Beginning in 1898, Reginald A. Fessenden worked to develop a complete radio transmission and receiving system, that didn't infringe on any competitor's patents, and could also transmit audio, not just dots-and-dashes. Fessenden was ultimately successful, and on December 21, 1906 gave a demonstration of the new alternator-transmitter to invited representatives from a number of organizations. However, the main target was the American Telephone & Telegraph Company, whose review of the test appeared as a front page article in The American Telephone Journal, an AT&T publication. Fessenden and his financial backers dearly hoped AT&T would be so impressed it would buy the rights to the patents which covered the new system. The outcome of this presentation is reviewed at the close of this article.

This AT&T review noted that wireless telephony was "admirably adapted to the transmission of news, music, etc." simultaneously to multiple locations. Three days after the presentation reviewed in this report, on the evening of December 24, 1906, Fessenden would use his new alternator-transmitter to give what is generally considered to be the first broadcast of entertainment by radio, as part of the ongoing promotion of the new system.

One item of interest is that this demonstration took place in the middle of winter. The review mentioned the "lack of susceptibility to the foreign influences which produce disagreeable noises", but had the test taken place in the summer, they would have heard a tremendous amount of static whenever there was a passing thunderstorm, due to the station's extremely low operating frequency.
 

The American Telephone Journal, January 26, 1907, pages 49-51:

Experiments and Results in Wireless Telephony

BY JOHN GRANT

WIRELESS transmission of speech over a distance somewhat greater than ten miles was satisfactorily accomplished in the presence of a number of persons invited to witness demonstration of a new system of wireless telephony at the experimental station of the National Electric Signaling Company, Brant Rock, Mass., on Dec. 21st, 1906.
    The representative of THE AMERICAN TELEPHONE JOURNAL who was present at these tests was furnished by Professor Reginald A. Fessenden, the inventor of the system, with many facts which have made it possible to trace in the present article the development of his work. For this purpose abstracts without quotation will be freely made from an article which has been furnished by him for publication in Electrical Review, London, and from descriptions embodied in United States patents which have already been issued.
alternator

    Speaking broadly, wireless telephony by this system is accomplished by generating a practically continuous succession of electromagnetic waves, modifying the character of the emitted impulses by means of sound waves without interrupting their continuity, and receiving them in a constantly operative receiver of suitable form which controls a local circuit containing a battery and a telephone receiver. The apparatus which was seen in successful use at the time of the recent tests is the result of a series of diligent investigations in which a large amount of work was done to show the necessity of rejecting plans which did not lead to the required quality of transmission.
    Beginning his work on the subject in 1898, Professor Fessenden made some experiments which were entirely unsuccessful. At this time the only recognized means for the practically continuous generation of electromagnetic waves capable of being propagated through space to affect a distant receiving instrument were:
    (a) The plain aerial with spark gap used by Marconi.
    (b) The plain aerial heavily loaded with inductance, used by Lodge.
    (c) The plain aerial in conjunction with a local oscillatory circuit having a period of a different order of magnitude from the period of the antenna, used by Braun.
    Lodge's method was found to be the only one adapted to produce prolonged trains of waves. Tietz at an early date had used Leyden jars connected across the spark gap, and later Braun described and used a Leyden jar and antenna sending circuit in which the natural period of the Leyden jar circuit was specified as of a different and lower order than that of the antenna circuit. [Braun, English patent No. 1,862, A. D. 1899]
    None of these methods, however, gave the results desired. Professor Fessenden conceived the idea that good results could be obtained in conjunction with a local circuit tuned to the same frequency as the aerial. [U. S. Patent No. 706,735]
    This method, used by him in association with Professor Kintner, proved to give a fairly satisfactory means of producing a long train of waves, and is now extensively used. After making various tests with a Wehnelt interrupter and other devices with which more or less encouraging results were obtained, an induction coil and commutator were settled upon as a make and break mechanism for the tuned circuit. With this circuit, and apparatus giving 10,000 sparks per second, the experiments in wireless telephony led to the transmission of speech, which was first accomplished in the Fall of 1900. The antennae were two masts, 50 feet high, set up one mile apart at Rock Point, Md. A commutator making 10,000 breaks per second in circuit with an induction coil was used for generating waves.
nitrogen spark gap diagram
nitrogen spark gap

