[21] The same precautions must be taken in decomposing water by potassium as have to be observed with sodium (Chapter II., Note [8]).

It must be observed that potassium decomposes carbonic anhydride and carbonic oxide when heated, the carbon being liberated and the oxygen taken up by the metal, whilst on the other hand charcoal takes up oxygen from potassium, as is seen from the preparation of potassium by heating potash with charcoal, hence the reaction K2O + C = K2 + CO is reversible and the relation is the same in this case as between hydrogen and zinc.

[22] Potassium forms alloys with sodium in all proportions. The alloys containing 1 and 3 equivalents of potassium to one equivalent of sodium are liquids, like mercury at the ordinary temperature. Joannis, by determining the amount of heat developed by these alloys in decomposing water, found the evolution for Na2K, NaK, NaK2 and NaK3 to be 44·5, 44·1, 43·8 and 44·4 thousand heat units respectively (for Na 42·6 and for K 45·4). The formation of the alloy NaK2 is therefore accompanied by the development of heat, whilst the other alloys may be regarded as solutions of potassium or sodium in this alloy. In any case a fall of the temperature of fusion is evident in this instance as in the alloys of nitre (Note [14]). The liquid alloy NaK2 is now used for filling thermometers employed for temperatures above 360°, when mercury boils.

[23] For accurate measurements and comparative researches more complicated spectroscopes are required which give a greater dispersion, and are furnished for this purpose with several prisms—for example, in Browning's spectroscope the light passes through six prisms, and then, having undergone an internal total reflection, passes through the upper portion of the same six prisms, and again by an internal total reflection passes into the ocular tube. With such a powerful dispersion the relative position of the spectral lines may be determined with accuracy. For the absolute and exact determination of the wave lengths it is particularly important that the spectroscope should be furnished with diffraction gratings. The construction of spectroscopes destined for special purposes (for example, for investigating the light of stars, or for determining the absorption spectra in microscopic preparations, &c.) is exceedingly varied. Details of the subject must be looked for in works on physics and on spectrum analysis. Among the latter the best known for their completeness and merit are those of Roscoe, Kayser, Vogel, and Lecoq de Boisbaudran.

[24] The arrangement of all the parts of the apparatus so as to give the clearest possible vision and accuracy of observation must evidently precede every kind of spectroscopic determination. Details concerning the practical use of the spectroscope must be looked for in special works on the subject. In this treatise the reader is supposed to have a certain knowledge of the physical data respecting the refraction of light, and its dispersion and diffraction, and the theory of light, which allows of the determination of the length of the waves of light in absolute measure on the basis of observations with diffraction gratings, the distance between whose divisions may be easily measured in fractions of a millimetre; by such means it is possible to determine the wave-length of any given ray of light.

[25] In order to give an idea of the size of the scale, we may observe that the ordinary spectrum extends from the zero of the scale (where the red portion is situated) to the 170th division (where the end of the visible violet portion of the spectrum is situated), and that the Fraunhofer line A (the extreme prominent line in the red) corresponds with the 17th division of the scale; the Fraunhofer line F (at the beginning of the blue, near the green colour) is situated on the 90th division, and the line G, which is clearly seen in the beginning of the violet portion of the spectrum, corresponds with the 127th division of the scale.

[26] The two most distinct lines of D, or of sodium, have wave-lengths of 589·5 and 588·9 millionths of a millimeter, besides which fainter and fainter lines are seen whose wave-lengths in millionths of a millimeter are 588·7 and 588·1, 616·0 and 615·4, 515·5 and 515·2, 498·3 and 498·2, &c., according to Liveing and Dewar.

[26 bis] In the ordinary spectroscopes which are usually employed in chemical research, one yellow band, which does not split up into thinner lines, is seen instead of the system of sodium lines, owing to the small dispersive power of the prism and the width of the slit of the object tube.

[27] The most accurate investigations made in this respect are carried on with spectra obtained by diffraction, because in this case the position of the dark and bright lines does not depend on the index of refraction of the material of the prism, nor on the dispersive power of the apparatus. The best—that is, the most general and accurate—method of expressing the results of such determinations consists in determining the lengths of the waves corresponding to the rays of a definite index of refraction. (Sometimes instead of this the fraction of 1 divided by the square of the wave-length is given.) We will express this wave-length in millionth parts of a millimetre (the ten-millionth parts are already doubtful, and fall within the limits of error). In order to illustrate the relation between the wave-lengths and the positions of the lines of the spectrum, we will cite the wave-lengths corresponding with the chief Fraunhofer lines and colours of the spectrum.

Fraunhofer line ABC D E b FG H
Wave-length 761·0687·5656·6 589·5–588·9 527·3 518·7 486·5431·0 397·2
Colour red orange yellow green blue violet