There is involved here, however, an assumption which is by no means self-evident, that every advancement gained by the variation in question constitutes a new centre for the variations occurring in the following generation. _That this is a fact_, is proved by such actual results of selection as are obtained in the case of the Japanese cock. But the question remains, Why is this the fact? Now here, I think, my theory of determinants gives a satisfactory answer. According to that theory every independently and hereditarily variable part is represented in the germ by a _determinant_, that is by a determinative group of vital units, whose size and power of assimilation correspond to the size and vigor of the part. These determinants multiply, as do all vital units, by growth and division, and necessarily they increase rapidly in every individual, and the more rapidly the greater the quantity of the germinal cells {37} the individual produces. And since there is no more reason for excluding irregularities of passive nutrition, and of the supply of nutriment in these minute, microscopically invisible parts, than there is in the larger visible parts of the cells, tissues, and organs, consequently the descendants of a determinant can never all be exactly alike in size and capacity of assimilation, but they will oscillate in this respect to and fro about the maternal determinant as about their zero-point, and will be partly greater, partly smaller, and partly of the same size as that. In these oscillations, now, the material for further selection is presented, and in the inevitable fluctuations of the nutrient supply I see the reason why every stage attained becomes immediately the zero-point of new fluctuations, and consequently why the size of a part can be augmented or diminished by selection without limit, solely by the displacement of the zero-point of variation as the result of selection.