    In these experiments the articulation was of a sort which left considerable room for improvement, and there was a noise, due to the irregularity of the spark, which was disagreeable and at times overpowering. This lead to the invention of the compressed gas spark gap, [U. S. Patent No. 706,741] which gave a steadier spark. This device is essentially a spark gap having its terminals, 4, 5 (Fig. 1), enclosed in a chamber in which the gas may be subjected to a pressure produced by the pump 8. This spark gap is connected between the ground and the antenna, shunting the source of energy, the circuit of which contains a make and break device. In practice the chamber was filled with compressed air, from which the oxygen was absorbed by lime in the bottom of the chamber, leaving compressed nitrogen. The appearance of the exterior of the apparatus is shown in Fig. 2. Later a mercury gap of the Cooper-Hewitt type was used, but with this the results obtained were not quite as good as with the compressed gas gap, even when the spark was localized as much as possible by small points of platinum-iridium wire projecting to the surface of the mercury.
    With these types of apparatus high speed breaks of various kinds were used. In 1901 and 1902 experiments were made, using Elihu Thomson's method of producing rapid oscillations by means of an arc and shunted resonant circuit. Better results were obtained by a modification of this method, using regulating resistance, compressed gas gaps and governing circuits for the purpose of making it more applicable to practical working, but there was still a very considerable amount of foreign noise in the telephone circuit.
    Work on high frequency alternating current dynamos had been begun in 1900, and in 1902 an alternator giving 10,000 cycles per second was completed at the works of the General Electric Co. and delivered to Professor Fessenden. This was a 1 kilowatt machine, delivering about 10 amperes at 100 volts. With it was used an air core transformer giving about 10,000 volts, and an interrupter producing 20,000 sparks per second. It was necessary to use the spark gap, as the frequency of the machine was not high enough for the direct production of electromagnetic waves. This combination, however, on account of the regularity of its action, gave much better results than the rotating break, and measurements made in Washington in 1904 led to the belief that transmission could be effected over a distance of 25 miles.
    Continued experiments were made with spark gap apparatus of various types, and in many cases fairly good articulation was obtained. With all these types of apparatus, using a spark gap however, while radiation was sufficiently continuous for the transmission of speech, it became more and more evident that the quality of articulation demanded for commercial telephony could not be obtained without a source of power which would give completely continuous radiation. Among the many methods for obtaining this which were tried was the very interesting method of producing high frequency oscillations commonly known as the musical arc, using a continuous current arc shunted by a condenser and inductance in connection with a magnetic blow out, invented by Professor Elihu Thomson.
    Professor Fessenden as early as 1898 had by his experiments verified the statement made by its inventor [In U. S. Patent No. 500,630] that frequencies as high as 50,000 or more can be obtained in this way. This statement, it is interesting to note, was controverted as late as 1903 by Duddell, [London Electrician, Vol. 51, Page 902] who seems to have not fully grasped the method of operation of the device, although many European scientists have even incorrectly attributed its invention to him.
spark gap regulator circuit
alternator

