It should be noted also that the germ-plasm is certainly not the immutable substance that the hypothesis originally postulated. Changes in the outer physical environment may certainly affect it; thus the larvæ bred from animals which live in abnormal physical conditions (temperature, moisture, etc.) may differ morphologically from the larvæ bred from animals belonging to the same species but living in a normal environment. The latter must therefore react on the germ-plasm, but the environment formed by the bodily tissues which surround the germ-cells may also so react: thus the germ-cells may be affected by such bodily changes as differences in the supply of nutritive matter, for instance. The offspring may deviate from the parental structure as the result of structural modifications acquired by the parent during its own lifetime, and, even if the filial deviation were not of the same nature as the parental modification, its inheritance would be an adequate cause of _some_ degree of transmutation. It is, however, certainly difficult to prove that organisms transmit to their progeny the _same_ kinds of deviation from the specific structure that they themselves acquire as the result of the action of the environment. Even if they did transmit such acquired deviations, it does not seem clear that this kind of inheritance alone would be a sufficient cause of the diversity of forms of life that we do actually observe in nature. Change of morphology would indeed occur, but we should expect to find insensible gradations of form and not individualised species. Let us suppose that Lamarckian inheritance acts for a considerable time on two or three originally distinct species inhabiting an isolated tract of land, and let us suppose that we investigate the variations occurring among all the organisms which are accessible to our observation with respect to some one variable character. The diagram _A_ represents what would seem to be the result of this process of transmutation. The numbers along the horizontal line are proportional to their distance from o, the origin, and represent the magnitude of the variation considered; and the height of the vertical lines represents the number of organisms exhibiting each degree of variation. We should expect to find that all the variations were equally frequent in their occurrence, but this is not what a study of variability in such a case as we have supposed--that of the animals inhabiting an isolated part of land--does actually indicate. What we should find would be the conditions represented by the diagram _B_. There would be two or more _modes_, that is, values of the variable character which are represented by a greater number of individuals than any other value of the variation. The environmental conditions _favour_ the individuals displaying this variation to a greater extent than they favour the rest. [Illustration: FIG. 24.]