« PreviousContinue »
bardment of the dust particles against the surfaces of these larger fragments.
A large quantity of fragments of cast-iron, steel, ebonite, sealing-wax, bone, &c., were tried as well as glass, but the attempt to obtain a luminous effect by the above method proved fruitless.
If metallic dust should produce a luminous effect when made to bombard the surface of glass, and if different effects should be produced by lead glass and German glass, the theory that solid particles emitted from the kathode would give rise to luminescence of the glass would be justified. The above experiments, however, do not seem to throw much light on Mr. Crookes' theory of luminescence due to the emitted particles of gas.
In the earlier stage of the investigation hypothesis (3) appeared plausible enough. Any experiments, however, that were made in connexion with it did not lead to results of an interesting character.
Whatever may be the real value of these experiments, it cannot be doubted that the inquiry into the cause of such a phenomenon as that noticed by Beccaria must prove of the greatest interest: firstly, on account of the obscurity enveloping the subject, and, secondly, on account of the interest attached to the artificial production (though on a small scale) of meteoric phenomena in the laboratory.
VIII. The Luminosity of Gases.
PART III.-Experiments on the Flame-Spectra of Salts of
N his well-known memoir on the spectra of coloured flames
to the fact that different spectral effects are obtainable from different parts of the flame of a Bunsen-burner. From his observations Gouy concludes that the inner cone of a Bunsenflame must be hotter than the outer one, and he remarks that, though this is in some respects in accord with theory, the mechanism of the phenomenon presents serious difficulties and demands fresh researches.
The mechanism alluded to by Gouy has, I think, been explained by recent investigations on the "Structure and * Communicated by the Author,
Chemistry of Flame" (Smithells and Ingle, Journ. Chem. Soc. lxi. p. 204, 1892), from which it is clear that the flame of a Bunsen-burner must be regarded as consisting of two distinct cones of combustion. In the inner one a partial combustion of gas takes place at the expense of the oxygen of the air admitted through the air-valves: in the outer cone, such of these products of partial combustion (chiefly carbon monoxide and hydrogen) as are capable of further oxidation are burnt in the external air; but, being mixed with a large quantity of fully oxidized products and with all the nitrogen passing through the inner cone, the flame produced is of a much lower average temperature than that of the inner cone. The apparatus used in the experiments to which allusion. has just been made secures a wide separation of the cones
of the flame of a Bunsen-burner, and so affords a simple means of ascertaining the different spectral effects obtainable in different parts of the flame.
The arrangement used in the experiments to be described is shown in the accompanying figure. Air under pressure is led to a Gouy sprayer (pulverisateur) at a and issues at c, carrying with it a fine spray or dust of the salt solution contained in b. The gas-supply by means of a tap at d can be diverted to any desired degree through a " saturator" e, which contains a roll of filter-paper or some asbestos moistened with any volatile acid or other liquid whose vapour it may be desired to add to the flame. The tube e is surrounded by a jacket through which warm water or steam may be passed, to aid in volatilizing the liquid in the saturator. The air and gas pass by a T-tube into the cone separator, which consists of two coaxial tubes connected by an india-rubber collar g and maintained symmetrical by the brass guide k. By adjusting the supply of air and gas, the flame may be so arranged that the inner cone rests at h and the outer one at i.
Flame-Spectra of Copper Compounds.
In a previous part of this paper I have commented on the difficulties of tracing the chemical processes which accompany the production of flame-spectra. The difficulties are specially great with salts of the alkalis, owing to the stability of their compounds at high temperatures. The case of chloride of copper, an easily altered substance, seemed likely to repay study, and I now give an account of the results which have been obtained.
The spectrum produced by the introduction of cupric chloride into a non-luminous flame was mapped by Bunsen and Kirchhoff in their original memoir.
In 1862 A. Mitscherlich, in a paper (Pogg. Ann. cxvi. p. 499) which raised the question as to the distinctive character of the spectra of compounds, described and mapped the spectra corresponding to what he called copper, copper chloride, and copper iodide. The copper spectrum was obtained by using a solution of copper oxide in acetic acid; the copper-chloride spectrum by supplying a mixed solution of copper chloride and ammonium chloride by means of a Mitscherlich-tube to a Bunsen-burner. Mitscherlich also noticed certain variations in the copper-chloride spectrum according to the quantity of hydrochloric acid present, and thought it probable that two spectra were obtainable, one of cuprous and one of cupric chloride.
