or of a gland, or of a nerve-cell, is to be communicated to the germ, and not to the whole germ-plasm, but only to so much of it as is necessary to cause variation in the corresponding group of cells in the child. It may perhaps be said that this still does not necessitate the assumption of special determinants for these cell-groups, and that one might, with Herbert Spencer, conceive of the germ-plasm as consisting of homogeneous units which vary in the development in accordance with the diverse regularly alternating influences to which it is exposed from step to step, and that, therefore, in each of these units of very complex structure only a single molecule, or perhaps only a single atom, would need to vary in order that, in the course of development, the resulting cell-group should appear in the rudiment in somewhat altered strength. But I do not believe that a chemical molecule, still less an atom, is sufficient for this, for reasons which I have already stated--yet we need not go into this now, but rather deduce the consequences of this admission. It follows that the 'unit' is made up of numerous 'molecules' or 'atoms,' of which each, by dint of changes it has undergone, causes particular parts of the body to vary in a definite manner; in other words, we have here again a theory of determinants, only they are on a much smaller scale, since each invisible little vital particle or 'unit' contains all the determinants within itself, while in my theory it is only the id, that is, the visible chromosome, which includes the determinant complex. Such a theory would be far from a simplification of mine, it would rather complicate it enormously, and that without anything being gained. At most it would be made more evident how inconceivably complex the nerve-paths must be which lead from the part that has been modified by exercise to the germ-plasm, and must also lead to all the innumerable 'molecules or atoms' of the individual 'units.' But even on my theory of the composition of the ids as aggregates of living determinants, such nervous transmission of qualities would be a monstrosity which no one would accept, and I think on this account that my argument as to the impossibility of conceiving of the transmission of the modifications of the personal part to the germinal part retains its force, notwithstanding Hering's interesting analogy. If the transmission of functional modifications were an indisputable fact, I repeat, we should have to give in, and then we might regard the 'memory of organized material' as affording a hint of the possibility of the unimaginable process. But as long as the occurrence of this transmission cannot be proved either directly or indirectly, such a vague possibility of explanation need not induce us to assume an unproved process. LECTURE XXV GERMINAL SELECTION