    The experiments made with this method of producing oscillations showed it to be hardly satisfactory. By the use of properly cooled electrodes and an air blast and magnetic blow out, very high frequencies were obtained, but it was found that neither frequency nor intensity was constant. The fact that a key could not be used to make and break the circuit, since the arc would not start itself, made it impracticable in its original shape.
    In order to overcome the difficulties arising from irregularity, the plan was modified by the substitution for a pure inductance in series with the arc of a coil having a considerable resistance with only a moderate amount of self-induction. This resistance was so adjusted and proportioned to the shunt resonant circuit as to maintain the frequency almost absolutely constant. [U. S. patent No. 730,753.]
    In Fig. 3 the coil 59 is shown in series with the arc in the sending circuit. It is so designed as to have a high resistance but low inductance, and any suitable means, such as a plug, 60, may be provided for shunting out more or less of the resistance. In operation, when condenser 12a has been charged to a sufficient potential, there will occur a discharge across the spark gap, discharging the condenser and setting up oscillations in the sending conductor. On account of the high resistance, 59, some time is required to recharge the condenser to sparking potential. The discharge is therefore intermittent, and may be made to occur many times per second as is desired, within recognizable limits, by plugging out more or less of the resistance. With this apparatus the periodicity depends upon the discharge voltage, which is not liable to fluctuate.
    To overcome the difficulty arising from the inability of the arc to start itself, a method of working was devised in which the arc operated continuously, and emitted radiation continuously, and the signalling was done by altering the frequency of the emitted waves. [U. S. patent No. 706,742]. Numerous experiments looking to the adaptation of this plan to telephony were made, but it was found that by none of the arrangements tried could the scratching and hissing noises in the receiver be eliminated. While these experiments were being carried on, work on the development of a new high frequency dynamo was making good progress. In the only patent which has yet been granted on this machine [U. S. Patent No. 706,737] its general characteristics are described as follows:
    It is necessary that it should give a pure sine wave, as such a form is the only one adapted to give perfect resonance. With a dynamo giving such a curve forming a part of a suitably constructed sending conductor, Professor Fessenden asserts that if the machine be wound to give a thousand volts on open circuit, it is possible by means of resonance effects to obtain a voltage of 100,000 volts on the sending conductor. These resonance effects are obtained by using a dynamo of low internal resistance as a portion of the sending conductor of large capacity or self-induction, or both, having these electrical constants suitably proportioned to give to the sending conductor, that is, to the whole conductor from the top of the antenna to the ground, including the armature of the dynamo itself a natural period identical with the periodicity of the dynamo. If the frequency of the dynamo were to be made lower than the periodicity of the radiating circuit the chief effects would be electrostatic and magnetic in their nature, and there would be practically no electromagnetic radiation. As it is only energy in the form of electromagnetic waves which may be transmitted to a great distance through the atmosphere, it is highly important that this effect should be predominant. The armature must have a low resistance, because if of a high resistance the oscillations will be dampened, making it impossible to produce high resonance voltages. Ventilation must be good, as the current may run up to a very high figure. The length of wire in the armature must be as small as possible, compared with the length of the sending conductor. If this relation were not maintained, the electrical constants of the entire sending conductor would be determined too largely by that part of the circuit between the armature terminals, and the amount of radiation
Brant Rock station mast
would be much less than would be the case if the armature had a relatively small length of wire. Another way of stating this requirement is that the self-induction and capacity of the armature must be as small a fraction as possible of the self-induction and capacity of the entire sending conductor in order to secure the highest radiating efficiency. It is also essential that all iron magnetically influenced by currents in the conductor should be so proportioned and distributed as not to affect the shape of the curve of voltage, or to cause loss of power by hysteresis, as in such a case there would be too much dampening. For these reasons the dynamo may be constructed with a fixed armature containing no iron, having the air gap as long as possible, consistent with a high magnetic flux density, and revolving pole pieces so shaped as to produce sine waves as closely as possible. The revolving parts may be formed of magnetic material of high tensile strength, such as nickel steel. A peripheral speed of five miles per minute, which can be safely maintained with properly constructed moving parts of nickel steel, would allow the machine to be arranged to give one hundred thousand cycles per second. Such a speed can be obtained with a steam turbine to drive the dynamo.
    The alternator which is at present in use is constructed along those lines, but embodies many ingenious mechanical arrangements due to the skill of several of the engineers of the General Electric Company, notably Dr. Steinmetz, Mr. Haskins, Mr. Alexanderson, Mr. Dempster, and Mr. Geisenhoner.
    This machine (Fig. 4) was originally designed for a frequency of 100,000 cycles, at an output of one kilowatt. It is now being driven by belting, the construction of a type to be driven by a De Laval turbine connected through gearing, and, on account of belt slipping, is never run at a speed to give more than 80,000 cycles. For most work it is run at 60,000 cycles, at which speed it has an output of about one-quarter of a kilowatt. The internal resistance of the armature is approximately six ohms, and the inductive drop at full load is about equal to the ohmic drop. At 60,000 cycles the voltage is about 60 volts. The armature makes 10,000 revolutions per minute, bearings being kept at a low temperature by lubrication controlled by oil pumps. The operation of the machine is said to be extremely satisfactory, it having been run daily for six or seven hours at a time with practically no attention. The design, and the method in which it has been worked out by the engineers and mechanics of the General Electric Company, mark a notable advance in dynamo-electric machinery, for which the highest credit is due those who have developed this machine, accomplishing what has been declared by Fleming, in his latest published work in wireless telegraphy, to be an impossibility.
(To be continued.)

