lOGO 
PROFESSOR W. N. HARTLEY ON 
do not appear quite similar to the Ijands of the metal, nor of pure compounds of 
manganese, and this is partly owing to the continuous spectrum of carbon monoxide, 
partly to carbon bands, such as are exhibited in hydrocarbon flames, and also in part 
to bands of iron. This will l^e readily understood from an examination of the photo¬ 
graphs and the wave-lengths of the edges of ca.rbon bands, for it is evident that as 
the bands of carbon overlie those of manganese, the latter are most certainly obscured 
by the former, and the former by the latter; because, while the former consist of 
narrow bands degraded towards the violet, the latter are composed of similar narrow 
bands degraded towards the red. One overlying the other can produce the appear¬ 
ance of a broad band, which is not degraded, but appears as a continuous spectrum, 
and in this the only part distinguishable is the more refrangible edge of the manga¬ 
nese band, and the less refrangible edge of that of carbon. Between the two, 
neither bands nor lines can be distinguished if the rays are strong. There are some 
few lines due to carbon monoxide, and certain bands due to an oxide of manganese, 
either MnO or Mn 304 . It is not likely that MnOo, which is easily decomposed by 
heat, could exist in the Bessemer flame; the vaporized oxide must be a substance of 
no small stability. Beferences to the lines, measured by Watts, have been inserted 
opposite to the line of nearest wave-length in the Bessemer spectra. Lines not 
identified by Watts are either lines of iron, of manganese, or carbon bands. The 
lines of metallic manganese were not identified, probably because Watts used 
manganese dioxide, or carbonate, for his comparison spectrum, instead of the metal 
heated in the oxyhydrogen flame. 
Lines in the spectrum of the Bessemer^flame, which are more refrangible than the 
solar line H, have not been examined before, and this portion of the spectrum is 
especially interesting. It extends to some small distance beyond the solar line 0 of 
Cornu’s Spectre Normal, or on some plates as far as P X 3361‘5. Most of the lines 
are veiy strong and sharp, forming very characteristic groups. They have nearly all 
been identified with iron ; they are all strong lines, as seen in the arc spectrum of 
iron, and they are coincident with lines in the sun. They have been identified with 
the lines photographed from the spectrum of Turton’s tool-steel, but steel in the 
oxyhydrogen blow-pipe yields no greater number of iron lines than occur in the 
Bessemer flame. Ferric oxide, under the same conditions, exhibits a spectrum with a 
band, and still fewer lines. It certainly appears as if the temperature of the 
Bessemer metal during the “boil” is as high as, or approaches that of, the 
oxyhjnlrogen blow-pipe flame, when the oxygen contains 10 per cent, of nitrogen, as 
the commercial oxygen supplied to us was found to do, and the flame is being used 
for spectroscopic purposes in the manner already described (‘Phil. Trans.,’ 1894.) 
Owing to the complicated nature of the spectra, the wave-lengths of bands and lines 
exhibitcid during different periods of the “blow” have been tabulated, with a 
descnprion of each feature of the spectrum to which the measurements belong, and in 
parallel columns there are references to Watts’ measurements, and those of other 
