the anode, by the circulation of water through the duct in the copper shown in Fig. 8, and the cathode by the rotation of the carbon, which is geared to a small motor. The are flame, which, as we have seen in Fig. 7, burns on the edges of the electrodes, is thus enabled to burn on a fresh and relatively cool portion of the cathode at all times. The circulation of water through the anode makes it impractical to revolve it.

Both electrodes are consumed in the burning of the arc and must be frequently renewed, especially in high power installations. The carbon, being the more volatile, obviously must be renewed much more often than the copper. From Fig. 8, it will be observed that it is possible to permit the consumption, through burning, of only a small portion of the anode since a hole will otherwise be burned through to the water supply.

FIG. 8.

To give some idea of the dimensions of the electrodes, it may be stated that while the diameter of the electrodes in the average commercial arc light is about one-half inch, that of the cathode in a 500 kw. radio converter is about one and one-half inches.

Hydrogen gas is admitted to the air-tight arc chamber to assist in deionization since the velocity of the hydrogen ions is higher than that of the gases comprising the atmosphere. This makes it possible to more quickly remove the ions from the space between the electrodes, with consequent increase in the rapidity of extinction of the arc. At radio stations ashore where illuminating gas is available, it is customary to employ same on account of its hydrogen content. Aboard ship, however, it is necessary to use a liquid hydrocarbon such as alcohol which may be volatilized by the heat of the arc chamber. This is supplied to the arc by means of a drip cup automatically controlled.

If hydrogen is not used in the arc, the presence of air will tend to form oxides on both electrodes. These incandescent oxides are very prolific in their ionic radiation-a familiar example to the physicist is the Wehnult cathode ray tube-and deionization by means of the magnets alone is very difficult. The hydrogen on

account of its strong affinity for oxygen serves as a reducing agent and quite effectively eliminates the formation of these troublesome oxides.

Since both methods of obtaining hydrogen described above entail the use of hydrocarbons, it follows that large quantities of carbon or soot are deposited within the arc chamber. This is increased by the natural disintegration of the incandescent carbon cathode. Frequent cleaning of the chamber must be performed, therefore, in order to maintain the insulation of the anode. (The cathode is grounded to the metallic chamber castings.) The carbon collects on the inner walls of the chamber and if present in sufficient quantity will cover the insulated bushing surrounding the anode holder, thus grounding the "hot" side of the generator. In high power sets, the arc chamber is cleaned daily.

In taking down an arc, a word of caution is necessary. Fully five minutes should elapse after the current is turned off-to enable the electrodes to cool-before the chamber door is opened. If the door be opened immediately, the electrodes will still be white hot and will ignite the explosive mixture formed from the hydrogen in the chamber and the inrushing oxygen of the atmosphere. Such explosions have many times resulted disastrously and too great caution in this respect cannot be exercised. In small arcs, an exhaust valve is provided to take care of the initial explosion occurring after each time the arc is disassembled. In large converters, the exhaust-blower system serves the same purpose.

It is interesting to note that with normal operation of the arc, i. e., the production of perfect oscillations in the antenna, the antenna current is always seven-tenths (0.707) of the input direct current. That is to say, the reading of the antenna ammeter should be seven-tenths of that of the direct current ammeter in the generator leads. This relation serves as a check on the calibration of the radio frequency antenna ammeter.

Referring again to Fig. 5, it is noted that the key is inserted in the antenna circuit. This is different from the spark type of transmitter in which the key is inserted in the lead to the alternator as shown in Fig. 9. In the Poulsen arc transmitter, it is not possible to open and close the power supply in the act of signaling since the arc electrodes must be closed to strike the arc each time the circuit is opened. (It is true that the arc flame is extinguished

at each cycle of the antenna current, but these interruptions occur at radio frequency intervals and, the electrodes being incandescent, the arc may be reignited by the antenna potential. This is not similar to the relatively slow key interruptions of signaling.) Accordingly, the arc must be left burning continuously and such variation as is made must be effected elsewhere. Usually, this is done in the antenna circuit.

The oldest method is to shunt some of the antenna loading inductance with a key. If connected as shown in Fig. 5, see K, the radiated wave will be shortened with each depression of the key since the shunted portion of the antenna inductance is in effect removed from the circuit. See equation (6). The signaling or shorter wave is called the working wave, the wave length radiated when the key is in the "up" position is termed the compensation wave. At the receiver, the operator of course tunes to the working wave and the compensation is ordinarily not heard since the receiver is not tuned to resonance with the longer wave. If the receiving operator should tune to the compensation wave, he would hear unreadable signals corresponding to the spaces between the dots and dashes of the transmitter.

This type of key may be arranged so as to make the compensation wave shorter than the working wave. Instead of placing the key contacts at the front of the key lever or forward of the fulcrum, as shown in Fig. 5, they are mounted at the rear end of the key. With the "back-connected" key, the circuit is closed, and the wave length shortened, when the key is up, and is openedwith increased wave length-when the key is depressed.

It will be observed that with the compensation type of key control two wave lengths are radiated from an arc transmitter-the compensation and the working. This results in just twice the interference from any one arc transmitter, so efforts have been made to design a key system which would radiate waves of but one length as with the modern spark systems. Experimentation led to the

"tank" system diagrammed in Fig. 10, the principle of whose operation is as follows: To the rear end of a key lever is fastened a contact, B. This engages with the contact A when the key is depressed and C when it is released. A is connected to the antenna while C is connected to a dummy antenna or tank circuit, L, R. and C and the earth, the inductance, resistance and capacity of which are adjusted to approximate the respective constants of the antenna circuit. The antenna current ammeter should show no variation in its reading when the arc is switched from the dummy to the real antenna. The tank circuit is necessary in order to keep stable the oscillatory conditions across the arc since perfect signaling cannot be obtained by merely switching the antenna on and off the arc. When not connected to the antenna the arc must be connected to a similar oscillatory circuit.

Such a circuit as described radiates of course but a single wave length since the tank circuit is a closed-and consequently a practically non-radiating-circuit. It happens, however, that in the act of switching from the tank to the antenna there is an instant in which the arc is not connected to either and since it has no oscillatory circuit to keep it functioning smoothly, irregularities in the oscillations develop which spoil signaling. A new system was developed, therefore, which was intended to keep the arc always connected to some form of oscillatory circuit-either the tank or the real antenna.

Fig. 1 shows this new system which embraces the use of two "reactance" keys, one in the antenna and the other in the tank circuit. This type of key is really a transformer, with a primary DC

through which is sent direct current and a secondary through which the radio frequency current flows. The primary of the antenna key is labeled P. the secondary S, while the primary and secondary of the tank key are correspondingly lettered P 1 and ST.

If direct current is sent through a coil of wire surrounding an iron core, lines of magnetic force parallel to the axis of the core are set up within it. An increase in the amount of current flowing through the coil will result in an increased magnetic field until finally a point is reached where an increase in current will produce practically no further magnetization. In such a condition, we say the core is saturated, just as a sponge may be so saturated with water that it can absorb no more. If, with the core saturated, we now pass an alternating current through the coil, no reactance will be offered this alternating current since it cannot set up