February 2, 1907, page 68-70, 79-80:
transmitter circuit
condenser transmitter
MODIFICATION of the character of the electromagnetic waves to impart the fluctuations characteristic of the current in a circuit containing an ordinary telephone transmitter has been the object of an exhaustive series of experiments by Professor Fessenden, second only in importance to those which led to his development of a satisfactory system for radiating energy. It is evident that the forms of the electromagnetic waves must be varied exactly in correspondence with the sound waves of spoken words at the transmitting station, and at the receiving station the apparatus must be capable of transforming the energy into sound waves of like character to those originated at the distant end of system. An early arrangement tried at the transmitting end of the line was of the form indicated in Fig. 5. [U. S. Patent No. 706,747] Here the conductor from the aerial passes through a winding 2 of the transformer 3 to the source of energy (here represented by induction coil the other terminal of which is connected by induction coil 6), the other terminal of which is connected to ground. Capacity 18 and in inductance 19 in series shunt spark gap 4-5 for the purpose of maintaining constant frequency, as previously described with referenced to Fig. 3. Transmitter 9 and battery 8 are serially included with a second winding 7, on transformer 3. Capacity 18 and inductance 19 are arranged to have the same period of oscillation as the sending conductor 1, and also as the receiving conductor. Advantage of the fact that if the resistance of a transformer secondary be changed it alters the inductance of the primary is taken to produce the required modifications the waves emitted. Thus by speaking into the transmitter the permeability of the core 3 is correspondingly modified, producing a change in the self-inductance of the winding 2. This in turn affects the natural period of vibration of the sending conductor, throwing it out of resonance with resonating circuit 18-19. Owing to this variable failure of resonance there is produced a series of corresponding changes in the intensity of the waves given off by the conductor 1, and these variations are reproduced in the circuit of the receiving conductor. It is to be noted that the essential point in the operation of this method of transmission is the throwing of the aerial out of tune with the resonant circuit 18-19, and an alternative method of doing this is to alter the capacity of conductor 1, instead of its inductance. To affect this type of variation, conductor 1 was connected to a fixed condenser plate 13 (Fig. 6), while plate 14 is formed by or connected to a diaphragm capable of vibrating in unison with sound waves, produced by words spoken into a transmitter mouthpiece. The latter arrangement has been termed by Professor Fessenden a "condenser transmitter." Relating the practical results obtained with this type of apparatus in conjunction with the high frequency dynamo for generating waves he states that with a diaphragm two centimetres in diameter a movement of 1-100 of an inch inwards reduced the current from 3.1 amps. to 2.5 amps. This result was obtained on a circuit used for telephoning from Brant Rock to Plymouth, a distance of about ten miles. The dynamo was connected to the aerial through a transformer with 10 and 100 turns respectively, stepping up the voltage from 45 volts to approximately 3,000 volts, with a frequency of 50,000 cycles. This result was obtained without a resonant circuit between the movable terminal of the condenser transmitter and ground. grapohone layout
