These rest mainly on the external and vital relations of species to species in a state of nature--on what has been well termed by Semper the "physiology of organisms," rather than on the anatomy or physiology of organs.
* * * * * It has always been considered a weakness in Darwin's work that he based his theory, primarily, on the evidence of variation in domesticated animals and cultivated plants.

I have endeavoured to secure a firm foundation for the theory in the variations of organisms in a state of nature; and as the exact amount and precise character of these variations is of paramount importance in the numerous problems that arise when we apply the theory to explain the facts of nature, I have endeavoured, by means of a series of diagrams, to exhibit to the eye the actual variations as they are found to exist in a sufficient number of species.

By doing this, not only does the reader obtain a better and more precise idea of variation than can be given by any number of tabular statements or cases of extreme individual variation, but we obtain a basis of fact by which to test the statements and objections usually put forth on the subject of specific variability; and it will be found that, throughout the work, I have frequently to appeal to these diagrams and the facts they illustrate, just as Darwin was accustomed to appeal to the facts of variation among dogs and pigeons.
I have also made what appears to me an important change in the arrangement of the subject.

Instead of treating first the comparatively difficult and unfamiliar details of variation, I commence with the Struggle for Existence, which is really the fundamental phenomenon on which natural selection depends, while the particular facts which illustrate it are comparatively familiar and very interesting.

It has the further advantage that, after discussing variation and the effects of artificial selection, we proceed at once to explain how natural selection acts.
Among the subjects of novelty or interest discussed in this volume, and which have important bearings on the theory of natural selection, are: (1) A proof that all _specific_ characters are (or once have been) either useful in themselves or correlated with useful characters (Chap.
VI); (2) a proof that natural selection can, in certain cases, increase the sterility of crosses (Chap.