258 
NATURE 
[| Fan. 11, 1883 
spectrum with acetylene, by observing that cyanogen compounds 
are continously formed when the are discharge takes place in 
gases containing nitrogen, and that in all probability their forma- 
tion is due, as Berthelot has shown, to a reaction taking place 
between acetylene and nitrogen. Berthelot is positive in his 
assertions that cyanogen is never formed by a direct combination 
between carbon and nitrogen, and any such apparent combination 
is due to impure carbon, and the presence of an imperfectly 
dried gas ; in other words, hydrogen is essential to the production 
of cyanogen under such conditions according to the views of 
Berthelot. 
The fact that carbonic oxide, which is one of the most stable 
binary compounds of carbon, forms a distinct spectrum of a 
character similar to that of the flame spectrum, tended to support 
the view that the flame spectrum might originate with acetylene. 
The similarity in the character of the magnesium-hydrogen 
spectrum to that of the hydro-carbon flame spectrum induced us 
to believe that they were due to similarly constituted compounds, 
and seeing we felt sure about the accuracy of the view, which 
assigns the former spectrum to some compound of magnesium 
with hydrogen, we accepted the analogy in favour of the sup- 
position that acetylene is the substance which produces the flame 
spectrum ; or, at any rate, that acetylene is a necessary concomi- 
tant of the reaction taking place during its emission, and con- 
sequently might give rise to this peculiar spectrum. 
Having examined this question in the way described, we 
adopted the view of Angstromand Thalén as to the genesis of this 
spectrum in opposition to the views of Attfield, Morren, Watts, 
Lockyer, and others, who held that this spectrum was really 
due to the vapour of carbon. The delicate character of the 
experiments which were required to discover the origin of the 
peculiar set of flutings in the more refrangible part of the 
spectrum of cyanogen made it apparent that, whatever views as 
to the origin of the hydrocarbon flame spectrum were adopted 
by different workers, it could not be regarded hitherto as experi- 
mentally proved which was the correct one. 
With the object of being able to exhaust this question, a 
special study was subsequently made of the ultra-violet line 
spectrum of carbon, in order to ascertain whether any of its 
lines could be found in the spectra of the arc or flame. We 
have found that the ultra-violet lines of metallic substances have 
as a rule the greatest emissive power, and are often present 
when no trace of characteristic lines in the visible part of the 
spectrum can be detected. If carbon resembled the metals in 
this respect, then we might hope to find ultra-violet lines 
belonging to its vapour, thus enabling us to detect the volatili- 
sation of the substance at the relatively low temperatures of the 
arc and flame. The test experiments made on this hypothesis 
are recorded in the paper entitled ‘‘ General Observations on the 
Spectrum of Carbon and its Compounds.” It is there shown 
that some seven of the marked ultra-violet spark lines of carbon 
occur in the spectrum of the arc discharge, although one of the 
strongest lines, situated in the visible portion of the spectrum at 
wave-length 4266, could not be fcund. Further, it is proved 
that the strongest ultra-violet line of carbon does occur in the 
spectrum of the flame of cyanogen fed with oxygen. Thus it seems 
probable that the same kind of carbon molecule exists, at least in 
part, in the arc and flame, as is found to be produced by the 
most poweriul electric sparks, taken between carbon poles or in 
carbon compounds. 
Now the spark gives us the spectrum which is associated with 
the highest temperatures, and therefore it is assumed that this 
spectrum is that of the simplest kind of carbon vapour. If that 
be the case, we cannot avoid inferring that denser forms of 
carbon vapour must exist in arc and flame, emitting, like other 
complex bodies, a fluted, in contrast to a line, spectrum ; or 
rather that the two distinct kinds of spectra may be superposed. 
Such considerations showed that a series of new experiments and 
observations must be made with the special object of reaching a 
definite conclusion regarding the origin of the flame spectrum, 
and the following paper contains a summary of the results of 
such an inquiry. 
Vacuous Tubes.—We have heretofore laid little stress on obser- 
vations of the spark in vacuous tubes on account of the great 
uncertainty as to the residual gases which may be left in them. 
The film of air and moisture adherent to the glass, the gases 
occluded in the electrodes, and minute quantities of hydrocarbons 
of high boiling-point introduced in sealing the glass, may easily 
form a sensible percentage of the residue in the exhausted tube, 
however pure the gas with which it was originally filled. The 
excessive difficulty of removing the last traces of moisture we 
learnt when making observations on the water spectrum, and the 
almost invariable presence of hydrogen in vacuous tubes is doubt- 
less due in great measure to this cause. Wesendonck (Proc. Roy. 
