Still the two are not solely a result of observation, deduced from paleontological data; they are also _a consequence of the theory of selection_, as was shown in the text. If the theory in its previous form was unable to fulfil this requirement, it is certainly now able to do so after germinal selection has been added, and it is not in any sense necessary to assume a difference of _character_ between phylogenetic and ontogenetic variations. Bateson and Scott are wrong in imagining that I ask them "to abrogate reason" in pronouncing the "omnipotence of natural selection." On the contrary, the theory seems to me to accord so perfectly with the facts that we might, by reversing the process, actually construct the facts from the theory. What other than the actual conditions could be expected, if it is a fact that selection favors only the useful variations and singles them out from the rest by producing them in {74} increasing distinctness and volume with every generation, and also in an increasing number of individuals? The mere displacement of the zero-point of useful variations alone must produce this effect, especially when it is supported by germinal selection. It is impossible, indeed, to see how considerable, that is perceptible, deviations could arise at all on the path of phyletic development if in each generation a large number of individuals always possessed the useful, that is, the phyletic variations? In fact, by the assumption itself, the difference between useful and less useful variations is merely one of degree, and that a slight one.