Flame Spectra observed at Swedish Bessemer Works. 89 



the spectrum of magnesium. According to Liveing and Dewar,* 

 Mg gives in the arc the following values for X: 5711, 5528, 5183, 

 5172, 5167, 4704. The line 5711 appears brightly, and generally 

 very distinctly, just when a little piece of magnesium has been 

 placed in the arc ; it then rapidly diminishes, and by degrees 

 becomes very faint, or completely disappears. The same line 

 appears also in the spark when taken from a solution of MgCl 2 . 

 In the spark all lines mentioned above are to be found. The 

 groups 5183 — 5172 — 5167 that correspond to 6 in the sun spectrum 

 appear, however, comparatively fainter than in the arc,| on account 

 of which one might conclude that with another temperature and 

 another pressure the lines 5710 and 5529 would remain longest, 

 that is to say, that these would differ from the b lines, and by 

 this one might also explain the reason why 5712 appears more dis- 

 tinctly in the Hagfors spectrum than in the Nykroppa spectrum. 



The Magnesium Oxide Spectrum. 



Magnesium oxide has the sharpest edge of its brightest band at 

 5006, X and corresponds better with the faint band 5002 in the Langs- 

 hytte spectrum than with the Mn 2 3 band, to which also its origin 

 might be attributed. 



The Barium Oxide Spectrum. 



This spectrum has a band with \ = 5536, which may account for 

 he line 5532 in the Nykroppa spectrum. 



II. 



The Spectrum of the Blue Flame. (Fig. 1, spectrum IV.) 



This spectrum was visible only for a few seconds during each 

 " blow," and was consequently difficult to deal with. Often when it 

 appeared the instrument was not properly adjusted, or, owing to 

 parallax, no reliable values could be obtained for the position of the 

 lines. The spectrum being of so short a duration, there was no time 

 to write down more than one or two lines during a " blow." It 

 seems, however, to be of interest in more than one respect, and on 

 this account I have examined the spectrum as well as circum- 

 stances allowed, only making one or two observations of the faintest 

 and most rarely appearing of the twenty- eight bands and lines that 

 are visible. In the following table the results of these observations 

 are given. 



* ' Koy Soc. Proc.,' vol. 30, p. 97. 



t See Capron, ' Photographed Spectra,' London, 1877, Plate XI. 

 X Lecoq, • Spectres Lumineux,' p. 86. 



