The distance of the whistle from the lens has then to be adjusted so as to produce on the other side a nearly parallel beam of sound. In other words, the whistle must be placed in the focus of the lens. A rule for doing this is as follows: If the balloon from which the segments of collodion were cut was nearly spherical, and had a diameter of 8 inches, then the whistle must be placed at slightly less than 8 inches from the side of the lens next to it.[23] The exact distance, however, will have to be found by trial, but it is somewhere near the point so determined. The sensitive flame should be about 4 or 5 feet away from the lens on the other side of the screen. These arrangements having been made and the whistle set in action, it will be found that the flame responds vigorously when it is placed on the _axis-line_ of the lens, but if moved a few inches to right or left of this line, it will cease to flare. This shows us that we have formed a beam of sound, and with some little care it is possible to make this a nearly parallel beam, so that when plunged in this stream of air waves the flame dips, but by removing it just outside the stream of sound it no longer flares. I have found it not difficult, when using a sound-lens 6 or 7 inches in diameter, to make a beam of sound from a whistle some 10 inches wide at about 4 feet from the lens. Supposing the sound-lens and sensitive flame so adjusted, it is then necessary for our purpose to provide a sound-prism, made in the following manner: A zinc box is made in wedge form, and the two inclined sides are cut out, and these windows are covered with thin collodion film. The box has two pipes connected with it, by means of which it can be filled with carbonic acid gas. Provided with this apparatus, it is now possible to show you a series of experiments which will leave no doubt in your minds that the external agency which creates in us the sensation of sound is a wave-motion in the air we breathe. Let me, in the first place, show you that a sound-beam can be reflected. We adjust our sensitive flame and set the whistle in action, and create, as described, by the lens, a beam of sound. At a little distance, say a couple of feet, outside the parallel beam we place the sensitive flame, and, being sheltered from the direct action of the whistle, it remains perfectly quiescent. Taking a sheet of glass in my hand, I hold it at an angle of 45° in the sound-beam, and you see the flame at once roars. The beam has been reflected on to the flame, but a very small angular movement of the glass is sufficient to reflect the sound-ray past the flame without touching it, and the flame then exhibits no agitation.