In such cases isolation is usually accompanied by selection: rarely has the first of a double, striped or variegated race well filled or richly striped flowers or highly spotted leaves. Usually minor degrees of the anomaly are seen first, and the breeder expects the novelty to develop its features more completely and more beautifully in subsequent generations. Some varieties need selection only in the beginning, in others the most perfect specimens must be chosen every year as seed-bearers. For striped flowers, it has been prescribed by Vilmorin, that seeds should be taken only from those with the smallest stripes, because there is always reversion. Mixed seed or seed from medium types would soon yield plants with too broad stripes, and therefore less diversified flowers. In horticulture, new varieties, both retrograde and ever-sporting, are known to occur almost yearly. Nevertheless, not every novelty of the gardener is to be considered as a mutation in the scientific sense of the word. First of all, the novelties of perennial and woody species are to be excluded. Any extreme case of fluctuating variability may be preserved and multiplied in the vegetative way. Such types are designated [608] in horticulture as varieties, though obviously they are of quite another nature than the varieties reproduced by seed. Secondly, a large number, no doubt the greater number of novelties, are of hybrid origin. Here we may discern two cases. Hybrids may be produced by the crossing of old types, either of two old cultivated forms or newly introduced species, or ordinarily between an old and an introduced variety. Such novelties are excluded from our present discussion. Secondly, hybrids may be produced between a true, new mutation and some of the already existing varieties of the same species. Examples of this obvious and usual practice will be given further on, but it must be pointed out now that by such crosses a single mutation may produce as many novelties as there are available varieties of the same species. Summarizing these introductory remarks we must lay stress on the fact that only a small part of the horticultural novelties are real mutations, although they do occur from time to time. If useful, they are as a rule isolated and multiplied, and if necessary, improved by selection. They are in many instances, as constant from seed as the unavoidable influence of vicinism allows them to be. Exact observations on the origin, or on the degree of constancy, are usually lacking, [609] the notes being ordinarily made for commercial purposes, and often only at the date of introduction into trade, when the preceding stages of the novelty may have been partly forgotten.