    In Fig. 9 is shown a third arrangement using a carbon microphone transmitter, 16-17, in circuit between the sending generator 15 and aerial 1. A proper type of transmitter for this purpose should be capable of carrying from 10 to 100 amperes. In the practical instrument which has been developed the metal enclosing the carbon chamber is made with two deep circumferential grooves, visible in Fig 11, permitting the rapid radiation of such heat as may be produced.
    In operation, the sending conductor has its natural period in resonance with the period of the dynamo, and the amount of resonant voltage depends upon the resistance of the microphonic contact. Speaking against the diaphragm therefore causes the voltage at the aerial terminal to change in correspondence with the sound waves. This microphonic contact may be substituted for the variable inductance or variable capacity in conjunction with the resonant circuit 18, 19 as shown in Figs. 5 and 6. While the condenser transmitter has given the best results, the carbon transmitter works very well in practice. The instruments as now constructed have platinum-iridium electrodes, and carry three amperes without injurious heating. To get the best results the ohmic resistance of the carbon transmitter should be equal to the radiation resistance of the aerial. In practice the carbon transmitter is usually placed between the exciting source and ground, as shown in Figs. 12, 14, for the purpose of preventing possible shocks. heterodyne receiver
    A carbon transmitter may also be placed in the field of the high frequency alternator. Still another method is to use the armature winding differentially, with a second field, to shift the position of the field. Many other forms of transmitting devices for varying the natural period of the sending aerial circuit through the action of a transmitter upon a spark gap, etc., were experimented upon, but were laid aside on account of the objectionable noises which they produced in the receiving circuits.
    Receiving apparatus at the time Professor Fessenden began his work was in an unsatisfactory state, all known forms of receiver being of the "imperfect contact" type. These were not considered satisfactory, as it is well established that a receiver adapted to reproduce speech must be constantly operative. Moreover the known types were all voltage operated devices, and the thing required was recognized to be a current operated receiver. Forms of current operated receivers were devised, to the number of more than one hundred. In all these the fundamental principle is that all constants are electrically good contacts, and the devices are capable of being operated by electromagnetic waves. [U. S. Patent No. 706,736] They are broadly distinguished from devices depending for their operation upon the varying of contact resistance, as in the "coherer" types of receiver. Amongst these types of receiver which have become known may be mentioned the hot wire barreter, the liquid barreter, the eddy current receiver, the mircobaric receiver, the repulsive disk, etc. Of these the most satisfactory for telephone work was found to be the liquid barreter. The type of instrument consists of a small vessel containing a liquid in which is immersed a diaphragm perforated with a minute hole, before which is placed a fine point connected with the antenna. Under the action of the electromagnetic waves the stratum of liquid contained in the perforation of the diaphragm becomes heated, its resistance is varied, and if the terminals be shunted by a battery and receiver sounds will be produced corresponding to such fluctuations in resistance. The inventor of these various forms of receivers believes, however, that they are all surpassed by what he terms his "heterodyne" receiver. Although this cannot be fully described on account of the condition of patents, the following data are available:
aerial circuit
transmitter
telephone microphone
transmitting circuit
telephone relay transmission
receiving circuit

