stokes.] INFLUENCE OF IMPURITIES. 31 
(No. 20), Littmitz (No. 22), and Crow Branch mine (No. 23). In the 
latter two eases the selected material showed no pyrite under a lens, 
although the specimen as a whole contained it in visible amount. 
It further appears from the curve that small amounts of marcasite 
in pyrite can be determined with greater accuracy than small quanti- 
ties of pyrite in marcasite. In the former case the determination can 
probably be made to 1 per cent with duplicate experiments and in the 
absence of notable amounts of other sulphides; in the latter it can 
hardly be more accurate than 2 or 3 per cent. It must be remem- 
bered, however, that there is a slight uncertainty close to both ends 
of the curve, due to the unknown influence of minute contaminations. 
In other cases, however, greater accuracy than is obtainable could 
hardly be desired. 
VIII. INFLUENCE OF IMPURITIES. 
When pyrite or marcasite is inclosed in rock, the extraneous mate- 
rial may be removed by digesting with hydrofluoric and hydrochloric 
acids; bnl small amounts of insoluble siliceous material can usually be 
neglected. 
MISCELLANEOUS IMPURITIES. 
The necessity <»l* extreme care in preparing the material is empha- 
sized above. Since /;>(r — r/), an inspection of 1 1 1 * - ('([nation 
a333_fc_,, 
c—a 
shows thai the presence of iron salts or oxides, or access of air during 
the decomposition, affects the ratio - - in such a manner as to lower 
c—a 
the value of /;, this effect being greater with pyrite than with marca- 
site. Fvw sulphur exerts very Little influence, and is moreover 
removed by the preliminary extraction with ether. Without influ- 
ence arc also sulphates, carbonates, or soluble silicates free from iron, 
insoluble silicates, quartz, and in general any substance neither con- 
tributing iron to nor effecting reduction of the ferric solution. Pyr- 
rhotite and limonite are easily eliminated. Zinc blende, galena, mag- 
netite, and hematite, if present in but small amounts and not inclosed 
in the fine particles of FeS 2 , are more slowly removed by prolonged 
digestion with hydrochloric acid. All sulphides are more or less 
readily attacked, and t hose free from iron, since they do not contribute 
to c— a, should, it might be expected, raise the value of p. If present 
in traces, they do not appreciably affect the result, but if existing in 
large amounts, they introduce considerable errors, and abnormal 
results are to be explained by a qualitative analysis. In reality their 
influence can not be always predicted, as pointed out above (p. 30), 
and some experiments bearing on this point are given below. 
Bull. 186—01 3 
