§ 3. Magnesium.

In the case of magnesium (which, as regards its chloride, bromide, and iodide, followed the behaviour of the alkali metals, to which I shall immediately refer, rather than that of barium and strontium), the fluoride exhibited the same stubborn resistance to the action of the spark. It is to be remarked that these three fluorides are non-volatile, and so infusible that even after long exposure to the current there was little or no indication of coherence; in fact, in the case of magnesic fluoride, the salt was distinctly seen to be blown out of the cup as a cloud of dust, and when one of these particles was converted into vapour in the spark, the spectrum exhibited fragments of lines sharp at both ends.

On blowing a cloud of magnesic fluoride in fine powder through the spark, b in particular was seen as a series of three pointed lines.

4. Sodium and Lithium.

Sodium and lithium, elements of low atomic weight, were the next metals experimented upon.

Some sodic fluoride was inserted into one of the aluminium cups, and opposite was placed a clean blunt aluminium point ;* the small coil and a jar were employed.

* Special precautions being taken to keep the poles the same distance apart in all the experiments.

The break

On passing the spark, D only was seen. was then readjusted and the spark made heavier, but the result was the same. Some new and moist sodic fluoride was then placed in the cup, but the result still remained as before.

Sodic Chloride was then treated in the same way, a fresh aluminium pole being reserved for it and placed opposite to its cup. D was present and was bright; the double line in the red once flashed in, but it was not again seen though the cup was charged and recharged with the salt repeatedly.

Sodic Bromide treated in the same way gave D, the red line being seen but once; D, however, was brighter than before.

Sodic Iodide treated as above gave D and the red double line, which remain constantly visible, the double green line near D being also occasionally seen. D was intensely brilliant, and the salt fumed away from the pole in a dense white smoke.

A fresh attack with more powerful apparatus was then made. One of App's 6-inch quantity coils and a jar with about 224 square inches of coated surface were employed with a powerful battery of five one-pint Grove's cells. The result, however, went exactly in the same way. The Sodic Iodide and Bromide gave all the metallic lines of sodium, which were very brilliant; whilst the Chloride gave D and the double red very bright, and stretching all across the spectrum. Fluoride gave also D and the double red line; but the latter only extended three quarters across the spectrum,

The

and neither D nor the red line was so bright as they were in the chloride. Further observations showed that under certain circumstances all the lines appeared even in these latter salts; but they were so dim as to be scarcely visible, and the fact of these compounds behaving in direct contravention to the observations with lead was established.

The following experiments were made with Lithic Iodide and Chloride in coal-gas:

Lithic Iodide gave the red line of this metal (W L 6705), extending all across the spectrum, but it was very faint, the orange line (6102) was very brilliant and about two thirds across, the blue line (4603) was very short and nebulous when seen, which was only on one occasion. This differs but little from the spectrum given by the metal itself, except that the lines are in the latter case much brighter, and that the red and orange lines extend right across the spectrum, and the blue three quarters across.

*

Lithic Chloride gave the red line (6705·2) thin and faint, but all across the spectrum; the orange very bright and across the spectrum; the blue line was also visible, but it did not extend across the spectrum.

It will be seen from the above that lithium, the least electro-positive of the alkali metals, approaches in its spectroscopic behaviour the metals of the alkaline

* A line at 4972 was seen also in the spectrum when the metal was used; but this has never been seen by Kirchhoff, Thalén, or any other observer except Huggins, in lithium.

There can be no doubt that it is the cæsium line, and that it is due to the presence of a trace of cæsium existing as an impurity in the lithium.

earths, strontium and barium, as it approaches them in some points of its chemical behaviour. Thus the spectrum of its iodide differs from that of the chloride as the spectrum of baric iodide differs from that of baric chloride, and not as the spectrum of sodic iodide does from sodic chloride, as might have first been supposed from the usual position, among the alkalies, assigned to the metal.

§ 5. Flame and Weak Electric Discharge Spectra.

In these experiments, then, in addition to information. on the points already referred to, we had accumulated some enabling a comparison to be made between the spectra of salts observed with a weak electric discharge, and those observed in a Bunsen flame, which were among the first to be studied.

Some flame spectra were specially observed in order

* These experiments were as follows:

BARIUM.-Baric Iodide.-This salt gave the spectrum proved afterwards to be due to the oxide and the line 5534'5 very distinctly; it coloured the flame a greenish yellow, and fused to a globule. Baric Bromide gave the oxide spectrum and 5534'5 with difficulty; the spectrum was not very bright, and the flame but little coloured. Baric Chloride gave the same spectrum as the two salts mentioned above; but the spectrum was much brighter, and the flame was coloured a bright pale green. Baric Fluoride gave scarcely a trace of the oxide spectrum, and 5534'5 was very faint indeed; but no signs of fusion were visible, no bead being formed, and the flame was only coloured slightly and in parts.

STRONTIUM.-Strontic Iodide, heated on platinum wire in the Bunsen flame, gives the spectrum in the red, so well represented in Bunsen's and Kirchhoff's drawing, and the great blue line 4607 5 of the metal. Strontic Bromide behaved much as the iodide did, but showed more of the structure in the red. 4607'5 was also always present, and very bright, a considerable change from its appearance in the iodide, in the spectrum of which it was for a time

to note how the long and short lines behaved, beads of the salts being heated in the Bunsen flame on loops of platinum wire.

It was seen, then, that when spectra produced by flames are compared with those produced by the lowtension spark, the spectra of the metals in the combination are in the former case invariably more simple than in the latter, and that they are simplified to such an extent that only the very longest line is left; thus:

Baric Iodide with the low-tension spark gives five-andtwenty lines. In the flame it gives but one, and that the longest, namely 5534'5.

Baric Bromide gives five-and-twenty lines with the spark; only one in the flame, the same longest line 5534'5.

Baric Chloride five-and-twenty lines in spark and one, 5534'5, in flame.

Baric Fluoride four lines in spark, 5534'5 alone in flame.

Again, taking the case of strontium, we find that in the case of strontic iodide thirty-two lines are observed in the spark, one alone in the flame, and that is the longest, namely 4607'5, a line by far the longest in the spectrum of strontium.

Strontic Bromide gives also thirty-two lines in the spark and but one in the flame, the same longest line 46075. faint and then became brighter. Strontic Chloride gave the bands very brightly at first, but not so brightly after a time; 4607'5 was fainter, but very distinct. Strontic Fluoride refused to give any trace either of the strontium or compound spectrum; it is, in fact, only capable of being heated to a white heat and giving a continuous spectrum.