I wrote in the last chapter of the recognition of that degree of "likeness" or kinship in animals and plants which we point to by the word "species," and of the grouping of several similar species to form a "genus," and of several genera to form a family, of families to form orders, and of orders to form classes–and of the giving of names to all these groups. Whilst the making of this or that lot of species into a distinct _genus_, and giving it a new name is a mere matter of convenience for the indication of more or less important agreements and divergences, and is to a large extent arbitrary or an expression of opinion–it has always been recognized among naturalists that the group called "a species" is not a mere convention, but has a real natural limitation. It is true that the actual things which we see in studying natural history are so many units or individuals. But the possibility of arranging these by pattern, colour and shape into ultimate companies of which all the units are alike and differ from all the units of another company, has been regarded as a natural fact of primary importance and not a mere convention or convenience. The conception of the "naturalness" of a species depends really upon a further qualification of great importance as to what we naturalists understand by it.
We find by rearing plants from seed and by causing animals to breed under actual observation that the individuals of a species pair with one another, and not with individuals of other species, and further, that the young which they produce are like the parents–show themselves, in fact, to be of the same "species." The species continually year after year reproduces itself with little variation, though some variation does occur. The faculty of pairing only within the group, of never naturally breeding with members of other groups, has accordingly been adopted as a test of species. Hybrids between two species do not occur, except in very rare cases, in the state of nature. It is not always the case that the members of two species cannot possibly pair together, but it is the fact that they do not do so. Man sometimes brings about such crossing or hybridization, and it is a curious fact that the hybrids are often infertile or give rise only to weakly offspring, which could not survive in the natural struggle for existence. Sometimes, however, when the two hybridized species happen to come from regions of the world remote from one another, the resulting hybrids establish a vigorous race. There are real obstacles (of which I will say more below) in natural conditions to hybrid-breeding between any two species which occur naturally in the same territory. Thus the idea of a species is expanded so as to be not merely "a group of individuals of constant likeness in form and characteristics," but we add to that definition a living or constitutional quality expressed by the words, "which produce fertile offspring by pairing with one another, but do not pair with the members of other species."