder.

*Wied. Ann. xxxix. p. 619 (1890).

To eliminate any possible errors due to the self-induction of the bridge-wire, the resistance of the standard arm was always varied until the reading of the bridge was not far from the

middle of the wire.

I no longer employed a tuning-fork as interrupter, but used an apparatus discussed in Addendum II., which enabled me to obtain any number of breaks per second up to 500.

To explain the fact, discovered by himself, that a bismuth wire in a strong magnetic field has apparently a higher resistance when measured with a telephone and an alternating current than with a galvanometer and constant current, Lenard frames the bold hypothesis that it is not the frequency of by far the greater part of the alternating current that has to do with the increased resistance, but accidental (so to speak) oscillations with a frequency of about 10,000.

With the bismuth spiral in a strong field absolute silence is never obtained in the telephone, but there is a decided minimum noise of a nondescript character.

No differences of resistance could be detected with frequencies between 60 and 500.

It is well known that an alternating current tends to travel along the surface of a conductor, especially when the conductor is magnetic; the result is that the resistance for an alternating current is greater than for a steady one.

Let be the length of a straight wire, R the resistance to steady currents, p/2π the frequency of vibration, μ the magnetic permeability for circumferential magnetization, R' the resistance to alternating currents; then if p222 is small,

[merged small][merged small][merged small][subsumed][ocr errors][merged small][merged small][merged small]


} approx.*,


where resistance per unit length.

r =

In the case of a copper wire 1 millim. in diameter, where μ=1, the increase of resistance with an alternating current of frequency 10,000 equals about one tenth.

[ocr errors]

The formula becomes unworkable when p2 is large;

but it seems probable that if a current of 10,000 exists side by side with one of 100, the use of sufficient iron wire of sufficient diameter would enable us to diminish the strength of * Lord Rayleigh, Phil. Mag. vol. xxi. 1886, p. 387.

the former considerably, while diminishing the latter to a much less degree.

I employed a length of iron wire which had a resistance equal to 1845 ohms, l=4675 centim.; taking μ=100, we get when p=2π × 10,000:

[ocr errors]

= 250 approx.

The wire was placed between the secondary of the inductioncoil (which had, as arranged, a resistance of about 24 ohms) and the Wheatstone bridge.

No very appreciable change was produced in the intensity of the sound of the telephone when the iron wire was introduced, nor was any change produced in the reading which gave the position of minimum sound.

The use of the telephone, as being too comprehensive in its record, seemed to me in some respects unsatisfactory; and I thought that possibly a method might be useful which enabled the telephone to be replaced by a galvanometer, whilst still retaining an alternating current.

It has already been mentioned that the interruptions were produced by means of a vibrating wire, a platinum point attached to the wire dipped in and out of mercury; to the same vibrating wire was attached, by means of an insulator, some platinum wire in the form of an inverted U, the lower tips of which just dipped in some mercury contained in two little vessels. The telephone was replaced by a galvanometer, and the latter was so connected to the bridge that the galvanometer-branch was only complete when the platinum wire (in the form of an inverted U) dipped into the mercury in the two little vessels.

This arrangement, as was to be expected, gave correct results when the resistance of a german-silver wire, for example, was determined by means of an alternating current; but when the attempt was made to determine the resistance of a bismuth spiral in a strong magnetic field, the method failed. Matters were sufficiently stable to allow of observations, but the results obtained varied with the frequencies, and even with the same frequency were not constant from hour to hour. Since readings could be obtained, and since of a second is no doubt small in comparison with the time the galvanometer-branch was complete, it seems tremely probable that the action of the bismuth spiral depends on some action upon the alternating current of the frequency I was endeavouring to employ (about 50). It seems likely also that the action of the bismuth is not a simple one, i. e. there is perhaps not merely (if at all) an




« PreviousContinue »