Cases of apparent reversion were often seen in nurseries, especially in flower culture, which under ordinary circumstances are rarely wholly pure, but always sport more or less into the colors and forms of allied varieties. Such sporting individuals have to be extirpated regularly, otherwise the whole variety would soon lose its type and its uniformity and run over to some other form in cultivation in the vicinity. For this reason atavism in nurseries causes much care and labor, and consequently is to be dealt with as a very important factor. From time to time the idea has suggested itself to some of the best authorities on the amelioration of plants, that this atavism was not due to an innate tendency, but, in many cases at least, was produced by crosses between neighboring varieties. It is especially owing to Verlot that this side of the question was brought forward. But breeders as a rule have not attached much importance to this supposition, chiefly because of the great practical difficulties attending any attempt to guard the species of the larger cultures against intermixture with other varieties. Bees and humble-bees fly from bud to bud, and carry the pollen from one [187 ] sort to another, and separation by great distances would be required to avoid this source of impurity. Unfortunately the arrangements and necessities of large cultures make it impossible to isolate the allied varieties from each other. From a theoretical point of view the origin of these impurities is a highly important question. If the breeders' atavism is due to crosses, and only to this cause, it has no bearing at all on the question of the constancy of varieties. And the general belief, that varieties are distinguished from true species by their repeated reversion and that even such reversibility is the real distinction of a variety, would not hold. For this reason I have taken much trouble in ascertaining the circumstances which attend this form of atavism. I have visited a number of the leading nurseries of Europe, tested their products in various ways, and made some experiments on the unavoidable conditions of hybridizing and on their effect on the ensuing generations. These investigations have led me to the conclusion, that atavism, as it is generally described, always or nearly always is due to hybridization, and therefore it is to be considered as untrue or false atavism.