I have formerly stated that Helmholtz was enabled to produce vowel sounds artificially by combining musical tones of different pitches and intensities. His apparatus is shown in fig. 18. Tuning-forks of different pitch are placed between the poles of electro-magnets (_a^1_, _a^2_, &c.), and are kept in continuous vibration by the action of an intermittent current from the fork _b_. Resonators 1, 2, 3, &c. are arranged so as to reinforce the sounds, in a greater or less degree, according as the exterior orifices are enlarged or contracted. Thus it will be seen that upon Helmholtz’s plan the tuning-forks themselves produce tones of uniform intensity, the loudness being varied by an external reinforcement; but it struck me that the same results would be obtained, and in a much more perfect manner, by causing the tuning-forks themselves to vibrate with different degrees of amplitude. I therefore devised the apparatus shown in fig. 19, which was my first form of articulating telephone. In this figure a harp of steel rods is employed attached to the poles of a permanent magnet N.S. When any one of the rods is thrown into vibration an undulatory current is produced in the coils of the electro-magnet E, and the electro-magnet E´ attracts the rods of the harp H´ with a varying force, throwing into vibration that rod which is in unison with that vibrated at the other end of the circuit. Not only so, but the amplitude of vibration in the one will determine the amplitude of vibration in the other, for the intensity of the induced current is determined by the amplitude of the inducing vibration, and the amplitude of the vibration at the receiving end depends upon the intensity of the attractive impulses. When we sing into a piano, certain of the strings of the instrument are set in vibration sympathetically by the action of the voice with different degrees of amplitude, and a sound, which is an approximation to the vowel uttered, is produced from the piano. Theory shows, that, had the piano a very much larger number of strings to the octave, the vowel sounds would be perfectly reproduced. My idea of the action of the apparatus, shown in fig. 19, was this: Utter a sound in the neighbourhood of the harp H, and certain of the rods would be thrown into vibration with different amplitudes. At the other end of the circuit the corresponding rods of the harp H´ would vibrate with their proper relations of force, and the _timbre_ of the sound would be reproduced. The expense of constructing such an apparatus as that shown in fig. 19 deterred me from making the attempt, and I sought to simplify the apparatus before venturing to have it made. [Illustration: Fig. 19.] [Illustration: Fig. 20.]