ut it also gives suck to its young, and in this feature it
differs from all fish, while it resembles all mammals. Now, looking at
those two features alone, should we say that a porpoise ought to be
classed as a fish or as a mammal? Assuredly as a mammal, and for this
reason: The number of teeth is a very variable feature both in fish and
in mammals, whereas the giving of suck is an invariable feature among
mammals, and occurs nowhere else in the animal kingdom. This, of course,
is purposely chosen as a very simple illustration; but it exemplifies
the general fact that the guiding principle of scientific
classification is the comparing of organism with organism, with the view
of seeing which of the constituent organs are of the most invariable
occurrence, and therefore of the most typical signification.

Now, since the days of Linnaeus this principle has been carefully
followed, and it is by its aid that the tree-like system of
classification has been established. No one, even long before Darwin's
days, ever dreamed of doubting that this system is in reality, what it
always has been in name, a _natural_ system. What, then, is the
inference we are to draw from it? An evolutionist answers, that it is
just such a system as his theory of descent would lead him to expect as
a natural system. For this tree-like system is as clear an expression as
anything could be of the fact that all species are bound together by the
ties of genetic relationship. If all species were separately created, it
is almost incredible that we should everywhere observe this progressive
shading off of characters common to larger groups, into more and more
specialized characters distinctive only of smaller and smaller groups.
At any rate, to say the least, the law of parsimony forbids us to
ascribe such effects to a supernatural cause, acting in so whimsical a
manner, when the effects are precisely what we should expect to follow
from the action of a highly probable natural cause. The classification
of animal forms, indeed, as Darwin, Lyell, and Haeckel have pointed out,
strongly resembles the classification of languages. In the case of
languages, as in the case of species, we have genetic affinities
strongly marked; so that it is possible to some extent to construct a
language-tree, the branches of which shall indicate, in a diagrammatic
form, the progressive divergence of a large group of languages from a
common stock. For instance, Latin may be regarde