    All forms of voltage operated receivers, and most forms of current operated receivers are very inefficient. Even the liquid barreter, which is recognized as an exceptional sensitivity instrument has an efficiency of only about 1-10 of one per cent for weak signals. The magnetic receiver of the types developed for wireless telegraphy is in the same class. While a liquid barreter or magnetic receiver will give an indication between 1/100 and 1/1000 of an erg., an ordinary telephone will indicate the passage of less than 1/1,000,000 of an erg. From this it is evident that a proper method for directly using an ordinary telephone receiver would increase the efficiency enormously. This has been accomplished in the heterodyne receiver, which is a combination of the "beats method," [U. S. Patent No. 706,740] and the method of operating by continuously generated waves, [P. P. S. Patent No. 706,737] which has already been described. The beats method requires the use at the sending station of two or more antennae, so constructed and proportioned as to have different periods of oscillation--in practice a difference of about 5 per cent being preferred. At the receiving station two or more conductors are connected to separate windings and of a receiver magnet. Separate alternators are used, tuned to frequencies corresponding with the periods of the aerials to which they are respectively connected. As the device is operated waves of different periodicities are generated by the respective sending conductors, and these waves produce in the corresponding receiving conductors correspondingly varying oscillations in potential. As the oscillations persist there follows a varying difference of potential at the receiver terminals, and corresponding signals caused by the electric "beats," analogous to sound "beats" will be heard. The heterodyne receiver (Fig 8) is built up of a telephone having a fixed magnetic core formed of iron wires .001 inch in diameter, and this core is excited by a high frequency current. A small coil, with or without a core, is cemented to a thin mica diaphragm, and this coil is arranged to be excited by the oscillations produced by the received electromagnetic waves. While it is impossible to make the frequency of waves generated at the sending station exactly equal to the oscillations generated at the receiving station, it is believed that regulation sufficient for all practical purposes may be obtained by automatic means. This gives an extremely efficient form of receiver. Advantages pointed out by the inventor of this type of receiver are that it is unaffected by atmospheric disturbances, or by disturbances from nearby stations, and that it is adapted to the reception of a message on the same aerial which is being used to transmit a message to another station.
    The apparatus as set up for experiments in talking from Brant Rock to Plymouth at the time of the recent test referred to at the opening of this article was set up in conformity with the circuits shown in Figs. 12 and 15. The armature in the transmitting circuit Fig. 12 is in series with a resistance and the primary of a variable transformer. This latter piece of apparatus consists of a pair of non-inductive cores, about which are wound a number of turns of wire, the number of turns on each core being varied to suit the requirements of transformation by the simple device of rotating the core with a crank. Examples of this type of transformer are visible at the front of the right hand table in Figs 7, 10, and in Fig. 13. A similar transformer is used in the receiving circuit, Fig. 15. The receiver and battery are connected across the terminals of the barreter in the manner indicated, a simple potentiometer arrangement being used to regulate the normal voltage at the receiver terminals. Inductances, not shown in these diagrams, were inserted between the aerial and the transformer winding for the purpose of tuning. With this arrangement of apparatus speech was clearly transmitted from Brant Rock to Plymouth by some of the men present at the tests made on December 21. These tests also included experiments in transmission from a phonograph and nearly all speech, as well as music was distinctly intelligible. All tests made were apparently satisfactory. Articulation was distinct, the quality of reproduced tones good, and the efficiency of transmission was high. An expert stated that he believed efficiency to be on that day rather better than transmission through twenty-five miles of standard cable, this judgement being, of course, based on his estimation unassisted by any of the devices for comparison which are available in laboratories for transmission testing. A modification of the circuit which is shown in Fig 12 is effected by the introduction of a telephone in Figure 14. For this purpose Professor Fessenden has designed a highly ingenious type of relay, using differential windings on the cores of magnets, between the poles of which is mounted an armature attached to the electrode of a microphonic transmitter chamber. Variation in the current traversing the windings causes a shifting of the magnetic field one side or the other, producing a corresponding series of changes in the position of the plate controlling the movable transmitter electrode. This relay has shown itself to be very sensitive in practice, but improvements made within the past few weeks are expected to materially improve its efficiency. As a call a double differential relay of this type has been used to operate either a loud-speaking transmitter, a bell or a Morse writer.
