In these elementary atoms he has, for the present, found the materials of which the heavens and the earth are made; but facts are being slowly gained which render it probable that these atoms are themselves structures--that they are built up of yet smaller parts, of yet simpler kinds of matter. To gather evidence for or against this supposition, the chemist has been obliged to go from the earth to the heavens, he has been obliged to form a new science, the science of spectroscopic analysis. This subject has been considered in "The Astronomers," belonging to this series of books; but the point of view from which the matter is there regarded is astronomical rather than chemical. I should like briefly to recall to the reader the fundamental facts of this branch of science. [Illustration: Fig. 4.] When a ray of light is allowed to pass through a glass prism and then fall on to a white surface, the image produced on this surface consists of a many-coloured band of light. The blue or violet part of this band is more bent away from the plane of the entering ray than the orange part, and the latter more than the red part of the band. This is roughly represented in Fig. 4, where _r_ is the ray of light passing through the prism P, and emerging as a sevenfold band of coloured lights, of which the violet, V, is most, and the red band, R, is least bent away from the plane of the ray _r_. If the surface--say a white screen--on which the many-coloured band of light, or _spectrum_, falls, is punctured by a small hole, so as to admit the passage of the violet, or blue, or orange, or red light only, and if this violet, etc., light is then passed through a second prism, no further breaking up of that light takes place. This state of matters is represented in the part of the figure towards the right hand, where the red ray, R, is shown as passing through the screen, and falling on to a second prism, P': the red ray is slightly bent out of its direct course, but is not subdivided; it falls on the second screen as a ray of red light, R'. But if a quantity of the metal sodium is vaporized in a hot non-luminous flame, and if the yellow light thus produced is passed through a prism, a spectrum is obtained consisting of a single yellow line (on a dark background), situated on that part of the screen where the orange-yellow band occurred when the ray of sunlight was split up by the action of the prism. In Fig. 5 the yellow light from a flame containing sodium is represented by the line Y. The light emitted by the glowing sodium vapour is said to be _monochromatic_. [Illustration: Fig. 5] Lastly, if the experiment is arranged so that a ray of sunlight or of light from an electric lamp passes through a layer of comparatively cool sodium vapour before reaching the prism, a spectrum is produced corresponding to the solar spectrum except that a black line appears in the position where the yellow line, characteristic of sodium, was noticed in the second experiment.