We have assumed passive fluctuations in nutrition to be the first cause in individual variation, and it is obvious that the action of this first cause of dissimilarity must be greatly restricted by the multiplicity of the ids and the corresponding homologous determinants. For although passive fluctuations in nutrition should occur continually in the case of all determinants, this would not imply that they would follow the same direction in all the determinants _X_ of all ids, for some determinants _X_ might vary upwards, and others downwards, and these might counteract each other in ontogeny; so that in many cases the fluctuations of the individual determinants will not be felt in their products at all. But since there are--as we shall see later--only two directions of variation, upwards and downwards, plus and minus, it must also sometimes happen that a majority take one direction, and this affords the basis on which germinal selection can build further, and on which it is materially supported by reducing division and the subsequent amphimixis. For reducing division removes half of the ids and thus of the determinants from the mature germ-cell, and according as chance leaves together or separates a majority of _X_-determinants varying in the same direction, this particular germ-cell will contain the primary constituents of a plus- or of a minus-variation of _X_, and it is possible that the presence of a majority or a minority may be entirely due to the reduction. The germ-plasm of the parent may contain, for instance, the determinant _X_ in its twenty ids 12 times in minus-variation form, 8 times in plus-variation form; and the reducing division, according to our view, may separate these into two groups of which one contains eight plus- and two minus-variations, the other ten minus-variations, or the one six plus- and four minus-variations, the other two plus- and eight minus-variations, and so on. Now every germ-cell which contains a majority of plus- or minus-variations--and this must be the case with most of them--may unite, if it attains to amphimixis, with a germ-cell which also contains a majority of plus or minus _X_-determinants, and if similar majorities let us say plus--meet together, the plus-variation of _X_ must be all the more sharply emphasized in the child. Thus, although the individual determinants _X_ may not be incited to further variation by their co-operation with others varying in the same direction, the collective effect of the plus-determinants will be greater, and adherence to the same direction of variation in the following generation will be assured, for if in the germ-plasm of the parent there be, for instance, sixteen out of twenty determinants possessing the plus-variation, a minus-majority can no longer result from reducing division.