1894.] Flame Spectra at High Temperatures. 197 



difference being chiefly in the iron line being stronger in the slag 

 spectrum. This explains the fact observed by Brunner, namely, 

 that when a converter is being heated with coke after it has been 

 used, but not re-lined, the spectrum of the Bessemer flame makes its 

 .appearance ; manifestly it comes from the adhering slag. 



The luminosity of the flame during the " boil " is due, not merely to 

 the combustion of highly heated carbonic oxide, but also to the pre- 

 sence of the vapours of iron and manganese in the gas. 



The disappearance of the manganese spectrum at the end of the 

 " fining stage," or third period, is primarily due to a reduction in the 

 quantity of heated carbon monoxide escaping from the converter, 

 which arises from the diminished quantity of carbon in the metal. 

 When the last traces of carbon are gone, so that air may escape 

 through the metal,- the blast instantly oxidises any manganese, either 

 in the metal or in the atmosphere of the converter, and, furthermore, 

 oxidises some of the iron. The temperature must then fall with 

 .great rapidity. 



The entire spectroscopic phenomena of the " blow " are undoubtedly 

 determined by the chemical composition of the molten iron, and of 

 the gases and metallic vapours within the converter, the temperature 

 of the metal and that of the issuing gases. 



The Temperature of the Bessemer Flame. 



The probable temperature of the Bessemer flame at the finish is 

 that produced by the combustion in cold air of carbonic oxide heated 

 to about 1580 C., that is to say, to the temperature which, according 

 to Le Chatelier (' Comptes Rendus,' vol. 114, p. '670), is that of the 

 bath of molten metal from which the gas has proceeded. The bath 

 of metal acts simultaneously as a means of heating the blast, producing 

 the gas, and as a furnace, on the regenerative principle, which heats 

 the gas prior to its combustion. The heating effect is therefore 

 cumulative. The temperature, as is well known, can easily rise too 

 rapidly, and the metal has then to be cooled by throwing cold pig- 

 iron, or even old ingot moulds, into it, 



If we may judge by the lines and bands belonging to iron and 

 manganese which have been measured in photographed spectra of the 

 Bessemer flame, the temperature must nearly approach that of the 

 oxyhydrogen flame, and may easily attain the melting point of plat- 

 inum, namely, 1775 C. (Violle). 



Marshall Watts observed ('Phil. Mag.,' 1870) that the sodium 

 lines 5681 and 5687 may be employed as an index of temperature, 

 .since they are present in the spectrum of any flame containing sodium 

 which is hot enough to melt platinum, but do not appear at lower 

 temperatures. The Bessemer flame does not show this double line, 

 but only the D lines. 



