SOSMAN AND HOSTETTERI REDUCTION OF IRON OXIDES 295 



tube, but that there is no approach to a complete reabsorption 

 of the oxygen. This decrease of pressure on cooling probably 

 represents the re-oxidation of a small amount of dissociated mag- 



iO 40 60 80 lOO -ttd 140 160 ISO ZOO 220 2+0 260 2BQ 



TIME IN MINUTES 



Fig. 1. Evolution of oxygen from magnetite heated in platinum (see Table I). 



netite; the dissociation product is as yet unknown. No metallic 

 iron was found in the charge either by solution in hydro- 

 chloric acid, which yields bubbles of hydrogen if metallic iron is 

 present, or by the more sensitive copper sulphate test under the 

 microscope. 



TABLE II 



Charge and crucible 



Crucible alone 



Charge alone 



INITIAL WEIGHT 



mg. 



4265.8 

 3261.4 

 1004.4 



FINAL WEIGHT 



mg. 



4264.7 

 3263.2 

 1001.5 



CHANGE IN 

 WEIGHT 



mg. 



-1.1 (loss) 

 + 1.8 (gain) 

 -2.9 (loss) 



The gain in weight of the crucible, shown in Table II, has been 

 caused partly by the absorption of metallic iron from the oxide, 

 and partly by the absorption of rhodium and also platinum- 



2 The analysis of the crucibles shows that the combined weight of rhodium 

 and iron is less than the total gain in weight. No appreciable amounts of other 

 metals seem likely to be present, nor were any found by the analysis. Heating a 

 part of crucible No. 1 in hydrogen for an hour at about 600° produced no 

 measurable loss of weight, showing that no oxide of platinum or iron is present. 

 Some platinum must therefore have been carried from the furnace tube to the 

 crucible, either on account of a temperature gradient between furnace and 

 crucible, or by some disintegrating effect due to the current, or by both causes 

 combined. 



