Now all incandescent solid or liquid substances, and even gases ignited under great pressure, give what is called a "continuous spectrum;" that is to say, the light derived from them is of every conceivable hue. Sorted out with the prism, its tints merge imperceptibly one into the other, uninterrupted by any dark spaces. No colours, in short, are missing. But gases and vapours rendered luminous by heat emit rays of only a few tints, which accordingly form an interrupted spectrum, usually designated as one of lines or bands. And since these rays are perfectly definite and characteristic--not being the same for any two substances--it is easy to tell what kind of matter is concerned in producing them. We may suppose that the inconceivably minute particles which by their rapid thrilling agitate the ethereal medium so as to produce light, are free to give out their peculiar tone of vibration only when floating apart from each other in gaseous form; but when crowded together into a condensed mass, the clear ring of the distinctive note is drowned, so to speak, in a universal molecular clang. Thus prismatic analysis has no power to identify individual kinds of matter, except when they present themselves as glowing vapours. A spectrum is said to be "reversed" when lines previously seen bright on a dark background appear dark on a bright background. In this form it is equally characteristic of chemical composition with the "direct" spectrum, being due to _absorption_, as the latter is to _emission_. And absorption and emission are, by Kirchhoff's law, strictly correlative. This is easily understood by the analogy of sound. For just as a tuning-fork responds to sound-waves of its own pitch, but remains indifferent to those of any other, so those particles of matter whose nature it is, when set swinging by heat, to vibrate a certain number of times in a second, thus giving rise to light of a particular shade of colour, appropriate those same vibrations, and those only, when transmitted past them,--or, phrasing it otherwise, are opaque to them, and transparent to all others. It should further be explained that the _shape_ of the bright or dark spaces in the spectrum has nothing whatever to do with the nature of the phenomena. The "lines" and "bands" so frequently spoken of are seen as such for no other reason than because the light forming them is admitted through a narrow, straight opening. Change that opening into a fine crescent or a sinuous curve, and the "lines" will at once appear as crescents or curves.