In a later paper (Pogg. Ann. cxxi. p. 459, 1864) Mitscherlich described the spectra of copper compounds in greater detail, and mapped carefully spectra which he attributed
respectively to cuprous chloride, cuprous or cupric oxide, and metallic copper. The spectrum previously attributed by him to metallic copper was now ascribed to one of the oxides of copper, and the chloride spectrum, though not particularized in the text, is indicated on the map as being due to cuprous chloride only.
In 1865 Diacon published (Ann. Chim. Phys.  vi. p. 1, 1865) a paper on "The Influence of the Electronegative Elements on the Spectra of Metals," in which he maps and describes the spectrum of cupric chloride as produced in a coal-gas flame. He mentions that the spectrum is due to the superposition of the spectra of oxide and chloride.
Lecoq de Boisbaudran, in his Spectres Lumineux (1874), gives a map of the spectrum obtained from copper chloride when the salt is introduced into a Bunsen-flame. In describing this map Lecoq de Boisbaudran distinguishes four cases in the use of copper chloride for the production of spectra, according to the quantity of salt employed and the length of time it is held in the flame. The particular case mapped is apparently given by the author as the one in which the normal spectrum of cupric chloride is obtained.
When in using the separator the spray of a dilute solution of cupric chloride is supplied with the air, the outer cone assumes a bright green colour; the lower cone is unaffected so far as the eye can judge (as soon as all air has been removed by diffusion from the space between the tubes), and nothing but the "candle" spectrum is seen in the spectroscope. If now a piece of asbestos soaked in hydrochloric acid be introduced into the upper cone, the green colour is replaced by a vivid blue. The same effect is obtained by introducing hydrochloric-acid gas with the gas-supply by means of the saturator. If the solution of copper chloride be replaced by one of copper sulphate or nitrate, similar effects are obtained. When these flames are examined by means of the spectroscope, the sulphate and nitrate of copper are seen to produce exactly the same result. The dilute solution of CuCl alone gives a spectrum containing the same lines and bands as the nitrate and sulphate with the addition of other faint bands. When hydrochloric acid is supplied, the blue flame obtained gives a spectrum in which the lines and bands common to the sulphate and nitrate are very faint, whilst the additional lines just referred to as pertaining to the CuCl2 flame are greatly intensified, and new lines also make their appearance. Using cold hydrochloric acid, it is impossible to entirely quench those lines which are found in the simple sulphate and nitrate spectra. But if the hydrochloric-acid
saturator be jacketed by steam, or, more simply, if chloroform be used in the saturator, these lines can be entirely eliminated. We have, then, to deal with two distinct spectra-one given by copper salts in the entire absence of hydrochloric acid; the other given when a large quantity of hydrochloric acid or chloroform vapour is supplied to the flame. In the intermediate case when there is some hydrochloric acid—but not an excess—a mixed spectrum is obtained. It is obvious that one of these spectra corresponds to the oxide of copper and the other to the chloride.
These spectra have been carefully mapped and compared with the maps of previous observers, with the following results (see Plate IV.).
1. The chloride spectrum corresponds almost exactly to that given by Mitscherlich. He failed to map three bands at the most refrangible end and to split up into three distinct bands a region between X 460 and 480.
2. The oxide spectrum bears a general resemblance to that given by Mitscherlich, but with the exception of three bands he did not map in any detail the vague bands of which this spectrum is largely composed. One band at the least refrangible end is entirely omitted by Mitscherlich.
3. Lecoq de Boisbaudran's map comprises all the chloride lines and bands, with the exception of two on the extreme violet which he mentions but does not map, and all the bands of the oxide except those which it is impossible to discriminate in the presence of a strong chloride spectrum.
4. Diacon's map contains all the chloride lines and bands, but the oxide bands are only vaguely marked.
In addition to the appearances above described, there is one other noticeable when using copper salts in the separator with hydrochloric acid. When the flame of the outer cone is turned blue by hydrochloric acid, a dull ruddy fringe is seen to surround it. This becomes more developed as the quantity of hydrochloric acid is increased. It gives a faint luminous spectrum, chiefly in the red. As this red fringe is never obtained except in the presence of a large quantity of hydrochloric acid or chloroform and a copper salt, it seems natural to ascribe it to a compound of copper and chlorine.
It seemed desirable to inquire further into the chemical changes involved in the production of the different spectral
* The readings of the instrument employed (a two-prism Steinheil spectroscope) have been plotted to the scale which while giving
readings referred to a natural standard presents the spectrum very much as seen with a glass-prism spectroscope.