468 Mr. E. Matthey. [Jan. 26, 



bismuth in contact with metallic iron at a dull-red heat and under 

 flnx A compound of iron and arsenic was thus formed and could be 

 removed as a scum; the disadvantages of this process being, 1st, 

 loss of bismuth by volatilisation, and, 2ndly,much loss of time m the 

 manipulation of any large quantity to be treated. 



Having occasion, a few months ago, to melt together a large quan- 

 tity of arsenical bismuth, some 700 or 800 kilos., that is, more than 

 three quarters of a ton, in order to obtain a homogeneous alloy upon 

 which to work subsequently by the process above alluded to, it be- 

 came evident that when the temperature was raised above the melt 

 ing point of bismuth, the surface of the metal being exposed to the 

 atmosphere, arsenical fumes appeared, and that these increased as 

 the temperature of the metal became more elevated, the result being 

 that the arsenic came off in dense white fumes (As 2 3 ). 



The observation of this fact led to further experiments, and it was 

 found that if the surface of the bath of fused arsenical bismuth was 

 freely exposed to the air at a temperature rather higher than its own 

 melting point, and if the molten metal was constantly stirred, it was 

 possible to eliminate the whole of the arsenic alloyed with the bismuth 

 by this simple process of fusion with stirring. 



As it is a matter of considerable interest for metallurgists to know,, 

 not only that this elimination does take place, but also at what tem- 

 perature it occurs, a series of experiments have been conducted with 

 a view of determining this accurately. 



The work of Roberts- Austen has shown that a thermo-junction is 

 practically the only form of pyrometer that can be used for delicate 

 thermal investigations of this kind, but the question arose which par- 

 ticular thermo-junction should be adopted. Was it well to use the 

 platinum-iridium one as advocated by Barus^ or the platinum-rhodium 

 one suggested by H. Le Chatelier ? My previous work on the alloys 

 of platinum and rhodium, lately published in the 'Philosophical 

 Transactions,' settled the question in favour of the rhodium-platinum 

 thenno- junction, for I was satisfied that the alloy of platinum, with 

 10 per cent, of rhodium is as homogeneous as any known alloy could 

 well be, and is therefore admirably adapted for use as a thermo- 

 junction, pure platinum being the opposing metal. 



The diagram shows the results obtained by calibrating the thermo- 

 jnnction by the aid of certain known melting points. The tempera- 

 ture at which arsenic is eliminated is also shown on the diagram. 



Analysis proved that the alloy operated upon in these experiments 

 contained 0'65 per cent, of metallic arsenic. 



From the diagram it will be seen that the melting point of this 

 alloy of bismuth is 278 C. By raising the temperature of the 

 alloy to 395 C. the arsenic freely sublimates from tne bismuth alloy, 

 and at a temperature of 513 C. the whole of the arsenic is eliminated. 



