f the earth is not to be
confuted by this argument alone, That there are, among natural bodies,
certain appearances which are not explained by the theory. We must
admit, that, not having all the data which natural philosophy requires,
we cannot pretend to explain every thing which appears; and that our
theories, which necessarily are imperfect, are not to be considered as
erroneous when not explaining every thing which is in nature, but only
when they are found contrary to or inconsistent with the laws of nature,
which are known, and with which the case in question may be properly
compared.

But we may have different theories to compare with nature; and, in that
case, the question is not, How far any of those theories should explain
all natural appearances? but, How far any one particular theory might
explain a phenomenon better than another? In this case of comparison, it
will be evident, that if one theory explains natural appearances, then
the opposite to that theory cannot be supposed to explain the same
appearances. If for example, granite, porphyry, or basaltes, should be
found naturally formed by fusion, the formation of those stones could
not be supposed in any case as formed by water, although it could not
be demonstrated that water is incapable of forming those mineral
productions.

In like manner, if those three bodies were proved to have been actually
formed by water alone, then, in other cases where we should have no
proof, they could not be supposed as having been formed by fire or
fusion. It must be evident, that an equal degree of proof of those two
different propositions would leave our judgment in suspence, unless that
proof were perfect, in which case, we would have two different causes
producing similar effects. But, if we shall have a sufficient proof
upon the one side, and only a presumptive proof or probability upon the
other, we must reject that probability or presumption, when opposed by
a proof, although that proof were only an induction by reasoning from
similar effects as following similar causes. _A fortiori_, if there be
on one side a fair induction, without the least suspicion of error,
and on the other nothing but a mere presumption founded upon a distant
analogy, which could not even properly apply, then, the inductive proof
would be as satisfactory as if there had not been any supposition on the
opposite side.

So far as a theory is formed in the generalization of natural
appearance