Soc., vol. 32, p. 380) has fully confirmed our observations as to 
this difficulty. By a method similar to that employed by him, 
we have, however, succeeded in so far drying tubes and the gases 
introduced into them that the hydrogen lines are not visible in the 
electric discharge. For this purpose the (Pliicker) tube was 
sealed on one side to a tube filled for some six or eight inches of 
its length with phosphoric anhydride, through which the gas to 
be observed was passed, and on the other side to a similar tube 
full of phosphoric anhydride, which was in turn connected by 
fusion to the (Sprengel) pump. To dry the gas it is not enough 
to pass it through such a tube or even a much longer one full of 
phosphoric anhydride ; it has to beleft in contact with the anhy- 
dride for several hours, and to get the adhering film of moisture 
out of the tube it has to be heated after exhaustion while con- 
nected, as above described, with the drying tubes up to the point 
at which the glass begins to soften, and kept at near this temper- 
ature for some time. To get most of the gases out of the 
electrodes, the tube must be exhausted and sparks passed through 
it for some time before it is finally filled with the gas to be 
observed. Even whenthese precautions have been observed, the 
lines of hydrogen can often be detected in tubes filled with gases 
which should contain no hydrogen. The general result of our 
observations on the spectra observed in tubes so prepared is that 
the channelled spectrum of the flame of hydrocarbons is not 
necessarily conrected with the presence of hydrogen ; it does not 
come and go according as hydrogen is or is not present along 
with carbon in the way that the channelled spectrum of cyanogen 
comes and goes according as nitrogen is present or absent. Our 
observations confirm those of Wesendonck on this point. This 
spectra given by various tubes containing carbon compounds are 
described in the paper. 
Tubes filled with carbonic oxide exhibit in general at different 
stages of exhaustion the following phenomena. When the 
exhaustion is commencing and the spark will just pass, the 
spectrum is usually that of the flame of hydrocarbons and nothing 
else. As the exhaustion proceeds, the spectrum of carbonic oxide 
makes its appearance superposed on the former, and gradually 
increases in brilliance until it overpowers, and at last at a some- 
what high degree of exhaustion, entirely supersedes the flame 
spectrum, This is when no jar is used. In the earlier stages of 
exhaustion, the effect of the jar is to increase the relative 
brilliance of the flame spectrum, and diminish that of the carbonic 
oxide spectrum, and at the same time to bring out strongly the 
lines of oxygen and carbon ; at a certain stage of the exhaustion, 
when the flame spectrum is very weak without the jar, the effect 
of the jar is to bring it out again, but without sensibly enfeebling 
the carbonic oxide spectrum, and without bringing out the 
cervon lines, Ata still higher stage of exhaustion, when the 
carbonic oxide spectrum is alone seen without the jar, the flame 
spectrum is sometimes, not always, brought out by putting on 
the jar, though the carbon lines again show well. At this stage, 
at which the flame spectrum is not seen at all, the distance 
between the strize in the wide part of the tube is considerable, 
and much metal is thrown off the electrodes, which are rapidly 
heated by the discharge. Ina tube filled with carbonic oxide 
mixed with a little air imperfectly dried, when not too highly 
exhausted, the carbonic oxide spectrum, that of the flame of 
hydrocarbons, and that of cyanogen, may all be seen at once 
superposed when no jaris used. With a jar and a tolerably 
high exhaustion the carbonic oxide spectrum, the hydrocarbon 
flame spectrum, and the carbon line spectrum, may al] be seen 
at the same time, 
Spectrum of the Spark in Compounds of Carbon at Higher 
Pressures.—In the spark taken between poles of purified:graphite 
in hydrogen, the spectrum of hydrocarbon flames is seen, and it 
increases in brilliance, as the pressure of the gas is increased up to 
ten atmospheres, and continues bright at still higher pressures so 
far as we have observed, that is, up to twenty atmospheres. 
The spark without condenser in carbonic oxide at atmospheric 
pressure, shows both the spectrum of carbonic oxide and that of 
the hydrocarbon flame; and as the pressure of the gas is 
increased, the former spectrum grows fainter, while the Jatter 
grows brighter, no jar being used. The line spectrum of carbon 
is also visible. At the higher pressures the flame spectrum pre- 
dominates and is very strong. The observations were carried up 
to a pressure of twenty-two and a half atmospheres. On letting 
