[Illustration: DIAGRAM SHOWING THE GENETIC RELATIONS OF THE CARBON
MINERALS.]

We may cut this triangle of residual products where we please, and by
careful analysis determine accurately the chemical composition of a
section at this point, and we may please ourselves with the illusion, as
many chemists have done, that the definite proportions found represent
the formula of a specific compound; but an adjacent section above or
below would show a different composition, and so in the entire triangle
we should find an infinite series of formulae, or rather no constant
formulae at all. We should also find that the slice, taken at any point
while lying in the laboratory or undergoing chemical treatment, would
change in composition, and become a different substance.

In the same way we can snatch a brand from the fire at any stage of its
decomposition, or analyze a decaying tree trunk during any month of its
existence, and thus manufacture as many chemical formulae as we like,
and give them specific names; but it is evident that this is child's
play, not science. The truth is, the slowly decomposing tissue of the
plants of past ages has given us a series of phases which we have
grouped under distinct names, and we have called one group peat, one
lignite, another coal, another anthracite, and another graphite. We have
spaced off the scale, and called all within certain lines by a common
name; but this does not give us a common composition for all the
material within these lines. Hence we see that any effort to define or
describe coal, lignite, or anthracite accurately must be a failure,
because neither has a fixed composition, neither is a distinct
substance, but simply a conventional group of substances which form part
of an infinite and indivisible series.