I am making arrangements for trying whether there are any degrees of sterility to be found between well-marked and constant varieties of plants. But as I have never done anything in the way of hybridising, perhaps you would be good enough to let me know whether the enclosed plan of experimenting represents the full and proper way of going to work. I know that you do not believe in the object of it, but, even supposing it to be a wild goose chase, there would be no harm in your telling me the best way to run. Then, whether the results prove positive or negative, it will not be open for any one to doubt them on the ground of any fault in the method.
Do any objections occur to you re my answer to critics in the 'Nineteenth Century'? Of course I might have said more about the swamping effects of free intercrossing (which appears to me the only point in which I deviate at all from the 'Origin of Species'), but it is much too large a subject to be dealt with in a review. My greatest difficulty here is to conceive the possibility of differentiation (as distinguished from transmutation in linear series) without the assistance of isolation in some form or another.
Yours very truly,
Geo. J. Romanes.
Dear Darwin,—Criticism of an intelligent kind is what I feel most in need of, and therefore it is no merit on my part to like it when it comes.
The point about the combined action of natural and physiological selection is, after all, a very subordinate one, and, as I said in 'Nature' some weeks ago, is the most highly speculative and least trustworthy part of the theory. Moreover, it is the only part that is directly opposed to an expressed conclusion in the 'Origin,' though, even here, the opposition is not real. If natural selection can do anything at all in the way of bringing about sterility with parent forms, it can only do so by acting on the type or whole community (for I quite agree with the reasoning in the 'Origin,' that it cannot do so by acting on individuals); and whether natural selection could in any case act on a type is a question which your father has told me he could never quite make up his mind about, except in the case of social hymenoptera and moral sense of man.
You will see what I mean by 'secondary variations' by looking at page 366 of my paper. It is merely a short-hand expression for all other specific differences save the sexual difference of sterility. My view is that these secondary differences are always sure to arise sooner or later in some direction or another wherever a portion of a species is separated from the rest, whether by geographical or physiological isolation, which, indeed, as regards the former, is no more than you (following Weismann, &c.) acknowledge. Now, to me it seems obvious that Weismann's 'variations' (i.e. slight changes in the form of shells) cannot possibly be themselves my 'physiological sports,' although they may very well be the consequences of such a sport leading to physiological isolation, and so to independent variation in two or three directions simultaneously, till afterwards blended by intercrossing. And my reason for thinking this is that 'Weismann's variations' always arose in crops at enormously long intervals of time. On the mere doctrine of chances it therefore becomes impossible to suppose that each of these variations was due to a separate physiological sport, although it is easy to see how each crop of them might have been so. For, if not, why should they always have arisen in crops, each member of which was demonstrably fertile with the other members of that crop, while no less demonstrably sterile with the original parent form? Therefore, what I see in these facts is precisely what, upon my theory, I should expect to see, viz. first, a 'primary variation,' or 'physiological sport,' arising at long intervals; secondly, closely following upon this, a crop of 'secondary variations' in the way of slight morphological changes affecting two or three different 'strains' simultaneously; and thirdly, an eventual blending of these strains by intercrossing with one another without being able to intercross with the surrounding and (at first) very much more numerous parent form.
But I can now quite understand why you thought these facts were 'dead against' me; you thought that every single slight change of morphology must (on my theory) have had a separate 'physiological sport' to account for it. This, however, most emphatically is not my theory. Physiological isolation I regard as having morphological consequences precisely analogous to those of geographical isolation; and you would not think of arguing that there must be a separate geographical isolation for every slight change of structure—for example, that a peculiar species of plant growing on a mountain top must have had one isolation to explain its change of form, and another isolation to explain its change of colour.
Lastly, if you will look up Hilgendorf's paper about these snails of Steinheim, I think you will find it impossible to suppose that all these little changes (thus arising at long intervals in crops) can have been useful. Or, if you can still doubt, look up the closely analogous but much larger case of the ammonites investigated by Neumayr and Wurtenberger.
What I meant about the sexual system being specially liable to variation is, that it is specially liable to variation in the way of sterility. In other words, changed conditions of life more readily effect variations in the primary functions of the sexual system than they do in general morphology. But at the same time, I quite agree with your view that in the last resort all changes of structure may be regarded as due to variations of this system. And, as you will see by turning to pp. 371-72 of my paper, important capital is made out of this doctrine.