240 THE SCIENTIFIC PAPERS OP 



the flame, which may oxidize the iron on the surface, forming an 

 oxide or cinder, which is then transferred to the carbon at the 

 bottom, in consequence of the general agitation of the mass. 



This view I am, however, in a position to disprove by my recent 

 experience in melting cast steel upon the open flame bed of a 

 furnace, having invariably observed that no oxidation of the 

 unprotected fluid metal takes place so long as it contains carbon in 

 however slight a proportion. 



But being desirous to ascertain by positive proof, what is the 

 behaviour of silicon, and carbon, in fluid cast iron, when contact 

 with the atmosphere, or the flame of the furnace, is strictly 

 prevented ; I instituted the following experiment at my Sample 

 Steel Works, at Birmingham : 



Ten cwts. of Acadian pig metal, and one cwt. of broken glass, 

 were charged upon the bed of a regenerative gas furnace (usually 

 employed for melting steel upon the open hearth). 



The bed of this furnace was formed of pure silicious sand, and 

 one object in view was to ascertain whether any reaction takes 

 place between silica and fluid cast metal, it being generally 

 supposed that metallic silicon is produced under such circum- 

 stances by the reducing action of the carbon in the metal upon the 

 silica or silicates present. The cast metal employed in this 

 experiment was Acadian pig, containing silicon, 1*5 per cent. ; 

 carbon, 4*0 per cent. 



In the course of an hour, the metal and glass were completely 

 melted. A sample was taken out, containing silicon, 1-08 per cent. ; 



, ( 0'6 per cent, combined carbon, 

 carbon, 2-90 per cent. | 2 . 3 ^ ^.^ 



At the end of the second hour, another sample was taken out 

 and tested, the result being silicon, '96 per cent. ; carbon, 2*40 

 per cent., combined. 



The physical condition of the metal had now undergone a 

 decided change. The carbon having wholly combined with the 

 iron, rendered it extremely hard. 



The amount of silicon having steadily diminished, these results 

 prove that no silicon is taken up ly fluid cast metal in contact with 

 silica or silicates. 



The reduction of the amount of silicon in the metal might be 

 accounted for by the presence of minute quantities of oxides of 



