tube B; is decomposed by the prism P, after its rays have emerged parallel by their passage through the lenses d and c, and is seen by the eye of the observer in the form of a spectrum at xx. At the anterior end m of the tube C is a microme

thrown upon the prism, and thence by reflection passes into the tube A and is seen upon the spectrum itself, thus serving to identify the position of any given line. The method of using this scale may be seen by reference to fig. 12, where 900, 1,000, etc., are the images of the figures on the scale seen in connection with the spectrum. This scale is entirely arbitrary. The position of the lines would vary with the numbers of prisms used in the dispersion of the light, for we may use a number of prisms instead of the single one shown in figs. 1 and 2. They also vary with the material of which the prism is made. For example, the relative lengths of spectra obtained under like conditions from similar prisms of flint glass, crown glass, and water, are approximately 3, 1, and 1. To avoid this ambiguity, the positions of rays in the spectrum have been recently determined and expressed by their "wave lengths." As yet, however, the former method, from its simplicity, seems to be preferred in practical work.

Let us now ignite a piece of common salt (sodic chloride) in the Bunsen burner, shown in fig. 2, before the slit of the instrument. Instead of seeing a continuous spectrum, the observer will now see a single yellow line (a prism of great dispersive power separates it into two lines, as seen in fig. 10 at D) represented by fig. 3, and which has its length in the same direction

as the parallel sides of the prism. This line corresponds with Frauenhofer's line D. No other light will appear unless

Fig. 3.

emanating from some foreign source. The salt, converted into a luminous gas by the heat of the Bunsen burner, sends forth vibrations of nearly uniform wave length, and hence of nearly equal refrangibility; and thus, instead of having all the colors of the continuous spectrum, we have but a single color, and this only occupying a small part of the zone of yellow. If, instead of igniting common salt, we use a piece of potassic chloride, we will have two red lines and one violet. Calcic chloride would give red, orange, yellow, green, and violet. lines. In short, the salts, or compounds of each element, their own color. These peculiar spectra afford analytical tests of surpassing delicacy. A mixture of ignited salts of various bases gives all the lines characteristic of each of them. Their position may be noted by means of the scale. Those salts are usually best adapted for spectral examinations, which are readily converted into luminous gas, and for this reason the chlorides are generally selected.

It will be necessary to considere some of the meth

ods that we possess of producing lights of the most intense before

luminosity speaking of spectra of the next species. The best of these is the electric light, represented by fig.

4. The carbon points at L, which are at first in contact with each other, and are joined by metallic connections with the poles of a powerful galvanic battery. As soon as the electric current is thus established the carbon points are drawn slightly FOURTH SERIES, VOL. XXIII.-7

asunder, when a dazzling arc light appears between them, and of a very high temperature. If a band of parallel rays of the light emanating from this source, and passing through the slit C, be thrown upon a screen after dispersion by means of a prism, a continuous spectrum will appear. If the lower and positive carbon be hollowed out, and a small piece of sodium or potassium be placed in it, upon producing the light again a spectrum of the bright lines of these metals will appear upon the screen. Fig. 4 shows the appearance of a spectrum thus formed. Instead of the electric light, Ruhmkorff's coil is frequently employed for converting into luminous vapor those metals that are volatilized only at very high temperatures. The spectra from any source may be viewed through a telescope, as shown in fig. 2, instead of being thrown upon a screen, as seen in fig. 4; and this remark may also apply to those represented in figs. 5 and 6.

