50 



Transactio)is. 



This process gives very constant results, duplicate determinations 

 agreeing excellently. Thus, in estimating the iron in fifty-five samples of 

 limonite, the greatest difference in the duplicate determinations, which 

 were made quite independently of one another, was 0*22 per cent, of iron, 

 and the average difference only 0'09 per cent. In many of these deter- 

 minations, after titrating wdth permanganate, the solution was reduced 

 with sulphuretted hydrogen, the excess of this gas being removed by pass- 

 ing carbon dioxide through the boiling solution (Hillebrand's method). The 

 iron so reduced was then estimated by titration with permanganate, and 

 in every case the result so found agreed closely with that previously found 

 by our method. This agreement shows that the whole of the iron dissolved 

 out of the reduced ore goes into solution as a ferrous salt, and is therefore 

 estimated in the subsequent titration. Moreover, the residues contained 

 no iron. This was proved by fusing with carbonates of soda and potash 

 and testing for iron. 



We had thus conclusive proof of the accuracy of our results, Ijut for 

 the sake of comparison with other methods the following determinations 

 on five limonites were made. The results given under A are by our method ; 

 those under B were determined by dissolving in hydrochloric acid and 

 stannous chloride, adding mercuric chloride to remove the excess of stannous 

 chloride, phosphoric acid and manganese sulphate to improve the con- 

 ditions for titration, and finally titrating with permanganate. In column C 

 of the table the results were obtained by dissolving in hydrochloric acid^ 

 reducing with lead and copper sulphate, and titrating with potassium 

 bichromate. The results are percentages of iron. 



A. 



B. 



Difference 

 from A. 



Difference 

 from A. 



44-03 

 49-21 

 36-57 

 26-73 

 3.5-.52 



4.3-13 

 49-06 

 36-99 

 26-46 

 35-97 



-0-90 



— 0-15 

 + 0-42 



— 0-27 

 + 0-45 



43-42 

 49-29 

 36-66 

 26-78 

 36-02 



— 0-61 

 -f 0-08 

 + 0-09 

 + 0-05 

 + 0-50 



Mean difference — 0-09 



Mean difference + 0-02 



The titration in the case of B is not nearly so satisfactory as in A, owing 

 to the presence of a large amount of precipitate, which somewhat masks 

 the ending. Duplicates by this method differ to a much greater extent 

 than do those by the A method. 



In making determinations by method C the solution of the ore is very 

 slow, sometimes taking hours to accomplish, and, owing to the necessity 

 for an external indicator, the titration is not nearly so simple and satis- 

 factory as in the A method. 



Our method is thus shown to be well adapted for the estimation of iron 

 in limonites. It may also be used wdth advantage for the much less soluble 

 magnetites, ilmenites, htematites, &c. In this case the ore requires fine 

 grinding, and half an hour's reduction in hydrogen — coal-gas cannot be 

 used, as it does not give complete reduction. If the hydrogen be made 

 from commercial zinc, it should be purified from antimony and arsenic, 

 which interfere with the subsequent titration. The purification is most 

 simply effected by passing through a hard glass tube wrapped in thin sheet 



