Scientific American, September 19, 1908, pages 185-186.

Thermo-Electric Detector     The longest distance wireless telephone tests yet made on this side of the Atlantic have just been completed between Newark, N. J., and Philadelphia, Pa., a distance of eighty-one miles, as wireless waves travel.
    The system by which this has been accomplished is due to A. Frederick Collins, a pioneer in the wireless telephone field. The first of his series of tests took place between his laboratory in Newark, where he has a high-power sending station, and the Singer Building in New York city, about twelve miles away, on the night of July 9, when spoken words were clearly and loudly transmitted across the intervening space. The following day the distance was increased to thirty-five miles, when the receiving station was located at Mr. Collins's country home at Congers, N. Y., and then, amplifying the power of the sending station and bringing the instruments into sharp resonance, the Newark-Philadelphia tests were made the following Tuesday at midnight, from the top of the Land Title Building.
    This system of wireless telephony is the culmination of work begun by the inventor in 1899, and in its modified and present form it consists of an apparatus for generating continuous oscillations and an instrument for reconverting the received oscillations into audible, articulate speech. For the overland tests the initial energy employed was a direct current at 500 volts furnished by the Newark Public Service Corporation. This was increased to 5,000 volts by a direct-connected motor-generator set, the dynamo of which was especially designed by Mr. Collins to stand high-potential strains.
Wiring diagram
Collins in Laboratory
Phone Receiver
    The latter current was used to energize a self-regulating arc lamp having revolving electrodes instead of the usual induction coil employed in spark telegraphy. A blow-out magnet was adjusted at right angles to the oscillation arc, and one of the ends of the magnet was placed in series with the positive wire, and the other coil in circuit with the negative wire. This magnet fixes the arc in the best position; besides the coils serve to choke back the oscillations from the high-tension generator. Across the 500-volt direct-current circuit, the terminals of a small transformer coil are shunted, but a condenser is interposed to check the high-voltage direct current from flowing through it. The primary of the transformer is connected in series with a source of current developing 25 volts direct current and a telephone transmitter.
    From the opposite sides of the arc the oscillation circuit leads off, and is completed by a battery of glass plate condensers on either side of the tuning induction coil. The choking effect of the induction coil causes the potential difference of the oscillations to be greatest on either side of the coil. Hence the aerial and ground wires are placed on opposite sides of the coil at the point where resonance is a maximum. An auto-transformer in the aerial serves further to step-up the potential to 100,000 volts or more of the oscillations surging through it. Not the least important, though a subsidiary piece of apparatus, is the resonance tube devised by Mr. Collins for the instant visual determination of the proper values of induction and capacity of the closed circuit, as well as when this latter circuit is in tune with the aerial wire system. The device consists of an exhausted glass tube 13 inches in length and 1 3/4 inches in diameter. Sealed in the ends are platinum wires 1/16 inch in diameter, and these extend longitudinally through the center of the tube until the ends almost touch each other. The outside terminals are connected in shunt with the induction coil. Now, when the first feeble oscillations begin to surge in the closed circuit, one or the other will glow, or both of the free ends of the inclosed wires will glow, depending on the oscillatory nature of the current. As the current strength of the oscillations increases, the glow-light extends farther and farther toward the ends of the tube, but always keeping close to the oppositely-disposed wires.
    The length of the glow on the wires is proportional to the current strength, and thus the tube may also be used as a measuring apparatus instead of the milammeter usually employed. The characteristics of the oscillations can also be easily observed; for if they are positive the light will appear almost wholly on the end of one of the wires, and if the current is reversed, on the opposite end; while if the current is oscillating with equal electro-motive forces, the light will have the same degree of intensity on both wires. By means of a revolving mirror the oscillations may be segregated, and it is then easy to see whether they are periodic or continuous, and if the latter, to analyze the wave form of the spoken words.
    The receptor comprises a thermo-electric detector of Mr. Collins's invention, the fuller details of which it is inadvisable to give out at the present time. It may be said, however, that the principles upon which it is based are entirely different from the numerous other detectors that have made their appearance since the original form of the Branly coherer. Roughly, the detector in question consists of two exceedingly fine wires of different metals crossed at right angles and forming a couple, somewhat on the order of a Boys radio-micrometer, the conduction losses, however, exceeding the radiation losses. Under the junction of these wires is placed a resistance wire, which is heated by the currents surging in the aerial wire system. The detector is sensitive to oscillations of 1/5000 of an erg, and is especially well adapted to the reception of articulate speech. A variable electrolytic resistance is used to modify the current, while the tuning inductance and condensers are very much the same as in other wireless systems.
    The highest degree of tuning is obtained by means of a thermo-galvanometer. This instrument comprises a single loop of silver wire suspended between the poles of a permanent magnet. The lower ends of the loop are connected with a bismuth-antimony thermocouple, which is heated by a fine filament of high specific resistance, through which filament the oscillating current passes, very much as in the detector just described. One end of the heater is connected with the frame of the instrument, in order to avoid electrostatic stress. The heat generated by the passage of the oscillations through the resistance falls on the thermo-junction, and the resulting electromotive force applied to the ends of the silver loop causes it to turn in the magnet field.
    In the Newark-New York tests the aerial wire at the sending station had a length of 250 feet, and was formed of four radiating phosphor-bronze wires, making a total of 1,000 feet of wire. At the Singer Building the receiving station was located on the twenty-fifth floor, and from the receptor an aluminium wire passed through a window and followed the perpendicular wall to the forty-first story, where it passed through a porcelain bushing, which was suspended at the end of an arm projecting five feet from the cornice of the building. The upper end of the antenna was likewise swung away from the top of the flagstaff, 612 feet above the street level, by means of a highly insulated arm, and the wire was thus kept free from the building.
    The receptor was grounded to the water pipe. In the Newark-Congers experiments the aerial consisted of 1,000 feet of aluminium wire held in the air by three kites which were connected in tandem. The same aerial was elevated from the top of the Land Title Building in Philadelphia. Hence in every case there was practically a clear visual line between the sending and the receiving instruments.