BY ]). MAWSON. 595 



material in which inclusions were made apparent during the 

 operation of crushing, a very pure sample was obtained. 



Several preliminary analyses of different samples showed, by 

 their eccentricities, the material first selected to be slightly con- 

 taminated. This led to a more stringent selection from the best 

 obtainable veinstuff for the final analysis quoted above. 



The results of the several preliminary partial analyses were of 

 special value in helping to draw up a formula for the mineral, 

 and so will enter into the following discussion. 



Silica : Besides that quoted in the analysis, the two following 

 were very accordant, 49 - 05 °/ Q , 50 - 00 °/ o . Titania almost quite 

 constant. Alumina : In one case a few centigrammes of the 

 sesquioxides precipitated by ammonium hydrate appeared to be 

 alumina when calculated by difference; more often though, the 

 ferric oxide, calculated from the total iron as determined on the 

 reduced solution which had already served for ferrous iron deter- 

 minations, was more than sufficient to account for the whole 

 sesquioxide precipitate. Ferrous and ferric iron were determined 

 quite a number of times. The earlier determinations varied 

 considerably; later on, with purer material, the following results 

 were obtained on different samples: FeO = 25-39%, 27 72%, 

 26-51 %. In order to determine this pi-oportion as accurately as 

 possible, a single crystal, after powdering, was digested with' 

 fairly concentrated H„S0 4 for five minutes, which, it was 

 expected, would help to free it from foreign matter, and yet not 

 appreciably affect the silicate itself. The washed and dried 

 mineral gave 2o - 50 % FeO which, calculating from the total iron 

 determined in the same sample, leaves 8-25 % Fe._,0 3 . The ratio 

 of ferrous to ferric iron thus appears to be FeO:Feo0 3 ::26 50:8-25. 

 j\/avj/anous oxide : Two estimations gave 109 °/ , 1-30 °/ Q . Lime : 

 2-30 %, 1-70 %, 1*48 %, rather variable. Magnesia appeared 

 as a minute trace in one only of the analyses. Soda -and 

 Potash : The importance of an accurate determination of 

 the alkalies merited an unusually careful estimation, the 

 results of blank tests on the reagents being duly regarded. 

 Lithia: Two estimations gave 0-60 7, 0-70 7. Fluorine though 



