If wild species, when crossed, give almost invariably intermediate forms, then it may appear that we are going against the only evidence that we can hope to obtain if we claim that discontinuous variation, of the kind that sports are made of, has supplied the material for evolution. If, furthermore, when distinct races of domesticated animals are crossed, we do not get discontinuous inheritance, it might, perhaps, with justness be claimed that this instance is paralleled by what takes place when wild species are crossed. And if domesticated forms have been largely the result of the selection of fluctuating variations, as Darwin believes, then a strong case is apparently made out in favor of Darwin’s view that continuous variation has given the material for the process of evolution in nature. Whether selection or some other factor has directed the formation of the new species would not, of course, be shown, nor would it make any difference in the present connection. Before we attempt to reach a conclusion on this point let us analyze the facts somewhat more closely. In the first place, a number of these cases of discontinuous variation are of the nature of abnormalities. The appearance of extra fingers or toes in man and other mammals is an example of this sort. This abnormality is, if inherited at all, inherited completely; that is, if present the extra digit is perfect, and never appears in an intermediate condition, even when one of the parents was without it. The most obvious interpretation of this fact is that when the material out of which the fingers are to develop is divided up, or separated into its component parts, one more part than usual is laid down. Similarly, when a flower belonging to the triradiate type gives rise to a quadriradiate form,—as sometimes occurs,—the new variation seems to depend simply on the material being subdivided once more than usual; perhaps because a little more of it is present, or because it has a somewhat different shape. My reasons for making a surmise of this sort are based on certain experimental facts in connection with the regeneration of animals. It has been shown in several cases that it is possible to produce more than the normal number of parts by simply dividing the material so that each part becomes more or less a new whole, and the total number of parts into which the material becomes subdivided is increased. It seems not improbable that phenomena of this sort have occurred in the course of evolution, although it is, of course, possible that those characters that define species do not belong to this class of variation. To take an example. There are nine neck-vertebræ in some birds, but in the swan the number is twenty-five. We cannot suppose that the ancestor of the swan gradually added enough materially to make up one new vertebra and then another, but at least one new whole vertebra was added at a time; and we know several cases in which the number of vertebræ in the neck has