precarious.' Or, as I have previously presented this formal aspect of
the matter while discussing the teleological argument with Professor Asa
Gray,--'I suppose it will be admitted that the validity of an inference
depends upon the number, the importance, and the definiteness of the
things or ratios known, as compared with the number, importance, and
definiteness of the things or ratios unknown, but inferred. If so, we
should be logically cautious in drawing inferences from the natural to
the supernatural: for although we have the entire sphere of experience
from which to draw an inference, we are unable to gauge the probability
of the inference when drawn--the unknown ratios being confessedly of
unknown number, importance, and degree of definiteness: the whole orbit
of human knowledge is insufficient to obtain a parallax whereby to
institute the required measurements or to determine the proportion
between the terms known and the terms unknown. Otherwise phrased, we may
say--as our knowledge of a part is to our knowledge of a whole, so is
our inference from that part to the reality of that whole. Who,
therefore, can say, even upon the hypothesis of Theism, that our
inferences or "idea of design" would have any meaning if applied to the
"All-Upholder," whose thoughts are not as our thoughts?'[30] And of
course, _mutatis mutandis_, the same remarks apply to all inferences
having a negative tendency.

As an outcome of the whole of this discussion, then, I think it appears
that the influence of Science upon Natural Religion has been uniformly
of a destructive character. Step by step it has driven back the apparent
evidence of direct or special design in Nature, until now this evidence
resides exclusively in the one great and general fact that Nature as a
whole is a Cosmos. Further than this it is obviously impossible that the
destructive influence of Science can extend, because Science can only
exist upon the basis of this fact. But when we allow that this great and