Flame Spectra observed at Swedish Bessemer Works, 91 



ning of the " blow," sometimes only after a while, and then the 

 spectrum was continuous. In the former case the "boil" was hotter 

 than in the latter. Experience has shown that the blue flame appears 

 earlier when the pig-iron is strong in silicon, independently of the 

 charge being cold or hot. And if the treatment is basic, the blue 

 flame will appear sooner than if it is acid, other conditions being 

 the same. 



Comparison Spectra. 



That several of the bands that are to be found in the spectrum 

 of the blue flame agree with the Bessemer spectrum is evident, 

 and likewise that these are owing to manganese oxide. From the 

 fact that the more refrangible rays relatively grow in brightness as 

 the temperature becomes higher,* one might suppose that the bands 

 of the group e with X = 4986 and X = 4963, and possibly even 

 other bands, are owing to manganese oxide. The spectrum which, 

 other than that of manganese oxide, agrees best with that of the 

 blue flame is that of silicon. 



The Silicon Spectrum. 



Saletf obtained a band spectrum by passing hydrogen over silicon 

 chloride, and surrounding the hydrogen name with an air current in 

 the same manner as in the production of the phosphorus spectrum 

 mentioned above. I have been unable to decide whether it is owing 

 to elementary silicon or to any of its compounds. I have tried to 

 produce this spectrum by Salet's method, but without success. 



The wave-lengths of the middle of the bands, thus observed by 

 Salet, approximate closely to the most important bands which occur in 

 the spectrum of the blue flame, viz., cc 522, ft 545, 7 501, c 567, and 

 c 487*5 (though faintly), Siu (X = 522) agrees as to its position with 

 the Mn 2 3 band c 3 (X = 5227) ; that in the blue flame appears with 

 greater brightness than the other bands of the c group. 



In addition to these complete agreement is also to be found be- 

 tween the bands 481 and 474 in the Si-spectrum and those of the 

 blue flame 4809 and 4778, though it must be remarked that the band 

 corresponding to Si e is fainter than those of the blue flame. 



There are other reasons, too, for the fact that silicon, or one of its 

 compounds, gives a spectrum where the blue flame appears. Daring 

 the slag-period really nothing takes place but an oxidation of Si and 

 Mn, which process of combustion occasions a considerable increase 

 of the temperature. These elements having passed completely into 

 the slag, the carbon begins to burn, and then the second period, 



* ' Spectres Lumineux,' par Lecoq de Boisbaudran, p. 43. 



f ' Annales de Chimie et de Physique,' vol. 4, No. 28, 1873, p. 66. 



