It will be observed that in accordance with equation (3), when the current is rising, the potential is falling and vice versa, as shown. In the lower curve, the peak to the left is the potential at the beginning of the charging period or the extinction voltage. of the arc. (So called because it is the potential across the arc at the instant the arc is blown out by the magnets.) The potential peak to the right is the ignition voltage as defined in a previous paragraph.

Fig. 4 represents a sine curve, or the current obtained from a theoretically perfect alternator. By comparing the upper or current curve of Fig. 3 with that of Fig. 4, a difference will be seen in that the lower half of the cycle in Fig. 3 differs from the upper half, while in Fig. 4 the two were identical.

From the study of sound in physics, it will be recalled that the fundamental vibration of a string or other sonorous body is often accompanied by overtones or harmonics. In fact, the rich quality of a note sounded on a musical instrument is often due to the pleasing combination of the harmonics with the fundamental. The frequencies of these harmonics are integral multiples of the fundamental frequency. The same phenomenon occurs in electrically vibrating or alternating current circuits. Thus, an alternator delivering current at the frequency of 60 cycles for which it is rated may often show harmonics of 180, 300, 420, 540 cycles and higher frequencies. Ordinarily, these upper harmonics are not noticeable since their high frequency causes the impedance of the average circuit, containing considerably. more inductance than capacity, to be so high as to limit their magnitudes and to make them practically non-existent.

When harmonics do exist in an alternating current circuit, however, an oscillograph of the current will show that a sine curve no longer obtains. Instead, the superimposition of the harmonics. on the fundamental frequency produces irregularities in the curve. When the same irregularities occur in each lobe or alternation,

A

it is an indication that odd harmonics only are present. This is typical of the current curve of an alternator whose inherent construction and rotation necessarily insure identical alternations. In circuits possessing distributed capacity and inductance, however, such as telephone and transmission lines, both odd and even harmonics are present. In such cases, the current curve is not symmetrical-the lower alternations being different in shape from the upper. This condition is present in the upper or current curve of Fig. 3 and indicates the presence of both odd and even harmonics in the current across the Poulsen arc.

A radio antenna by virtue of its construction contains distributed inductance and capacity, that is to say, its inductance is not lumped in a coil nor its capacity in a condenser. As a result, its fundamental oscillation is accompanied by harmonics, but-like the alternator-its inherent construction, in this case a grounded lower extremity and an opened upper end, limits its oscillation to odd harmonics only.

It happens, therefore, that when an arc is inserted in an antenna circuit, as is done in the operation of the Poulsen arc, certain of the antenna harmonics, and in most cases all of the lower ones, are reinforced by those harmonics of the arc which nearest approximate them in frequency. This results in the radiation of several wave lengths from the aerial of a Poulsen arc transmitter-the working wave length to which the antenna circuit is adjusted and which is determined by the fundamental formula

where à represents the wave length, v is the velocity of the radio wave, (see fifth paragraph), C is the capacity and L the inductance, and many odd harmonics whose wave lengths are practically integral subdivisions of the working wave length. (It is only in an unloaded antenna, i. e., one in which no inductance coil or condenser is inserted, .that the fundamental wave length is an integral multiple of the wave lengths of the harmonics.)

While the harmonics radiated from an antenna are not comparable in energy magnitude with the working wave length and accordingly produce no effect at a distant station, it happens that they are sufficiently strong to produce considerable interference at nearby receivers, which while not tuned to the working wave length of the transmitter, may happen to be in resonance with one

of its many harmonics. In addition, these short waves may be in resonance with the metallic circuits of the stays and other rigging, thus setting them into oscillation and consequent radiation, with the production of still further local interference.

A complete diagram of the Poulsen arc as actually used in practice is shown in Fig. 5. It will be observed that the shunt. circuit L,, C, of Fig. I has been replaced by the antenna circuit. which includes the antenna, the loading inductance L1, the arc electrodes Cu and C, the antenna ammeter and the ground. In addition, the blow-out magnets, M, have been inserted in the direct current leads supplying the arc.

