To be visible, a body must either emit light itself, must shine, or must affect in some way the light which falls upon it, must take up that light entirely or partly, absorb it, or must have a deflective effect upon it, that is, reflect or refract it. We cannot see the air as we can a flame, for it shines only exceptionally, as in a Geissler's tube. The atmosphere is extremely transparent and colorless; it cannot be seen, therefore, as a dark or colored body can, or as chlorine gas can, or vapor of bromine or iodine. Air, finally, has so small an index of refraction and so small a deflective influence upon light, that the refractive effect is commonly imperceptible altogether. A glass rod is visible in air or in water, but it is almost invisible in a mixture of benzol and bisulphuret of carbon, which has the same mean index of refraction as the glass. Powdered glass in the same mixture has a vivid coloring, because owing to the decomposition of the colors the indices are the same for only one color which traverses the mixture unimpeded, whilst the other colors undergo repeated reflexions.[111] Water is invisible in water, alcohol in alcohol. But if alcohol be mixed with water the flocculent streaks of the alcohol in the water will be seen at once and _vice versa_. And in like manner the air, too, under favorable circumstances, may be seen. Over a roof heated by the burning sun, a tremulous wavering of objects is noticeable, as there is also over red-hot stoves, radiators, and registers. In all these cases tiny flocculent masses of hot and cold air, of slightly differing refrangibility, are mingled together. In like manner the more highly refracting parts of non-homogeneous masses of glass, the so-called striæ or imperfections of the glass, are readily detectible among the less refracting parts which constitute the bulk of the same. Such glasses are unserviceable for optical purposes, and special attention has been devoted to the investigation of the methods for eliminating or avoiding these defects. The result has been the development of an extremely delicate method for detecting optical faults--the so-called method of Foucault and Toepler--which is suitable also for our present purpose. [Illustration: Fig. 49.]