106.]

SPECTRA OF COLOURED FLAMES.

189

relation to Kirchhoff's theory of the cause of the dark lines, which requires that the position of the bright lines thus obtained, should coincide with the black lines produced by absorption when light is transmitted through these different gases. Plücker's expe

riments show distinctly that this is not the case in these gases. Dr. Robinson subsequently investigated the effect of varying the pressure upon the nature of the electric spectrum of an incandescent gas (Phil. Trans. 1862, 939): he finds that the light is the most intense under ordinary pressures, though at low pressures bright lines appear which are not visible at the ordinary pressure of the gases.

Van der Willigen (Poggendorff's Annal. 1859, cvi. 617) made the interesting remark that, by placing in succession upon a pair of wires consisting of a metal which, like platinum, possesses few special bright lines of its own, small quantities of a weak solution of chloride of calcium, barium, or strontium, or of nitrate of calcium, &c., new bright lines are produced, and these lines are characteristic of the particular metal contained in each of these several compounds.

d. Spectra of Coloured Flames.-The first person who seems to have analysed coloured flames by means of the prism, was Sir J. Herschel, who describes briefly (Edin. Phil. Trans. 1822, ix. 455), the spectra of strontic, calcic, and cupric chloride, of cupric nitrate, and of boracic acid. The same observer, in the article LIGHT, Encycl. Metrop. 1827, 438, says :-" Salts of soda give a copious and purely homogeneous yellow, of potash a beautiful pale violet" and he then gives a general statement of the results with the salts of calcium, strontium, lithium, barium, copper, and iron. He further continues,-"Of all salts the muriates [chlorides] succeed best, from their volatility. The same colours are exhibited also when any of the salts in question are put in powder into the wick of a spirit-lamp. The colours thus communicated by the different bases to flame, afford in many cases. a ready and neat way of detecting extremely minute quantities of them."

The analysis of the spectra of artificial lights was resumed by Fox Talbot (Brewster's Journal of Science, 1826, v. 77). He there describes a method of obtaining a yellow monochromatic light by the use of an ordinary spirit-lamp with a cotton wick fed with dilute alcohol holding common salt in solution. He found the same effect, whether chloride, sulphate, or carbonate of sodium was employed.

Nitrate, sulphate, chlorate, and carbonate of potassium agreed

190

SPECTRUM ANALYSIS.

[106. in giving a bluish-white tinge to the flame. By burning a mixture of nitre and sulphur, he observed a red line of low but definite refrangibility, which he regarded as characteristic of the salts of potassium, as the yellow line is of the salts of sodium. He concludes his paper with the following observation, which follows some remarks upon certain experiments of Herschel's :-" If this opinion should be correct and applicable to the other definite rays, a glance at the prismatic spectrum of a flame may show it to contain substances which it would otherwise require a laborious chemical analysis to effect."

In the Phil. Mag. for 1834 [3], iv. 115, Fox Talbot further showed how, notwithstanding the similarity in colour of the light of lithium and strontium, they can at once be distinguished by means of the prism. He says, "The strontia flame exhibits a great number of red rays, well separated from each other by dark intervals, not to mention an orange and a very definite bright blue ray. The lithia exhibits one single red ray. Hence I hesitate not to say that optical analysis can distinguish the minutest portions of these two substances from each other with as much certainty, if not more, than any other known method."

The spectra of coloured flames were further examined in 1845 by myself, and an account of these experiments was given in a paper read that year before the Chemical Section of the British Association, at Cambridge. (Phil. Mag. 1845 [3], xxvii. 81.)

In these experiments an alcohol lamp, fed with the solution of the compound the flame of which was to be examined, and a common wick supported in a small glass tube, furnished the flame. The lamp was placed opposite the vertical slit, through which diffused daylight could also be transmitted at pleasure. Fraunhofer's lines thus served as points of comparison of the different flames. The paper was illustrated by coloured lithographs of various spectra, including those of cupric chloride, boracic acid, strontic nitrate, calcic and baric chlorides, in minute detail. The green light produced by burning a solution of cupric chloride in alcohol, for instance, gives the spectrum shown in fig. 81, No. 5, and that furnished by an alcoholic solution of boracic acid is represented in No. 6. Numerous other spectra were also described, including those of the chlorides of sodiuni, manganese, and mercury, and of a large number of other metals.

(107) Spectrum Analysis.-But it is to Kirchhoff and Bunsen (Poggendorff's Annal. 1860, cx. 161) that we are indebted for reducing the prismatic observation of flame tinged by the salts of different metals to a simple and systematic method of qualitative analysis for the alkalies and alkaline earths; and they have contrived a spectroscope by which the different spectra may be conveniently examined and compared with one another.

Fig. 82 exhibits a very complete form of the instrument adapted to a single prism (Poggendorff"s Annal. 1861, cxiii. 374). It is an improvement on the

107.]

SPECTROSCOPE.

