id produced; but in the case of the sample which led me to
seek for another method, this way could not be employed, for the specimen
had been taken from the surface, and was covered and penetrated by
vegetable growths which could not be entirely removed mechanically. Add
to this the fact of the presence of iron pyrites and the probable
occurrence of carbonates in the rock, and it will be at once seen that no
reliance could be placed on the results obtained by this suggested
method.

As the problem thus resolved itself into finding a way by which all
interfering substances could be destroyed without affecting the graphite,
it at once occurred to me to try the effect of caustic potash. I melted a
few pieces of potash in a silver crucible until it had stopped spitting
and was in quiet fusion. I then transferred the weighed sample to the
crucible, the melted potash in which readily wetted the graphite rock.
The mass was then gently heated, and occasionally stirred with a piece of
silver wire. The heat never need be much above the melting point of the
potash, though toward the last I have been in the habit of raising the
temperature slightly, to insure the complete decomposition of the melt.
When the decomposition is complete, which can be known by the complete
absence of gritty particles, the crucible is cooled and then soaked out
in cold water. This is very quickly accomplished, and we then see that we
have an insoluble residue of graphite and a flocculent precipitate of
lime, magnesia, iron hydrate, etc., while the organic matters have
disappeared. The sulphides of iron, etc., have given up their sulphur to
the potash, and everything except the graphite has suffered some change.
The solution is now filtered through a weighed Gooch crucible, the
residue washed a few times with water, and then treated with dilute
hydrochloric acid (followed by ammonia to remove any silver taken up from
the crucible), which will dissolve all the constituents of the residue
except the graphite, and after washing will leave the latter free and in
a condition of great purity.

As evidence of the accuracy of the method, I subjoin the results I
obtained on a sample whose gangue was free from all organic and other
impurities, consisting chiefly of quartz:

New Method. Combustion in Oxygen, Weighing CO_{2}.
15.51 15.54

It is plain that such a result leaves nothing to be desired for the
accuracy of the method, wh