hydrogen-apparatus; the substance was then volatilized, and the gas kindled at the other end of the tube.

7. Or the spark was taken between poles containing the metal or compound in any gas; or between,

8. Liquid electrodes, in which the temperature is much lower than in 7.

From the beautiful series of researches carried on by these several methods, he concludes "that every compound of the first order which is not decomposed, and is heated to a temperature adequate for the production of light, exhibits a spectrum peculiar to this compound, and independent of other circumstances."

§ 9. Light Thrown by the Long and Short Lines.

Some experiments of my own, commuuicated to the Royal Society in 1873, taken in conjunction with my determination of the long and short lines of metallic vapours, and the consequent simplification of the spectra by the reducton of pressure, set this question at rest, I think and in the direction indicated by Mitscherlich, Clifton, Roscoe, and Diacon; while much light was thrown upon all the prior observations, as a consequence of which they are brought much more into harmony than at first appeared.

These experiments have been given at length in Chapter VI. I will here summarize them as a step to the work now to be discussed :

First. I observed that whether the spectra of iodides, bromides, &c., be observed in the flame of weak spark,

in air, the spectrum is in the main the same, as maintained by Kirchhoff and Bunsen; but that this is not the spectrum of the metal is established by the facts, that with a low temperature only the longest lines of the metals are present, showing that only a small quantity of the simple metal is present as a result of partial dissociation, and that by increasing the temperature, and consequently the amount of dissociation, the other lines of the metals appear in the order of their length with each rise of temperature.

Secondly. I convinced myself that this is the spectrum of the oxide, because in air, after the first application of heat, the spectra and metallic lines are in the main the same, while in hydrogen the spectra are different for each compound, and true metallic lines are represented rccording to the volatility of the compound, only the very longest lines being visible in the spectrum of the least volatile compound.

In proof of this statement I append a drawing (Fig. 46), representing the spectra of the chloride, bromide, and iodide of strontium. In order to avoid the introduction of the oxide spectrum, and so to secure the differentiation of the three spectra if possible, they were observed in hydrogen, which gas had been carefully treated in such a manner as to secure as far as possible the exclusion of any trace of oxygen. It will be seen at a glance that the spectra differ not only from the spectrum given by the metal, or by its salts in air at a high temperature, but considerably amongst themselves. In the experiments care was taken to keep the

V. In some cases where the percentage of a constistuent was so small that none of its lines were visible, there yet seemed to be an effect produced upon the lines of the other constituent as compared with those of the spectrum of the same vapour of the opposite pole.

VI. Changes in the relative lengths of a pair of lines belonging severally to the constituents of the alloy.

These conclusions were derived from observations of the alloys, which I had made in the manner indicated in my second Paper; and I saw that it would be important to observe series in which the change of percentage composition between the specimens was not so great, and of which actual assays had been made.

I therefore begged Mr. C. Freemantle, the Deputy Master of the Mint, to allow me the use of specimens. of the gold-copper and silver-copper alloys prepared for the coinage, as in them I had exactly what the research required—namely, ranges with small variations and of undoubted accuracy. That gentleman, at once, with the greatest promptitude and courtesy, acceded to my request.

§ 5. Experiments with Large Variations.

Before, however, I proceed to consider the cases of the Mint alloys, it will be well to briefly notice some experiments with alloys in which the variations in the proportions of constituents were greater.

As an instance of a moderately small difference, an

spark temperature as low as possible; and of course no jar was used, its presence in the circuit being instantly felt by the introduction of large numbers of metallic lines and the disappearance of the banded spectrum.

§ 10. General Statement.

These and other facts which I have observed can be included in a general statement such as the following.

I. A compound body, such as we have been considering, has as definite a spectrum as a simple one; but while the spectrum of the simple metal consists of lines, the number and thickness of some of which increase with molecular approach, the spectrum of the compound consists in the main of channelled spaces and bands, which increase in like manner. In short, the molecules of a simple body and a compound one are affected in the same manner by their approach or recess, in so far as their spectra are concerned; in other words, both spectra have their long and short lines, the lines in the spectrum of the element being represented by bands or channelled lines in the spectrum of the compound; and in each case the greatest simplicity of the spectrum depends upon the greatest separation of molecules, and the greatest complexity (a continuous spectrum) upon their nearest approach.

2. The heat required to act upon a compound, so as to render its spectrum visible, dissociates the compound according to its volatility; the number of true metallic lines which thus appear is a measure of the dissocia

tion; and as the metal lines increase in number, the compound bands thin out.*

11. The Problem of the Dissociation of the so-called Elements.

I have now shown historically how we have been led to the conclusion that binary compounds have spectra of their own, and how this idea has been, if not established, at all events strengthened by considerations having for a basis the observations of what I have termed long and short lines.

I now proceed to show how absolutely similar observations and similar reasoning may, in the future, succeed in establishing the compound nature of, to say the least, some of the chemical elements themselves.

* The above statement is confirmed by the following experiments. A bead of strontic chloride was interposed between two aluminium electrodes; the induced current, without a jar, was then passed. The red-band spectrum of the oxide was very intense, and the only metallic line of any strength was 4607'5. The wire and bead soon became red-hot, and the latter evaporated, the spectrum disappearing. A jar of 186 square centims. coated surface was then introduced into the secondary current. The metallic lines appeared all along the spectrum, the "structure" (oxide spectrum) became fainter, and its intervals wider; the bead soon became red-hot. A jar of 467 centims. gave lines only and no structure, and one of 2214 centims. the same result, the bead remaining cold.

On using the bead as the electrode, the results were nearly the same; but the heating-effect continued when somewhat larger jars were used than the one with 362 centims., which did not show this effect with the former arrangement. When the slit was very narrow it was observed that several of the bands of the oxide spectrum broke up into masses of fine lines, exactly like those of the iodine vapour absorption-spectrum; and this remarkable resemblance was rendered still more striking by the appearance of a bead like that shown by iodine.