It is an often recurring assertion that white varieties of colored species are the most stable of all horticultural races. They are often said to be at least as constant as the species itself, and even to surpass it in this quality. With our present state of knowledge, the explanation of this general experience is easily given. For selection removes the effect of spontaneous crosses from the variety in each year, and renders it practically pure, while it is wholly inadequate to produce the same effects on the species, because of the concealed hybrids. The explanation given in this simple instance may be applied to the case of different varieties of the same species, when growing together and crossed naturally by insects. It would take too long to go into all the details that present themselves here to the student of nature and of gardens. I will only state, that since varieties differ principally from their species by the lack of some sharp character, one variety may be characterized by the lack of color of the flowers, another by the lack of pubescence, a third by being dwarfed, and so on. Every character must be studied separately in its effects on the offspring [213] of the crosses. And it is therefore easily seen, that the hybrids of two varieties may resemble neither of them, but revert to the species itself. This is necessarily and commonly the case, since it is always the older or positive characters that prevail in the hybrids and the younger or negative that lie hidden. So for instance, a blue dwarf larkspur, crossed with a tall white variety, must give a tall blue hybrid, reassuming in both characters the essentials of the species. Keeping this rule in view, it will be easy to calculate what may be expected from spontaneous crosses for a wide range of occurrences, and thus to find an explanation of innumerable cases of apparent variability and reversion in the principle of vicinism. Students have only to recollect that specific characters prevail over varietal ones, and that every character competes only with its own antagonist. Or to give a sharper distinction: whiteness of flowers cannot be expected to be interchanged with pubescence of leaves.