Since the form _abc_ occurs once in the series of eight components, it is consequently little likely that it would be missing among the experimental plants, even were these raised in a smaller number, and the transformation would be perfected already by a second fertilisation. If by chance it did not appear, then the fertilisation must be repeated with one of those forms nearest akin, _Aabc_, _aBbc_, _abCc_. It is perceived that such an experiment must extend the farther _the smaller the number of experimental plants and the larger the number of differentiating characters_ in the two original species; and that, furthermore, in the same species there can easily occur a delay of one or even of two generations such as Gärtner observed. The transformation of widely divergent species could generally only be completed in five or six years of experiment, since the number of different egg cells which are formed in the hybrid increases in square ratio with the number of differentiating characters. Gärtner found by repeated experiments that the respective period of transformation varies in many species, so that frequently a species _A_ can be transformed into a species _B_ a generation sooner than can species _B_ into species _A_. He deduces therefrom that Kölreuter’s opinion can hardly be maintained that “the two natures in hybrids are perfectly in equilibrium.” It appears, however, that Kölreuter does not merit this criticism, but that Gärtner rather has overlooked a material point, to which he himself elsewhere draws attention, viz. that “it depends which individual is chosen for further transformation.” Experiments which in this connection were carried out with two species of _Pisum_ demonstrated that as regards the choice of the fittest individuals for the purpose of further fertilisation it may make a great difference which of two species is transformed into the other. The two experimental plants differed in five characters, while at the same time those of species _A_ were all dominant and those of species _B_ all recessive. For mutual transformation _A_ was fertilised with pollen of _B_, and _B_ with pollen of _A_, and this was repeated with both hybrids the following year. With the first experiment _B_/_A_ there were eighty-seven plants available in the third year of experiment for the selections of individuals for further crossing, and these were of the possible thirty-two forms; with the second experiment _A_/_B_ seventy-three plants resulted, which _agreed throughout perfectly in habit with the pollen parent_; in their internal composition, however, they must have been just as varied as the forms of the other experiment. A definite selection was consequently only possible with the first experiment; with the second some plants selected at random had to be excluded. Of the latter only a portion of the flowers were crossed with the _A_ pollen, the others were left to fertilise themselves. Among each five plants which were