Correspondence with Charles Darwin 171
of the patent and the latent elements selection took place, so that not only are the somatic elements a selection of all possible somatic elements of an individual of the same ancestry, but the latent elements or germ-plasm were themselves a selection. This selection he termed `class representation.' That the somatic or bodily characters are a selection is, of course, obvious ; that the germ-plasm is selected also is extremely probable, but less easily demonstrated. Galton represented to himself the 'structureless elements' as a vast congeries of individual elements-like balls of a great variety of colours in a bag. A selection is made of these ('class representation') for the embryonic elements which by development become the adult elements, the somatic characters; that is the simple explanation of variation in the somatic characters of individuals of the same ancestry and reared under the same environment. Another selection from the same bag gives the germ-plasm of the individual on which his gametic characters depend, i.e. the possibilities of his descendants. Thus the continuity of the `latent elements' or as we might say of the germ-plasm was in Galton's mind broken by continual selection. The `class representation' of the somatic characters giving the phenomenon of visible variation, and the `class representation' of the=-germplasm the variation of stocks or stirps.
Galton did not in this paper, I do not think he ever did, carry out his hypotheses to their legitimate conclusions. In the first place the two selections from our `bag' cannot be treated as wholly independent; the somatic characters are not perfectly correlated with the gametic characters, but they are correlated with them, and as we descend to highly specialised races highly correlated with them. It would not be unreasonable to suppose that the somatic characters arise from a sub-selection of the gametic group, or from leaving a portion of this drawing 'on the table.' But the selection of the germ-plasm must lead to its simpler and simpler structure, especially in the case of unisexual reproduction. The course of evolution must on this hypothesis start with a highly complex germ-plasm and tend to break this up into simpler and simpler groups as generation by generation more elements are differentiated, i.e. organism differs from organism by having fewer and fewer common latent elements. We should see genera breaking up into species, species into local races, and ultimately races into stirps and possibly stirps into the merely ideal `pure lines,' or organisms in the ease of which it would be impossible to carry germ-plasm selection further for it would have become of one type only; the innumerable balls of immense variety in our bag would have been reduced to a single colour!
Darwin's natural selection acts only on evolution through the definite correlation of somatic and gametic characters. Galton's germinal selection, a random selection at the output of each new individual, must-if there be isolation-tend to produce species, races and sub-races. A pure race could only be one in which all latent elements were so substantially represented that there was little chance of a ' class representation' excluding any of them'. This is not the place to discuss at length the bearing of Galton's
' Purity of race might also be preserved by much intra-racial crossing.