tal and present
it with an outfit of _heterogeneous_ constituents. Try, therefore, the
effect of such a gift; fling into the pre-existing cauldron the whole
list of recognized elementary substances, and give leave to their
affinities to work." The intended implication obviously is, that there
must exist the separately-created elements before evolution can begin.

Here, however, Mr. Martineau makes an assumption which few, if any,
chemists will commit themselves to, and which many will distinctly deny.
There are no "recognized elementary substances," if the expression means
substances known to be elementary. What chemists, for convenience, call
elementary substances, are merely substances which they have thus far
failed to decompose; but, bearing in mind past experiences, they do not
dare to say that they are absolutely undecomposable. Water was taken to
be an element for more than two thousand years, and then was proved to
be a compound; and, until Davy brought a galvanic current to bear upon
them, the alkalies and the earths were supposed to be elements. So
little true is it that "recognized elementary substances" are supposed
to be absolutely elementary, that there has been much speculation among
chemists respecting the process of compounding and recompounding by
which they have been formed out of some ultimate substance--some
chemists having supposed the atom of hydrogen to be the unit of
composition, but others having contended that the atomic weights of the
so-called elements are not thus interpretable. If I remember rightly,
Sir John Herschel was one, among others, who, some five-and-twenty years
ago, threw out suggestions respecting a system of compounding that might
explain these relations of the atomic weights.

What was at that time a suspicion has now become practically a
certainty. Spectrum-analysis yields results wholly irreconcilable with
the assumption that the conventionally-named simple substances are
really simple. Each yields a spectrum having lines varying in number
from two to eighty or more, every one of which implies the intercepting
of ethereal undulations of a certain order by something oscillating in
unison or in harmony with them. Were iron absolutely elementary, it is
not conceivable that its atom could intercept ethereal undulations of
eighty different orders. Though it does not follow that its molecule
contains as many separate atoms as there are lines in its spectrum, it
must clearly be a