Flame Spectra observed at Swedish Bessemer Works. 83 



towards 6350. Sometimes a line or faint band appeared at 6106. 

 When the converter was hot the appearance of the group varied 

 incessantly, now being that of a bright band from 5585 to 5718, 

 shading away towards the red, now being a much narrower one from 

 5585 to 5608. The broad band often divides for a moment into two, an 

 absorption line 5625 flashing out alternately with what appears to be 

 an absorption band 5648 — 5600. If the converter was not so hot, /3 X 

 only was visible as usual, 7, was always visible, and sometimes also 

 7 2 . Of the group b there was nothing seen but the brilliant edge 

 5153. 



The Nykroppa Bessemer pig-iron generally contained — 



70 per cent, of combined carbon, 

 4*07 ,, graphite, 



0*75 „ silicon, 



1*36 ,, manganese, 



0*015 ; , phosphorus, 

 and traces of sulphur. 



Two samples of this pig-iron were examined for manganese ; the 

 first on one occasion when the group c had appeared very faintly, 

 and the other one when four bands of the group e had been distinctly 

 visible. The result was that the former sample contained a little 

 more manganese than the latter, which proves that the changes in 

 the proportion of manganese have no effect upon the more or less 

 distinct appearance of the group <?. 



The Hacffors Spectrum. 



This spectrum differs from the former only in some small details. 

 The group a consists here of a broad bright band with distinct limits 

 X = 6277 and X = 6179. This band has on the side of less refrangible 

 rays the continuous background of rays, but on the more refrangible 

 side of 6179 it has, as it seems, an absorption band, which extends to 

 the narrow and bright band 6106 (Li/s) and then the bright background 

 adjoins it again. In the group /3 there is a new line visible 5860 

 ( Mn 2 3 ), and the band 5712 appears more distinctly limited already at 

 5585. The iNykroppa and Hagfors spectra fully agreed, except in the 

 following details, namely, the group e of the latter spectrum was 

 more often indistinct, and that it never exhibited so many bands as 

 the former. 



When the converter was being warmed there appeared certain 

 bands, which were not visible during the " blow," viz., a faint 

 narrow band with X, = 5923, and a distinct one with X = 5525 — 5491 

 (5525 being the most brilliant edge), not, however, quite so brilliant 

 as /3l 



g 2 



