28 
SCIENCE. 
“II est bien Evident que les cas dont nous venons de 
parler, ne forment pas une exception a la loi g£n6rale 
dnonc6e ci-dessus, savoir que chaque corps simple ne pent 
donner qu’un seul spectre. En effet, si Ton suppose que 
l’etat allotropique est du 4 la constitution mol6culaire du 
corps, soit que les moldcules se combinent les lines avec les 
autres, soit qu’elles s’arrangent entre dies d’une certaine 
maniere, cet 6tat allotropique possfedera au point de vue 
spectroscopique, toutes les proprifetfes significatives d’un 
corps compost, et par consequent il doil etre d6composdde 
la meme fa$on que celui-ci par les efFets de la dfechargc dis- 
ruptive de l’electricite.” * 1 
I say that in this paper Angstrom recalled his own in favor 
of Plucker’s view, because (as it has been remarked by Dr. 
Schuster' 2 ) the word “ element ” is used in a special sense — 
because in reality allotropic states are classed as compounds, 
that particular allotropic state which is to be regarded as 
truly elemental not being stated, nor any reason given why 
one should be thus singled out. 
In the letter to which I have just referred Dr. Schuster 
gives an instance in which in order to show that elementary 
bodies did not really possess two spectra, a double spec- 
trum was assigned to an acknowledged compound ; the 
fluted spectra of hydrogen and carbon which differ from each 
other as widely as fluted spectra can, being both ascribed 
to acetylene. 
Salet in his admirable work on the Spectra of the Metal- 
loids, 3 was driven to the conclusion that many of these 
bodies must be held to possess two spectra. His conclusions 
are thus expressed : — 
“ Nous avons compare le spectre d’absorption du brome 
et de l’iode a leur spectre electrique, et cette comparaison 
nous semble mettre hors de doute la possibility des spectres 
doubles. . . . 
“Nous avons obtenu, par voie electrique, un spectre 
primaire de l’iode correspondant a son spectre d’absorption. 
Le soufre, le selenium et le tellure nous ont offert des spec- 
tres de combustion tres-analogues aux spectres primaire 
obtenus par voie electrique, mais different essentiellement 
des spectres des lignes. . . . 
“ Nous avons produit le spectre primaire de 1 ’azote avec 
differents corps qui n’ont absolument de commun que 
l’azote ; nous pensons done avoir demontre qu’il appartient 
bien r<fellement a ce metallolde.” ( Annales de Chemie et 
de Physique, 4 serie, tome xxviii. pp. 70, 71). 
In 1868 Wullner 4 gave his attention to this subject, and 
strongly supported Plucker’s view of the existence of 
double spectra, indicating at the same time that the differ- 
ence of temperature must be regarded as the sole cause of 
the phenomenon, adding, however, “ a decomposition with 
further elements is not to be thought of.” In the case of 
hydrogen he showed that the banded spectrum ascribed to 
acetylene really depended upon a change in the emissive 
power brought about by an alteration of temperature. 
Touching oxygen, he showed that three distinct spectra may 
be obtained, while in nitrogen two are observed. 
I may say that in my early laboratory experiments I was 
at first led to think that, in the case of metallic vapors, 
Angstrom’s first expressed opinion was correct, and I said 
so. But after more experience and knowledge had been 
acquired, I was compelled by the stern logic of facts to 
abandon it, and I showed, first, that more “orders” of 
spectra — to use Plticker’s term — were necessary, and then 
that the line spectrum itself was in all probability com- 
pound ; that is, that it was in some cases built up by the 
vibration of dissimilar molecules, some of which might even 
give us a fluted spectrum, if we could study them alone. 
Although, however, in the views 1 have expressed on 
former occasions I have had the advantage of the support of 
the opinion of Pliicker and Angstrom, and later of Dr. 
Schuster,' 2 not to mention others, I am aware that though 
there is a general consensus among spectroscopic workers 
that double spectra cannot be ascribed to impurities, it is not 
absolute. 
I propose therefore in this place to refer to a special case 
in which this question has been recently brought prominently 
forward. 
O 
'Angstrom and Thalen’s 11 Recherches sur les Spectres des Metalloides,” 
p, 5. 2 Nature, vol. xv. p. 447. 
3 Ann. de Chirnie et de Physique , 1873, V °J- xxviii. p. 1. 
1 Phil . Mag., sec. 4, vol. xxxvii. p. 405. 
I have already stated that Angstrdm, who was the first to 
map the line-spectrum of carbon, ascribed the flutings 
ordinarily seen in the carbon compounds to acetylene. 