Let us now pass to a fundamental experiment, illustrated by fig. 5. Upon the left are seen the carbons of the electric light apparatus; at E, the slit through which a band of rays from it passes. In the first part of the experiment the parts represented by G, L, and S are removed. Let now the carbons be dipped in a weak solution of sodic chloride and dried, thus leaving them slightly coated with the salt, as explained in fig. 4. Upon establishing the electric current, the bright yellow sodium line will appear upon the screen Let this position be carefully marked. Soon the salt upon the carbons is completely volatilized, the yellow line disappears from the screen, and an ordinary continuous spectrum appears. Now place directly before the slit, E, any source of light, as a Bunsen burner, G; in front of this place a large screen, S, perforated by an opening as represented, so as to prevent the light from from sending rays to all parts of the screen on the extreme right; and now, with the intense electric light burning, introduce a piece of sodium, 7, into the flame of the Bunsen burner. We might, perhaps, expect the same yellow line that we obtained in the former experiment;

at m.

but, on the contrary, we now have an intensely black line in the precise location where our bright yellow previously appeared. If we remove the sodium the black line instantly disappears; replace it, again it appears. It is thus proved experimentally that sodium destroys rays of light of the same refrangibility as its own rays. In a manner entirely analogous, this principle may be demonstrated in the case of substances generally, and hence the principle deduced from the experiment with sodium may be stated as a general law. Lines of this nature are termed absorption lines. They occupy the same locations as their corresponding bright lines occupy in spectra of the second class, and constitute the third class of spectra. To produce this class distinctly, it is necessary that the absorbing flame possess less light than that whose rays pass through it; and the greater the difference the more striking the results. If, instead of the electric light in this experiment, the less intense calcium light had been employed, the blackness of the absorption line would have been less intense. But why do absorption lines thus appear? Let us take the case of sodium as an illustrative answer. The electric light sent out rays of all wave lengths and refrangibilities from ultra red to ultra violet. The sodium light, 7, emits rays of only slightly varying wave lengths. These waves strike down or destroy to a great extent those rays of the same wave lengths which come from the electric light. The other rays from this latter source pass along unharmed and fall upon the screen. It is clear, then, that there must be more light upon every other part of the spectrum than in the zone at D, where a large part of the light has been destroyed; m D must, therefore, be comparatively dark, and this relative darkness is greater the more light we give to the other portions of the spectrum, which condition is best fulfilled when the more remote light is as intense as possible. If the sodium light were the more intense of the two, a part of its rays would be destroyed by the comparatively faint rays of like refrangibility passing through it; but, owing to the faintness of the remote light, and the consequent faintness of its spectrum, there would still be a sufficient number of unharmed sodium rays to produce a faintly bright line.

We must now become acquainted with a method of comparing spectra. In fig. 6 we have one source of light, F,

placed directly before the upper part of the slit (s) of the tube. If a salt be converted into luminous gas in this flame, its Fig. 6. spectrum of bright lines

P

will appear at u on the screen. Let us now volatilize a salt of known composition. in the flame E placed directly in front of the prism d, which stands. before the lower half of

the slit. The rays from E are thus bent, and pass through the tube, forming a spectrum of bright lines at o, directly above the former spectrum. If we find the bright lines of these two spectra exactly coinciding, as represented upon the screen at the left, we know that the bases of the two substances are identical. We may thus compare a spectrum of absorption lines with one of bright lines by substituting such an arrangement as that shown in fig. 5 for one of the sources of light in fig. 6.

We could thus, as also by the micrometer scale, determine with regard to the coincidence of the two classes of lines. The light of the sun, stars, nebulæ, etc., might thus be examined. Before speaking of these bodies, however, let us repeat here that the spectra shown in figs. 4, 5, and 6 are usually thrown into a tube of the spectroscope after dispersion by the prism, and observed by means of a lens, as shown in figs. 1 and 2.

By far the most interesting and important part of spectrum analysis is its application to the heavenly bodies. Its leading principles are simple. If we examine spectroscopically the light which they send to us and find their spectra, (1) continuous, (2) of bright lines, or (3) of dark lines, we infer from experimental data that they are respectively (1) incandescent solid or liquid bodies, or gaseous bodies subjected to great heat, (2) bodies consisting of luminous gas, or (3) incandescent bodies -solid, liquid, or gaseous-surrounded by an atmosphere of lower temperature and less luminosity. And if in the two latter cases we can establish the identity of the lines, and prove their exact coincidence with those of known terrestrial elements, we have thus found a means of ascertaining the chemical constitution of these distant worlds. It is but natural that