S/A' U'lI.UAM SIEMENS, F.R.S. 241 



iron, produced in melting the pig metal, which oxides were now 

 innvased by the addition of hematite ore in small doses. 



At the end of the third hour, another sample was taken, 

 containing silicon, '76 per cent. ; carbon, 2*4 per cent, combined, 

 the metal being extremely hard, as before. 



Additional doses of red ore were added gradually without 

 agitating the bath, and the effect upon the fluid metal was 

 observed from time to time. 



At the end of the fifth hour, the samples taken from the fluid 

 bath assumed a decidedly mild temper ; when the addition of ore 

 was stopped, and exactly six hours after being charged, the metal 

 was tapped and run into ingots ; it now contained silicon, 0*046 

 per cent. ; carbon, '25 per cent., thus both the silicon and the 

 carbon had been almost entirely removed from the pig metal by 

 mere contact with metallic oxide under a protecting glass covering. 



The quantity of red ore added to the bath amounted to two 

 hundredweight, and the weight of metal tapped to ten hundred- 

 weight, five pounds, being slightly in excess of the weight of pig 

 metal charged. 



But the pig metal had contained 1/5 per cent, of silicon, and 

 4 per cent, of carbon, or a combined total of 5*5 per cent., 

 whereas the final metal contained collectively only *296 of silicon 

 and carbon, showing a gain of metal of 5'5 - '296 = 5*204 per cent., 

 or including the five pounds of increased weight, a total gain of 

 5*7 per cent, of metallic iron. 



Supported by these observations, I venture to assert that tJte 

 removal of the silicon and carbon from the pig iron in the ordinary 

 inuldlinij or " boiling " process is due entirely to the action of the 

 fluid ojcide of iron present, and thai an equivalent amount of metallic 

 iron is reduced and added to the bath, which gain, however, is 

 generally and unnecessarily lost again in the subsequent stages 

 of the process. The relative quantity of metal thus produced 

 from the fluid cinder admits of being accurately determined. 



The cinder may be taken to consist of Fe 3 0* (this being the 

 fusible combination of peroxide and protoxide), together with 

 more or less tribasic silicate (3FeO, SiO 3 ), which may be regarded 

 as a neutral admixture, not affecting the argument, and silicic 

 acid or silica is represented by Si O 3 , from which it follows that 

 for every four atoms of silicon leaving the metal, nine atoms of 

 VOL. i. n 



