The chances of the transmission of the saltatory variation will be greater or less according to whether the variation of the relevant determinants involves a bare majority of ids or a large majority, for the more ids that have varied, the greater is the probability that the majority will be maintained throughout the course of ensuing reducing divisions and amphimixis, that is, that the seeds of the plant will reproduce the variation, and will not revert to the ancestral form. Although one of the most satisfactory results of the id-theory lies precisely in the interpretation of these conditions, I do not wish to enter into the matter here, but will refer to the details in my _Germ-plasm_, published in 1894, which I consider valid still. At that time I had not formulated the idea of germinal selection, but the explanation of the occurrence of such sport-variations which I gave was based upon the assumption of nutritive fluctuations in the germ-plasm, which gave rise to variations in certain determinants. There was still lacking the recognition that the direction of variation once taken must be adhered to until resistance was met with, and that the determinants stand in nutritive correlation with one another, so that changes in one determinant must re-act upon the neighbouring ones, as I shall explain more fully afterwards. I also showed from definite cases that such sports, though they are sudden--'saltatory'--in their mode of occurrence, are long being prepared for by intimate processes in the germ-plasm. This 'invisible prelude' of variation depends on germinal selection. When a wild plant is sown in garden-ground it does not require to vary at once; several, even many, generations may succeed each other which show no sports; suddenly, however, sports appear, at first singly, then, perhaps, in considerable numbers. It is not, however, by any means always the case that considerable numbers occur, for some varieties of our garden flowers have arisen only once, and then have been propagated by seed; and such saltatory sports in plants which are raised from seed are usually constant in their seed, and if they are fertilized with their own pollen they breed true--a proof that the same variations must have taken place in the relevant determinants in a large majority of ids.