The operation of the magnets is as follows: We have seen that for the requisite charging of the antenna capacity, it is necessary that the arc be extinguished in order that the direct current from the generator may be diverted into the antenna. The arc when burning is in an ionized state, that is to say, the incandescent electrodes freely give off ions. These ions serve as the medium of conduction of the electric current across the arc just as they do in the electrolyte of a storage battery. When the ions are removed from a liquid or a gas and electrical conduction cannot take place, we say it is deionized. Accordingly, to extinguish the arc, it is necessary to deionize it or to remove the conducting ions from the air gap between the electrodes.

A current of electricity flowing in a wire sets up a circular magnetic field around the conductor. The direction of the lines of force of this field is determined by the familiar right-hand. rule which states that when the thumb of the right hand is pointed in the direction of the current flow, the curved fingers, held at right-angles to the thumb, point in the direction of the lines of force. It has been demonstrated that a stream of ions or isolated

electric charges behaves similar to the electric current in a conductor. (In fact, it is not altogether clear that electrical conduction in a wire does not actually consist of such a stream of electrons.) As a result, a stream of ions sets up about it a circular magnetic field similar to that around a wire.

Let us assume that a stream of ions is flowing away from the reader through the page at right-angles to it. We shall represent the cross section of this stream by a dot. From the right-hand rule, it is evident that there will be set up around this stream a clockwise flux. This is represented in Fig. 6 by the dot and circular arrows.

N

S

FIG. 6.

Let us now set up a magnetic field at right-angles to this ionic flow. This will require the north and south poles of a magnet and the direction of the lines of force of this additional field will be as indicated by the vertical arrows in the figure. It will be observed that the flux to the right of the dot is added to that set up by the ions, while to the left-the ionic flux opposes that of the magnets. This results in a stronger magnetic field to the right of the dot than to the left with the result that the ionic stream is deflected from the stronger to the weaker field as indicated by the arrow at the bottom of the figure pointing to the left.

In the arc, a transverse field is set up at right-angles to the arc flame and in accordance with the principles set forth in the preceding paragraphs, the flame or ionic stream is blown to one side as shown in Fig. 7. The arc starts at the edges of the electrodes at the ignition point I of the figure, and is blown out to the extinction point, E. In connection with the oscillatory current of the

antenna capacity, the magnetic field thus serves to automatically extinguish the arc at radio frequency intervals.

We have seen that at the longer wave lengths, the frequency is lower, consequently the time period, or the length of time in which it takes one oscillation or cycle to complete itself, is greater. It will be apparent, then, that at the longer wave lengths, the greater time period gives us more time in which to deionize or " scavenge" the arc. Consequently, a weaker field is used with the longer tunes than with the shorter.

Aboard ship, where the arc tunes do not cover a wide range, probably one adjustment of the magnetic field will suffice. Ashore, however, where the arc may be used on widely different wave lengths, it is essential that means be provided for varying the field strength. This is usually effected by employing a switch to cut in or out more or less turns of the magnets. Contacts of the switch are marked for the different tunes to which the transmitter is adjusted.

It will be obvious that the field strength must be properly adjusted for the time period-or wave length-of the antenna since if the field be too strong, the arc will be blown out into a fan so quickly that another arc will be started before the first is extinguished. This will result in two or more arcs burning at the same instant with consequent deleterious irregularities in the oscillations.

On the other hand, should the field strength be too weak, the arc will be extinguished before it reaches the point E of Fig. 7. The next arc will be ignited at the extinction position of the first. This secondary arc may also be extinguished prior to the point E, with-if the field be weak enough-the possible formation of a tertiary arc. This ignition of the secondary and tertiary arcs at points which make the arc lengths longer than normal requires an increase in the supply potential with consequent decrease in the over all efficiency.

In addition to the use of the blow-out magnets, two other agencies are employed to assist in the deionization of the arc. One is effective cooling, and the other the use of hydrogen.

An incandescent metal is a prolific radiator of ions. We see this exemplified in receiving apparatus in the radiation of electrons from the filament of the vacuum tube detector. To assist in deionization, therefore, it becomes necessary to cool the electrodes--