END-ON VIEWS OF GAS SPECTRA UNDER HIGH DISPERSION. 151 
band-system of the blow-pipe-flame, on the contrary, supersedes it almost entirely. In other 
tubes the two systems are simply superposed upon each other, or mingled together in various 
proportions of intensity. 
Similar to the indepeudent variations of the two orange-citron-green band-combinations, I 
noticed a marked character of individuality in the dark blue blow-pipe-flame band at the solar 
line G, which with its faint precursor, and with one strong blue band (near F) between it and 
those two sets just noticed, also constitute together very persistent features of carbon impurities 
in gas vacuum tubes. This most refrangible blow-pipe-flame band makes its appearance to- 
gether with a prodigious development, just following it between G and H, of the six or seven- 
rayed violet line-cluster which Professors LivEInc and DEWwaR have recently ascribed, in a 
Paper presented to the Royal Society of London, to Cyanogen, with extraordinary luminosity 
in a marsh-gas tube. The latter line-cluster I have observed as a single and solitary lwcidwm 
in the beautifully blue arc of flame between the pure carbon poles burned in the Brush’s, or 
Anglo-American Company’s Electric light. Its freedom in each of these two cases from any 
simultaneous traces of the less refrangible band or cluster midway between Hf and Hy, also 
referred by the same writers to Cyanogen, but which I have never yet detected in gas vacuum 
tubes, makes me doubt the correctness of their interpretation that it belongs to cyanogen, 
and to venture to attribute the six-lined blue-violet clusture just beyond G, and perhaps 
also the most refrangible band at G of the blow-pipe-flame spectrum, to the incandescence of 
marsh-gas. 
Ably supported as the assumption is, no doubt, that there exist low-temperature spectra 
of the chemical elements, particularly of the metalloids, and sound as some of the evidence is, 
without question, by which the important theory has been established, yet the identification of 
the low-temperature spectrum of carbon, if it exists, cannot be said yet to be unanimously 
represented as accomplished. The independent radiancies of the several individual bands and 
band-combinations which together constitute the carbon impurities of gas vacuum tubes, 
including that presented in the blow-pipe flame, are so strikingly various and unconnected, 
that a choice among the band-series produced respectively by olefiant-gas, by carbonic oxide, 
by marsh-gas, by cyanogen, and it may be by other carbon compounds, is one of some 
difficulty, before it can be positively affirmed which of all these is the low-temperature 
spectrum of elemental carbon by itself. That several spectra of different degrees of tempera- 
ture may exist, will scarcely explain the predominance under the same conditions, of three 
different spectral systems in such tubes as those of carbonic oxide, olefiant-gas, and marsh-gas, 
nor for the arbitrary admixtures of these three separate systems which high dispersion and 
accurate measurements easily detect as present in various abundances as common impurities in 
ordinary gas vacuum tubes. 
The following measurements of the linelets and of some intruding shaft-lines in the 
citron-band of a carbonic acid tube, made a year later with the improved tubes allowing end- 
on vision, but with the same prisms and micrometer-screw of the aurora-spectroscope, will 
show the precision of detail of which the instrument was capable, and at the same time the 
regular and definite character of these two kinds of carbon lines and bands which present 
themselves in vacuum tubes as intruders one upon the other. A translation of the readings 
into wave-numbers per British inch, made at the time, although not possessing the accuracy 
which Professor Piazzt Smyru’s later conversion tables for the instrument’s readings would 
have given them, is added to the measures of the list. A certain regularity of the intervals 
among the linelets (although not among the intruding lines) is discernible, which may, perhaps, 
