direct environmental effects. Every systematist also is perfectly aware that in dealing with collections from little explored countries the occurrence of polymorphism or even of sporadic variation may make the practical business of distinguishing the species difficult and perhaps for the time impossible; still, conceding that a great part of the diversity is due to geographical differentiation, and that some is sporadic variation, our experience of our own floras and faunas encourages the belief that if we were thoroughly familiar with these exotic productions it would usually be possible to assign their specific limitations with an approach to certainty.

For apart from any question of the justice of these wider inferences, if we examine the phenomenon of specificity as it appears in those examples which are nearest to hand, surely we find signs in plenty that specific distinction is no mere consequence of Natural Selection. The strength of this proposition has lain mainly in the appeal to ignorance. Steadily with the growth of knowledge has its cogency diminished, and such a belief could only have been formulated at a time when the facts of variation were unknown.

In Darwin's time no serious attempt had been made to examine the manifestations of variability. A vast assemblage of miscellaneous facts could formerly be adduced as seemingly comparable illustrations of the phenomenon "Variation." Time has shown this mass of evidence to be capable of analysis. When first promulgated it produced the impression that variability was a phenomenon generally distributed amongst living things in such a way that the specific divisions must be arbitrary. When this variability is sorted out, and is seen to be in part a result of hybridisation, in part a consequence of the persistence of hybrids by parthenogenetic reproduction, a polymorphism due to the continued presence of individuals representing various combinations of Mendelian allelomorphs, partly also the transient effect of alteration in external circumstances, we see how cautious we must be in drawing inferences as to the indefiniteness of specific limits from a bare knowledge that intermediates exist. Conversely, from the accident of collocation or from a misleading resemblance in features we deem essential, forms genetically distinct are often confounded together, and thus the divergence of such forms in their other features, which we declare to be non-essential, passes as an example of variation. Lastly, and this is perhaps the most fertile of all the sources of confusion, the impression of the indefiniteness of species is created by the existence of numerous local forms, isolated geographically from each other, forms whose differences may be referable to any one of the categories I have enumerated.

The advance has been from many sides. Something has come from the work of systematists, something from cultural experiments, something from the direct study of variation as it appears in nature, but progress is especially due to experimental investigation of heredity. From all these lines of inquiry we get the same answer; that what the naturalists of fifty years ago regarded as variation is not one phenomenon but many, and that what they would have adduced as evidence against the definiteness of species may not in fact be capable of this construction at all.

If we may once more introduce a physical analogy, the distinctions with which the systematic naturalist is concerned in the study of living things are as multifarious as those by which chemists were confronted in the early days of their science. Diversities due to mechanical mixtures, to allotropy, to differences of temperature and pressure, or to degree of hydration, had all to be severally distinguished before the essential diversity due to variety of chemical constitution stood out clearly, and I surmise that not till a stricter analysis of the diversities of animals and plants has been made on a comprehensive scale, shall we be in a position to declare with any confidence whether there is or is not a natural and physiological distinction between species and variety.

As I have said above, it is in the cases nearest to hand that the problem may be most effectively studied. Comparison between forms from dissimilar situations contributes something; but it is by a close examination of the behaviour, especially the genetic behaviour, of familiar species when living in the presence of their nearest allies that the most direct light on the problem is to be obtained. I cannot understand the attitude of those who, contemplating such facts as this examination elicits, can complacently declare that specific difference is a mere question of degree. With the spread of evolutionary ideas to speak much of the fixity of species has become unfashionable, and yet how striking and inscrutable are the manifestations of that fixity!

Consider the group of species composing the agrestis section of the genus Veronica, namely Tournefortii, agrestis, and polita.

These three grow side by side in my garden, as they do in suitable situations over a vast area of the temperate regions. I have for years noticed them with some care and become familiar with their distinctions and resemblances. Never is there any real doubt as to the identity of any plant. The species show some variability, but I have never seen one which assumed any of the distinguishing features of the others. A glance at the fruits decides at once to which species a plant belongs. I find it impossible to believe that the fixity of these distinctions is directly dependent on their value as aids in the struggle for existence. The mode of existence of the three forms in so far as we can tell is closely similar. By whatever standard we reckon systematic affinity I suppose we shall agree that these species come very near indeed to each other. Bentham even takes the view that polita is a mere variety of agrestis.

Now in such cases as this it has been argued that the specific features of the several types have been separately developed in as many distinct localities, and that their present association is due to subsequent redistribution. Of these Veronicas indeed we know that one, Tournefortii (= Buxbaumii) is as a matter of fact a recent introduction from the east.[8] But this course of argument leads to still further difficulties. For if it is true that the peculiarities of the several species have been perfected and preserved on account of their survival-value to their possessors, it follows that there must be many ways of attaining the same result. But since sufficient adaptation may be ensured in so many ways, the disappearance of the common parent of these forms is difficult to understand. Obviously it must have been a plant very similar in general construction to its modern representatives. Like them it must have been an annual weed, with an organisation conformable to that mode of life. Why then, after having been duly perfected for that existence should it have been entirely superseded in favour of a number of other distinct contrivances for doing the same thing, and—if a gradual transition be predicated—not only by them, but by each intermediate stage between them and the original progenitor? Surely the obvious inference from such facts is that the burden cast upon the theory of gradual selection is far greater than it can bear; that adaptation is not in practice a very close fit, and that the distinctions between these several species of Veronica have not arisen on account of their