Diminishing the pressure a little, the flame is again 20 inches long, but it is on the point of roaring and shortening. Like the singing flames which were started by the voice, it stands on the brink of a precipice. The proper note pushes it over. It shortens when the whistle sounds, exactly as it did when the pressure is in excess. The action reminds one of the story of the Swiss muleteers, who are said to tie up their bells at certain places lest the tinkle should bring an avalanche down. The snow must be very delicately poised before this could occur. It probably never did occur, but our flame illustrates the principle. We bring it to the verge of falling, and the sonorous pulses precipitate what was already imminent. This is the simple philosophy of all these sensitive flames. When the flame flares, the gas in the orifice of the burner is in a state of vibration; conversely, when the gas in the orifice is thrown into vibration, the flame, if sufficiently near the flaring point, will flare. Thus the sonorous vibrations, by acting on the gas in the passage of the burner, become equivalent to an augmentation of pressure in the holder. In fact, we have here revealed to us the physical cause of flaring through excess of pressure, which, common as it is, has never been hitherto explained. The gas encounters friction in the orifice of the burner, which, when the force of transfer is sufficiently great, throws the issuing stream into the state of vibration that produces flaring. It is because the flaring is thus caused that an infinitesimal amount of energy in the form of vibrations of the proper period can produce an effect equivalent to a considerable increase of pressure. § 10. _Seat of Sensitiveness_ [Illustration: FIG. 131.] That the external vibrations act upon the gas in the orifice of the burner, and not first upon the burner itself, the tube leading to it, or the flame above it, is thus proved. A glass funnel R, Fig. 131, is attached to a tube 3 feet long and half an inch in diameter. A sensitive flame _b_ is placed at the open end T of the tube, while a small high-pitched reed is placed in the funnel at R. When the sound is converged upon the root of the flame, as in Fig. 131, the action is violent; when converged on a point half an inch above the burner, as in Fig. 132, or at half an inch below the burner, as in Fig. 133, there is no action. The glass tube may be dispensed with and the funnel alone employed, if care be taken to screen off all sound, save that which passes through the shank of the funnel.[54] § 11. _Influence of Pitch_ [Illustration: FIG. 132.] [Illustration: FIG. 133.]