But leaving harmonic telegraphy and its “tuned bars,” both Gray and Bell applied themselves to the old problem of transmitting human speech. What was their very first step? They threw away their “tuned bars” and “steel springs,” and returned _to the tympanum_! Elisha Gray devised the receiver shown in Fig. 46, _G_, taken from his caveat of date February 14, 1876.[51] In that document Gray says: “My present belief is that the most effective method of providing an apparatus capable of responding to the various tones of the human voice, is a _tympanum_, drum, or diaphragm,” stretched across one end of a chamber. He adds that in the receiver there is (see Fig. 46, _G_) an electro-magnet, acting upon a diaphragm to which is attached a piece of soft iron, and which diaphragm is stretched across a vocalising chamber. Graham Bell’s receiver (the American specification of which was filed the same day as Gray’s caveat) is shown (in the form patented in Great Britain, Dec. 9, 1876) in Fig. 46 _H_, which is taken from Fig. 19 of Bell’s British patent. “The armature,” says the inventor, “is fastened loosely by one extremity to the uncovered leg, _h_, of the electro-magnet _E_, and its other extremity is attached to the centre of a stretched membrane.” The armature, in fact, was capable of vibrating like a pendulum on its pivot, but was elastically restrained by its attachment to the tympanum; the armature would therefore vibrate in perfect correspondence with any vibrations forced upon it by the electro-magnet. This instrument as also that of Gray, was admirably adapted to receive speech, for it embodied the three essential points which Reis had already discovered: viz., firstly, that the armature must be of iron, or capable of being acted upon by magnetic induction; secondly, that it must be elastically mounted; thirdly, that it should present an extended surface. Bell’s form of receiver had the advantage over Reis’s (compare p. 158), that its extended surface was a true tympanum of membrane, and not a mere broad thin plank. Being a tympanum, it therefore realised Reis’s fundamental notion of imitating the human ear more fully than even Reis’s own receiver did.