Now Attfield, in 1862, as a result of a most carefully con- 
ducted and admirably-planned set of experiments, came to 
the conclusion that the flutings were really due to carbon : 
in short, that carbon, like hydrogen, iodine, sulphur, nitro- 
gen, and other bodies, had a fluted spectrum as well as one 
consisting wholly of lines. 
The work of Attfield will be gathered from the following 
extract from his paper {Phil. Trans., vol. clii. part 1, p. 221 
et seq.) : — 
“ On iccently reading Swan’s paper by the light that 
Professors Bunsen and Kirchoff have thrown on the sub- 
ject, I came to the conclusion that these bands must be 
due to the incandescent carbon vapor; that, if so, they 
must be absent from flames in which carbon is absent, and 
present in flames in which carbon is present ; that they 
must be observable equally in the flames of the oxide, sul- 
phide, and nitride as in that of the hydride of carbon ; and, 
finally, that they must be present whether the incand- 
escence be produced by the chemical force, as in burning 
jets of the gases in the open air, or by the electric force, as 
when hermetically-sealed tubes of the gases are exposed to 
the discharge of a powerful induction-coil. . . . 
“To establish the absolute identity of the hydro- and 
nitro-carbon spectra, excluding of course the lines due to 
nitrogen, they were simultaneously brought into the field of 
the spectroscope: one occupying the upper, and the other 
the lower half of the field. 
“ This was readily effected after fixing the small prism, 
usually supplied with spectroscopes, over half of the nar- 
row slit at the further end of the object-tube of the instru- 
ment. The light from the oxyhydrocarbon flame was now 
directed up the axis of the tube by reflection from the little 
prism, while that from the oxynitrocarbon flame passed 
directly through the uncovered half of the slit. A glance 
through the eye-tube was sufficient to show that the char- 
acteristic lines of the hydrocarbon spectrum were perfectly 
continued in the nitrocarbon spectrum. A similar ar- 
rangement of apparatus, in which the hydrocarbon light 
was replaced by that of pure nitrogen, showed that the re- 
maining lines of the nitrocarbon spectrum were identical 
with those of the nitrogen spectrum. In this last experi- 
ment the source of the pure nitrogen light was the electric 
discharge through the rarefied gas. 
“ The above experiment certainly seemed to go far to- 
wards proving the spectrum in question to be that of the 
element carbon. Nevertheless, the ignition of the gases 
having been effected in air, it was conceivable that hydro- 
gen, nitrogen, or oxygen had influenced the phenomena. 
To eliminate this possible source of error, the experiments 
were repeated out of contact with air. A thin glass tube 
1 inch in diameter and 3 inches long, with platinum wires 
fused into its sides, and its ends prolonged by glass quills, 
having a capillary bore, was filled with pure dry cyanogen 
and the greater portion of this gas then removed by a good 
air-pump. Another tube was similarly prepared with ole- 
fiant gas. The platinum wires in these tubes were then so 
connected with each other that the electric discharge from 
a powe-ful induction-coil could pass through both at the 
same time. On now observing the spectra of these two 
lights in the simultaneous manner previously described, 
the characteristic lines of the hydrocarbon spectrum were 
found to be rigidly continued in that of the nitrocarbon. 
Moreover, by the same method of simultaneous observa- 
tion, the spectrum of each of these electric flames, as they 
1 Dr. Schuster’s recently published investigations are as follows: — 
Mr. Lockyer’s investigations have shown that most bodies give us a 
continuous spectrum, as a gas, before they condense, and many at a con- 
siderable temperature above the boiling point. Mr. Lockyer has rightly 
drawn the conclusion from these facts, that the atomic aggregation of 
the molecules is the. cause of the different orders of spectra. 
That the d scontinuous spectra of different orders (line and band spectra) 
are due to different molecular combinations, I consider to be pretty well 
established, and analogy has led me (and Mr. Lockyer before me) to ex- 
plain the continuous spectra by the same cause ; for the change of the 
continuous spectrum to the line or band-spectrum takes place in exactly 
the same way as the change of spectra of different orders into each other. 
Analogy is not a strong guide, yet some weight may be given to it in a 
case like the one under discussion, where experiment hitherto has failed to 
give a decided answer. (Dr. A. Schuster on the Spectra of Metalloids, 
Phil. Trans. Royal Society, 1879. Part i. page 38 and 89, note). 