191

instrument used by Swan and by Masson; including a scale for ascertaining the position of the lines in different cases, as well as a reflecting prism, by which two spectra can be compared side by side. P represents a flint-glass prism, supported on the cast-iron tripod F, and retained in its place by the spring, c. At the end of the tube, A, nearest the prism, is a lens, placed at the distance of its focus for parallel rays from a vertical slit at the other end of the tube. The width of the slit can be regulated by means of the screw, e. One-half of this slit is covered by a small right-angled prism designed to reflect the rays proceeding from the source of light, D, down the axis of the tube, whilst the rays from the source of light, E, pass directly down the tube. By this arrangement the observer stationed at the end of the telescope B, is able to compare the spectra of both lights, which are seen one above the other, and he can at once decide whether their lines coincide or differ. a and b are screws for adjusting the axis of the telescope so as to bring any part of the slit at e into the centre of the field of vision. The telescope, as well as the tube, c, is movable in a horizontal plane, around the axis of the tripod. The tube c contains a lens at the end next to the prism,

and at the other end is a scale formed by transparent lines on an opaque ground; it is provided with a levelling screw, d. When the telescope has been properly adjusted to the examination of the spectrum, the tube c is moved until it is placed at such an angle with the telescope and the face of the prism, that when light is transmitted through the scale, the image of this scale is reflected into the telescope from the face of the prisin nearest the observer. This image is rendered perfectly distinct by pushing in the tube which holds the scale nearer to the lens in c, or withdrawing it to a greater distance, as may be required. The reflected lines of the scale can then be employed for reading off the position of the bright or dark lines of the spectrum, as both will appear simultaneously, overlapping each other, in the field of the telescope. By turning the tube c round upon the axis of the tripod, any particular line of the scale can be brought to coincidence with any desired line of the spectrum. Stray light is excluded by covering the stand, the prism, and the ends of the tubes adjoining it, with a loose black cloth. The dispersive power upon the spectrum may be much increased by using several prisms instead

192

SPECTROSCOPE.

[107.

of one in the experiments of Kirchhoff upon the solar spectrum, he used four prisms; Huggins (Phil. Trans. 1864, 139) in his observations on the spectra of the metals obtained with the electric spark employed six prisms, and Gassiot has used an apparatns with nine prisms. Much care is required in placing the prisms: the refracting edge of each prism must be truly vertical, and the position of minimum deviation for the rays to be observed must be obtained.

Mr. Browning has constructed an ingenious form of automatic spectroscope in which the light passes through a train of six prisms. The prisms are mounted on triangular supports which are linked to one another and to the observing telescope in such a manner that on moving the latter all the prisms are maintained at the angle of minimum deviation for the particular ray under observation. The action of this spectroscope will be seen by an inspection of fig. 82 a, which is a plan of the latest form of the instrument. A is the collimator with FIG. 82 a.

the slit and small reflecting prism at a, the rays transmitted by the slit are rendered parallel by the lens at b, the beam then traverses the prisms 1, 2, 3, 4, 5, and 6, and the spectrum is observed through the telescope B The prism I is attached to the plate c by a screw passing through the triangular support at the angle c, nearest to the collimator, so that it may turn on this axis through a small arc. The prism is also attached to prism 2 by a screw d placed at the other end of its base, and the other prisms are similarly linked together at e, f, g, and h, the last prism 6 being linked by the screw i to the arm D which bears the observing telescope. The support of each prism carries, rigidly fixed to it, a flat brass bar placed at right angles to the base of the prism and which has a slot at its extremity. The arm D is bent at right angles, and its shorter portion E is also slotted. In the plate c a slot is cut in the direction 7 c, that is, towards the fixed angle of prism I; and under the plate there is a dovetail piece sliding parallel to the slot and carrying a pin k, which is passed through

107.]

SPECTRUM ANALYSIS.

193

the slot in the plate and through the seven slotted pieces attached to the prisms and telescope.

If it is required to observe the more refrangible end of the spectrum, the telescope B must be moved in the direction F G. This moves the train of prisms, causing all the prisms to approach the central pin k, diminishing the diameter of the circle of which k is the centre, and to which the bases of the . prisms are tangents. The first prism being fixed at one angle can only move to a small extent, the second will move a little more, and so on, and by arranging the circle of the prisms to a size which will accord with the refractive index of the glass, the instrument is so fitted that the refracted ray is always parallel to the base of the prism through which it is passing, or in other words, each prism is at the angle of minimum deviation.

Still more powerful spectroscopes have been constructed, in which the light is passed first through the lower half of the prisms, and is then reflected through the upper half by means of a right-angled prism with its hypothenuse vertical placed opposite to the last prism. Mr. Browning has made an instrument in which the light passes no less than four times through the prisms. In these spectroscopes both the collimator and the observing telescope are fixed, and the different portions of the spectrum are brought into view by altering the position of the train of prisms. The prisms can be readily detached from the stand, and to each of them the reflecting prism can be fixed, so that any number less than the whole train can be employed.

The extraordinary delicacy of certain of these spectrum reactions was indicated by Swan (Edin. Phil. Trans. 1857, xxi. 413), who measured it for sodium, by the only accurate method, namely, by dissolving a weighed quantity of the salt in a known quantity of water, and he thus determined with precision the limit of the reaction. Bunsen and Kirchhoff attempted to estimate the sensitiveness of the reaction by deflagrating a given weight of the various salts in the room in which they were experimenting, and diffusing the vapour mechanically through the air, increasing the quantity of the salt, until a gas flame showed the reaction of the peculiar metal, due to particles in suspension. But it is obvious that this ingenious method does not admit of precision, and is liable to lead to an exaggerated estimate of the delicacy of the reaction, from the impossibility of ensuring uniformity in the diffusion of the salt,

:

The sodium reaction is the most sensitive of all by its means Swan could detect of a grain of sodium; and so extensively is common salt diffused, that scarcely any flame can be obtained in which the indication of sodium is absent.

Having observed the position of the bright lines produced by introducing into the flame of a Bunsen gas-burner the chlorides of the metals of the various alkalies and alkaline earths, each of which had been purified for these experiments with great care, Bunsen and Kirchhoff constructed a chart in which the different lines were laid down for each, and they were able, by observing

1