In this case the tube is both generator and "modulator." In some cases these operations are separately performed by different tubes. This was true of the transmitting set used in 1915 when the engineers of the Bell Telephone System talked by radio from Arlington, near Washington, D. C., to Paris and Honolulu. I shall not draw out completely the circuit of their apparatus but I shall describe it by using little squares to represent the parts responsible for each of the several operations. First there was a vacuum tube oscillator which generated a small current of the desired frequency. Then there was a telephone transmitter which made variations in a direct-current flowing through the primary of a transformer. The e. m. f. from the secondary of this transformer and the e. m. f. from the radio-frequency oscillator were both impressed upon the grid of an audion which acted as a modulator. The output of this audion was a radio-frequency current modulated by the voice. The output was amplified by a two-stage audion amplifier and supplied through a coupling coil to the large antenna of the U. S. Navy Station at Arlington. Fig. 121 shows the system. [Illustration: Fig 121] The audion amplifiers each consisted of a number of tubes operating in parallel. When tubes are operated in parallel they are connected as shown in Fig. 122 so that the same e. m. f. is impressed on all the grids and the same plate-battery voltage on all the plates. As the grids vary in voltage there is a corresponding variation of current in the plate circuit of each tube. The total change of the current in the plate-battery circuit is, then, the sum of the changes in all the plate-filament circuits of the tubes. This scheme of connections gives a result equivalent to that of a single tube with a correspondingly larger plate and filament. [Illustration: Fig 122] Parallel connection is necessary because a single tube would be overheated in delivering to the antenna the desired amount of power. You remember that when the audion is operated as an amplifier the resistance to which it supplies current is made equal to its own internal resistance of _R_{p}_. That means that there is in the plate circuit just as much resistance inside the tube as outside. Hence there is the same amount of work done each second in forcing the current through the tube as through the antenna circuit, if that is what the tube supplies. "Work per second" is power; the plate battery is spending energy in the tube at the same rate as it is supplying it to the antenna where it is useful for radiation. [Illustration: Pl. XI.--Broadcasting Equipment, Developed by the American Telephone and Telegraph Company and the Western Electric Company.]