96 An Investigation of the Spectra of Bessemer Flames. 



(2.) The plienomeim of the " basic " Bessemer blow differ considerably 

 from those of the " acid "process. 



First, a flame is visible from the commencement of Mowing, or as 

 soon as the cloud of lime dust has dispersed. We conclude that the 

 immediate production of this flame is caused by carbonaceous matter 

 in the lining of the vessel, that its luminosity is due partly to the 

 volatilisation of the alkalies, and to the incandescence of lime dust 

 carried out by the blast. 



Secondly, volatilisation of metal occurs largely at an early period 

 in the blow, and is due to the difference in composition of the 

 metal blown, chiefly to the smaller quantity of silicon. There is 

 practically no distinct period when silicious slags are formed in the 

 " basic " process, and metals are volatilised readily in the reducing 

 atmosphere, rich in carbon monoxide." 



Thirdly, a very large amount of fume is formed towards the close of 

 the second period. This' arises from the oxidation of metal and of 

 phosphorus in the iron phosphide being productive of a high tempera- 

 ture, but little or no carbon remaining. The flame is comparatively 

 short, and the metallic vapours carried up are burnt by the blast. 



Fourthly, the " over-blow " is characterised by a very powerful 

 illumination from what appears to be a brilliant yellow flame : a dense 

 fume is produced at this time composed of oxidised metallic vapours, 

 chiefly iron. These particles are undoubtedly of very minute dimen- 

 sions, as is proved by the fact that they scatter the light which falls 

 on them, and the cloud casts a brown shadow, and, on a still day, 

 ascends to a great height. The spectrum is continuous, but does not 

 extend beyond wave-length 4000. This indicates that the source of 

 light is at a comparatively low temperature, approaching that of a 

 yellowish-white heat. We conclude, therefore, that the light emanates 

 from a torrent of very small particles, liquid or solid, at a yellowish- 

 white heat. The " flame " can have but little reducing power at this 

 stage, and this, together with its low temperature, accounts for the 

 very feeble lines of lithium, sodium, potassium, and manganese seen in 

 the photographs, or by eye observations. 



Fifthly, the spectra of flames from the first stage of the " basic " 

 process differ from those of the " acid " process in several particulars. 

 The manganese bands are relatively feeble, and lines of elements, not 

 usually associated with Bessemer metal, are present. Both the 

 charges of metal and of "basic" material contribute to these. Lithium, 

 sodium, potassium, rubidium, and caesium have been traced mainly to 

 the lime ; manganese, copper, silver and gallium to the metal. Other 

 metals, such as vanadium and titanium, are not in evidence, because 

 they do not yield flame spectra ; they, together with chromium, pass 

 into the slag in an oxidised state. 



(3.) Differences in the intensity of metallic lira's. The intensity of 