    In the system shown it is necessary, as in the early Bell telephone system, to throw a switch to change from talking to listening. At the tests a method of overcoming this defect was explained. Although patent considerations prevent the publication of the method of accomplishing this at the present time, it has been in successful operation. In general, Professor has found that where no spark is used for transmitting and a carbon transmitter is used for modifying the strength of waves the speech is as distinct as over a short open wire and rather more distinct than over cables, owing to the absence of any capacity effect, and there is a total absence of extraneous noise. With the present methods of transmission, there appears to be no distortion of sounds with increase of distance, as is found in all wire lines. Although this might have anticipated, it has been experimentally demonstrated by comparison of the relative intensities of notes of different frequencies at different distances. These characteristics have led to the prediction that wireless telephony may operate over longer distances than is possible with wire lines. The difficult problem in increasing the range of transmission is at present the modulation of the large amount of energy given out by the antenna. Where an ordinary granular carbon transmitter is used about one-half ampere of current is all that can successfully be modulated and even with special transmitter buttons 2½ amperes seems to be about the limit. With multiple buttons the limit is reached at about 10 amperes. For currents larger than ten amperes a number of telephone relays may be placed in series and operated by a single transmitter. The practical limits for this method have not as yet been determined. Much depends upon the possible improvements in the efficiency of the relays. transformers and condensers
    Possible uses of wireless telephony cover a variety of important fields. At sea the wireless telephone may be used as a safeguard in foggy weather. On land it is doubtful if wireless transmission will ever supplant the local exchanges with wires. So far as the subscriber is concerned, the simplicity of present systems is an advantage which is not likely to be overcome. For trunking, however, it apparently has a field, owing to its comparatively low first cost, faculty of working multiplex for the transmission of several conversations simultaneously by methods which are now being developed, and to its lack of susceptibility to the foreign influences which produce disagreeable noises in open wire lines. Its ultimate adaptability to long distance transmission and its comparative low cost is a factor which should not be overlooked. For supplanting submarine cables the system has an obvious advantage in transmission owing to the absence of capacity effects. A practical application along this line which has been suggested is the use of the wireless system for transmitting speech across the English Channel. It is admirably adapted to the transmission of news, music, etc. as, owing to the fact that no wires are needed, simultaneous transmission to many subscribers can be effected as easily as to a few.
    Methods of automatic relaying from ordinary telephone lines to wireless transmitting lines and from a wireless receiving station to a wire line are obviously simple and have already been tested with success. On sea and on land wireless telephony has the immense advantage over telegraphy that no expert operator is required either for transmission or for sending.

Unfortunately for Fessenden and his backers, AT&T decided -- correctly -- that Fessenden's system, while revolutionary, was not yet refined enough for commercial telephone service, and so did not purchase the patents. It would not be until 1920 that the first U.S. telephone link by radio would be installed, at Catalina Island, California. And although the equipment used by the Catalina link was based on the same basic principles -- continuous-wave AM signals -- first developed by Fessenden's 1906 Brant Rock station, instead of alternator-transmitters and liquid barreter receivers, the Catalina link would employ vacuum-tube transmitters and receivers, which had been developed in the interim and were much more efficient.

Fessenden had a falling-out with his backers, and eventually left radio work. But the alternator-transmitter continued to be developed by General Electric, under the supervision of Ernest F. W. Alexanderson. Alternator-transmitters, because of their complexity, high cost, and limited range of frequencies, would never be employed by broadcasting stations, but they did make superb longwave radiotelegraph transmitters, and would be used for transoceanic service through the nineteen-forties. In fact, by 1919 the alternator-transmitter patents, with their application for international radiotelegraph service, would be considered so valuable that the question of their ownership triggered the formation of the Radio Corporation of America, because for national security reasons the U.S. government didn't want the British-owned Marconi company to gain control of the alternator-transmitter rights.