The fixed stars, especially those near the horizon, shine with an unsteady light, sometimes changing color as they twinkle. I have often watched at night, upon the plateaux of the Alps, the alternate flash of ruby and emerald in the lower and larger stars. If you place a piece of looking-glass so that you can see in it the image of such a star, on tilting the glass quickly to and fro, the line of light obtained will not be continuous, but will form a string of colored beads of extreme beauty. The same effect is obtained when an opera-glass is pointed to the star and shaken. This experiment shows that in the act of twinkling the light of the star is quenched at intervals; the dark spaces between the bright beads corresponding to the periods of extinction. Now, our singing flame is a _twinkling_ flame. When it begins to sing you observe a certain quivering motion which may be analyzed with a looking-glass, or an opera-glass, as in the case of the star.[50] I can now see the image of this flame in a small looking-glass. On tilting the glass, so as to cause the image to form a circle of light, the luminous band is not seen to be continuous, as it would be if the flame were perfectly steady; it is resolved into a beautiful chain of flames, Fig. 116. § 3. _Experimental Analysis of Musical Flame_ [Illustration: FIG. 117.] With a larger, brighter, and less rapidly-vibrating flame, you may all see this intermittent action. Over this gas-flame, _f_, Fig. 117, is placed a glass tube, A B, 6 feet long and 2 inches in diameter. The back of the tube is blackened, so as to prevent the light of the flame from falling directly upon the screen, which it is now desirable to have as dark as possible. In front of the tube is placed a concave mirror, M, which forms upon the screen an enlarged image of the flame. I turn the mirror with my hand and cause the image to pass over the screen. Were the flame silent and steady, we should obtain a _continuous_ band of light; but it quivers, and emits at the same time a deep and powerful note. On twirling the mirror, therefore, we obtain, instead of a continuous band, a luminous chain of images. By fast turning, these images are drawn more widely apart; by slow turning, they are caused to close up, the chain of flames passing through the most beautiful variations. Clasping the lower end, B, of the tube with my hand, I so impede the air as to stop the flame’s vibration; a continuous band is the consequence. Observe the suddenness with which this band breaks up into a rippling line of images the moment my hand is removed and the current of air is permitted to pass over the flame. § 4. _Rate of Vibration of Flame: Toepler’s Experiment_