FLAME SPECTRA AT HIGH TEMPERATGRES. 
1089 
no indication, except at the close of the ‘‘ fining stage,” when the carbon remaining 
unburnt is reduced to a very small proportion of that originally present, it is quite 
evident that the most rational and practicable mode of operating is to remove it all, 
and then to add a sufficiency of a carburized metal which will produce a steel of the 
required hardness and composition; since both by the appearance of flame and fume, 
as well as by spectroscopic analyses, the complete termination of the “fining stage” 
is clearly indicated. 
The constitution of the Bessemer-flame spectrum, as established by this investiga¬ 
tion, and the cause of the continued ap|)earance of the manganese bands and iron 
lines, even after all the manganese has been removed from the metal, which has been 
explained, affords scientific reasons for not only continuing to j)ursue the course which 
has been universally adopted, but of not departing therefrom. The practice of 
complete decarburization is most rapidly and exactly carried out, and it has yielded, 
and continues to yield, enormous quantities of mild steel or soft iron, in a high state 
of purity and of remarkable constancy in composition. 
Summarij and Conclusions. 
1. The complex nature of the Bessemer-flame spectrum is oMung to the superposition 
of bands of manganese, carbon, carbonic oxide, possibly also of manganese oxide, and 
of the lines of iron, manganese, potassium, sodium, lithium, and hydrogen. The bands 
of manganese are to some extent obscured, first, by the strong continuous spectrum 
of the carbonic oxide flame, secondly, by the bands of carbon ; for, while the man¬ 
ganese bands are degraded towards the red, the overlapping carbon bands are- 
degraded in the opposite direction, that is, towards the blue. 
2. The cause of the non-appearance of the lines in the spectrum at the beginning of 
the “ blow ” is the comparatively low temperature at this period, very little above that 
of the molten metal, and the free oxygen that escapes with carbon dioxide, giving 
a gaseous mixture which contains too small a proportion of carbonic oxide. The 
alkalies come from the ganister brick lining of the converter, and therefore exist as 
silicates present in very small proportion. Silicates—such, for instance, as felspar— 
do not readily disclose the alkalies they contain until heated in the oxyhydrogen 
flame, but at this high temperature the metals potassium, litliium, and rubidium have 
been detected with the greatest ease in such silicates. Similarly, the alkali metals 
do not show themselves in the flame until a layer of slag has been formed, and the 
temperature has risen sufficiently high fon the constituents to be vaporized. 
3. There can be no ddubt that at the temperature of the “ boil” both metallic man¬ 
ganese and iron are freely vaporized in a current of carbonic oxide, which, in a highly 
heated state, rushes out of the bath of molten metal. The evidence of this is the 
lines and bands of iron and of manga-nese, photographed and compared with the lines 
and bands in the spectra of various alloys of iron and manganese. 
MDCCCXCIV.—A. G Z 
