small piece (c) of cylindrical hard-rubber rod, in which a groove was made down to the axis, along a plane through

the axis, was fastened to the hard-rubber plate (p) by wax, and the detector was then hung on the stretched wire at any place desired.

The curves are plotted from the mean of at least three sets of readings.

The published accounts of the exploration of wires about which electrical disturbances are produced as in wireless telegraphy, are not numerous, and, as far as I can learn, in no case has the exploration been at all minute.

In the present investigation three methods, well-known in practice, have been used to excite the oscillations.

Marconi's Simple Method.

This arrangement is illustrated in fig. 2. The " oscillator" in this case consisted of two cylindrical brass rods, AB, CD, 9.5 mms. in diameter and 12.5 cms. long, ending in spherical knobs B, C, 3.8 cms. in diameter. One half of the doublet is shown on a larger scale at the left-hand of fig. 2. From Fig. 2.

D led off the antenna DE. In some experiments A was connected to earth, in others a wire similar to DE was attached to A, while in one series this end of the doublet was left entirely free.

The knobs B, C were not kept polished, and the spark was about 1.9 mm. long.

For earth, in the case of wires of lengths 500, 1500, and 2000 cms., A was joined immediately to a large sheet of tin which, along with about two square metres more of sheet metal, was firmly connected to a steam-heating radiator near by. For the wire 1000 cms. in length the connexion from A to the sheet of tin was about 75 cms. long.

The wire joined to A in place of the earth connexion was precisely the same as that acting as antenna and attached to D; and in order to prevent inductive effects between these two wires the former was drawn up in a vertical direction by a cord over a pulley in the ceiling.

The readings were taken at points, usually 20 cms. apart, from one end of the antenna to the other, the readings in some cases beginning at the free end, in the rest ending at it.

A general view of the results obtained is given in Table I. and fig. 3. In the table the distances of the minima from the free end are given in centimetres, and the less-marked minima are inclosed in brackets. In all the curves ordinates denote magnetometer-deflexions, abscissæ distances from free end of wire.

1000

1500

260. None.

(120), (375), (500). None. (150), (660). None. 130, 425, 715, (1000?) (320?). None.

2000 None.

It is seen from the curves, that joining to earth one pole of the oscillator is equivalent to adding to that pole a wire similar to the antenna; or, in other words, the earth acts like a plane mirror in optics. This view has been put forward by several writers, especially by Slaby*, when offering an explanation of his system of syntonic telegraphy.

A. Slaby, Funkentelegraphie, 2nd ed. p. 86, and fol. Berlin, 1901.

In the curve obtained with the antenna of 1000 cms. connected to earth (see fig. 3), there is a deep minimum at

In the curve obtained with no earth or other connexion the natural oscillation of the wire as a whole is practically absent, but there are minima at distances

130, 425, 725 (1000?) cms.

Phil. Mag. S. 6. Vol. 7. No. 38. Feb. 1904.

K

That at 1000 cms. is not decisive from the curve, and so it is omitted in the following calculation (though including it would make no difference in the result) :

These, I believe, are half wave-lengths of overtones. In the first case the wire was grounded and so only odd overtones would be possible, the one present being probably the ninth, counting the fundamental the first. If such was the case, the entire length of the oscillating wire from free end to earth should be

a result requiring the oscillator to be equivalent to

1120 (1000+75), or 45 cms.

of the wire. This explanation seems to me the most probable. I may remark that the curves obtained with the wire 1000 cms. long, connected to earth, were the most irregular of all secured during the investigation, especially in the space between 100 and 300 cms. from the free end. A possible cause contributing to this may have been that the electrical disturbance was not produced immediately at the earth end.

In the second case the oscillating wire was free at each end, and so the entire system of overtones was possible. The one present, with half wave-length of 282 cms., seems to be the fourth, in this case the oscillator adding to the wire one-fourth of a wave-length.

It may be questioned why these particular overtones were present, and the others not noticeable. I think it was because the natural period of the oscillator alone was in approximate accord with them, being about one-half that of those exhibited. This would agree with the results of Lindemann*, who found that the waves proper to the oscillator as well as those of the entire system of oscillator and wires should be present.

I have not been able to identify the other ripples of the

curves.

Slaby and Braun ‡ have both studied the simple Marconi

* A. Lindemann, Ann. der Physik, ii. p. 376 (1900).

†A. Slaby, Elektrotechnische Zeitschrift, 1902, p. 168; extended abstract in Lond. Electrician, vol. xlix. p. 6 (1902).

F. Braun, Phys. Zeitschrift, iii. p. 143 (1900).

Coil

system. The former used a wire about 10 metres long, and explored it with a spark micrometer in which a blunt metal cone was opposed to a flat face of arc carbon. According to the curve he obtained (fig. 1 of his article), there was a standing wave, with potential loops at the ends and a relative node in the middle. In my experiments there is a node at the end of the wire attached to the coil. Slaby concluded that the overtones present were very trifling and that the oscillation emitted was almost a pure fundamental. The fundamental is certainly present in great intensity, but the readings giving it are sometimes scattering, as mentioned above, and the curve is not very smooth. In some cases, too, as already seen, overtones show in considerable strength. Slaby also found that when the pole of the induction-coil was joined, not to an end of the antenna, but to some other point, the oscillation produced showed considerable distortion. This effect is similar to that noted above in the case of the 1000 cms. earthed wire.

Braun used a wire 15 metres long stretched horizontally, and from it suspended five small Geissler tubes, each with a wire 50 cms. long hanging below it. When the coil was in action the tubes lighted up, but there was no trace of a node or a ventral segment.

Inductive Method of Excitation (Braun, Marconi). The experimental disposition used for inductively exciting the oscillations about the antenna is illustrated in fig. 4.

Fig. 4.

Barth or

Capacity

Ante

C1, C2 are two condensers. From the inner coatings conductors lead off and end in knobs, between which sparks are made to pass by an induction-coil. The outer coatings are joined by a thick wire bent into a single turn which acts as the primary of a transformer. The secondary of this transformer (AB) consists of a few turns. To one end of it (A) the antenna is joined, and to the other end (B) the earth, any desired capacity, or a wire similar to the antenna.

The apparatus actually used in the investigation was the transmitter of the experimental set supplied by the Gesellschaft für drahtlose Telegraphie, Berlin, Germany, of the system Prof. Braun and Siemens & Halske.

Each

condenser consisted of four small tubular jars, 17.5 mms. in diam., 2 mms. thick, and with coatings approxinately 7.5 cms. high. The spark-gap was from 1 to 2 mms.