All sounds are not equally effective on the flame; waves of special periods are required to produce the maximum effect. The effectual periods are those which synchronize with the waves produced by the friction of the gas itself against the sides of its orifice. With some of these flames a low deep whistle is more effective than a shrill one. With others the exciting tremors must be very rapid, and the sound consequently shrill. Not one of these four tuning-forks, which vibrate 256 times, 320 times, 384 times, and 512 times respectively in a second, has any effect upon the flame from our iron nipple. But, besides their fundamental tones, these forks, as you know, can be caused to yield a series of overtones of very high pitch. The vibrations of this series are 1,600, 2,000, 2,400, and 3,200 per second, respectively. The flame jumps in response to each of these sounds; the response to that of the highest pitch being the most prompt and energetic of all. To the tap of a hammer upon a board the flame responds; but to the tap of the same hammer upon an anvil the response is much more brisk and animated. The reason is, that the clang of the anvil is rich in the higher tones to which the flame is most sensitive. The powerful tone obtained when our inverted bell is reinforced by its resonant tube has no power over this flame. But when a halfpenny is brought into contact with the vibrating surface the flame instantly shortens, flutters, and roars. I send an assistant with a smaller bell, worked by clockwork, to the most distant part of the gallery. He there detaches the hammer; the strokes follow each other in rhythmic succession, and at every stroke the flame falls from a height of 20 to a height of 8 inches, roaring as it falls. [Illustration: FIG. 134.] [Illustration: FIG. 135.] The rapidity with which sound is propagated through air is well illustrated by these experiments. There is no sensible interval between the stroke of the bell and the ducking of the flame. When the sound acting on the flame is of very short duration a curious and instructive effect is observed. The sides of the flame half-way down, and lower, are seen suddenly fringed by luminous tongues, the central flame remaining apparently undisturbed in both height and thickness. The flame in its normal state is shown in Fig. 134, and with its fringes in Fig. 135. The effect is due to the retention of the impression upon the retina. The flame actually falls as low as the fringes, but its recovery is so quick that to the eye it does not appear to shorten at all.[55] § 12. _The Vowel-flame_