It is difficult to give any confident answer to this question. We cannot reach clearness on this point through our present knowledge of the germ-plasm, because we possess no insight into its structure; we can only draw conclusions as to the processes in the germ-plasm from the observed phenomena of variation and inheritance. But two facts stand in direct antithesis to one another, first, the high power of adaptation possessed by all species, and the undoubted occurrence of unrestricted persistence in a given direction of variation, as seen in artificial selection, and in the disappearance of parts which have ceased to function; and, secondly, the great constancy of old-established species which do indeed always exhibit a certain degree of individual variability, but without showing marked deviations as a frequent occurrence or in all possible directions, as they certainly would if every determinant favoured by a chance increase in the nutritive stream necessarily and irresistibly went on varying further in the same direction. Or can the constancy of such species be maintained solely by means of personal selection, which is continually setting aside all the determinants which rise above the selection-value by eliminating their possessors? I was for long satisfied that this was the true solution of the difficulty, and even now I do not doubt that personal selection does, in point of fact, maintain the constancy of the species at a certain level, but I do not believe that this is sufficient, but rather that it is necessary to recognize an equalizing influence due to germinal selection, and to attribute to this a share in maintaining the constancy of a species which has long been well adapted. I am led to this assumption chiefly by the phenomena of variation in Man, for we find in him a thousand kinds of minute hereditary individual variations, of which not one is likely to attain to selection value. Of course the constant recurrence of reducing divisions prevents any particular id which contains a varying determinant from being inherited through many generations; for so many ids are being continually removed from the genealogical tree by the constant rejection of the half of all ids of every germ-plasm, that only a small part of the ancestral id remains in the grandchild, great-grandchild, and so on. Certainly some of the ids of the ancestors compose the germ-plasm of the descendants, and if all the determinants of one of these ids had begun to vary persistently upwards or downwards in an ancestor, then all the determinants of the relative id in the descendants would possess the variation in an intensified degree; and however slowly the variation advanced it would attain selection-value in some one or other of the descendants, and would thus break the previously stable type of the most perfectly adapted species. The descendant in question would then succumb in the struggle for existence. But as the number of the determinants in the germ-plasm is probably much greater than that of the descendants of one generation, every descendant would in the course of time deviate unfavourably in some one character from the type of the species, and then either all the descendants would be eliminated or the type would become unstable. But neither of these things happens, and there are undoubtedly species which remain constant for long periods of time, therefore the assumption must be false and every variation of a determinant does not of necessity go on in the same direction without limit.
I therefore suppose that although slight variations are ceaselessly taking place upwards or downwards in all determinants, even in constant species, the majority of these turn again in the other direction before they have attained to any important degree of increase, at least in the germ-plasm of all species which have had a definitely established equilibrium for thousands of generations. In such a germ-plasm, or to speak more precisely, in the id of such a germ-plasm, marked fluctuations in the nutritive stream will not be likely to occur as long as the external conditions are unchanged, but slight fluctuations, which will not be wanting even here, may often alternate and turn in an opposite direction, and thus the upward movement of a determinant may be transformed into a downward one. Every determinant is surrounded by several others, and we can imagine that the regular nutritive stream which we have assumed may be partially dammed up by a slight enlargement of the determinant, and that this will drive the surplus back again. But however we may picture these conditions, which are for all time outside of the sphere of observation, the assumption of a self-regulation of the germ-plasm, up to a certain degree, cannot be regarded as inconceivable or unphysiological.
But there are limits to this self-regulation; as soon as the increase or decrease of a determinant attains a certain degree, as soon as it has got beyond the first slight deviation, it overcomes all obstacles, and goes on increasing in the direction in which it has started. This must happen even in the case of old and constant species, and frequently enough to admit of an apparent capacity for adaptation in all directions. Every part of a species can vary beyond the usual individual fluctuations, and as this is possible only by means of intra-germinal processes, we must assume that even in the case of germ-plasms which have long remained in a state of stable equilibrium there may occasionally be marked fluctuations in the nutritive stream, and thus more than usually pronounced variations of the determinants affected by it will occur. These yield the material for new adaptations if they are in the direction of fitness, or they are eliminated either by the chances of reducing division or by personal selection if new adaptations are not required.
The old-established hereditary equilibrium of the germ-plasm must be most easily disturbed when the species is in some way brought into new conditions of existence, as, for instance, when plants or animals are domesticated, and when in consequence, as we have already assumed, the nutritive currents within the id gradually alter, quantitatively and qualitatively; and on this account alone certain kinds of determinants are favoured, while others are at a disadvantage. In this way there arises the intensified general variability of domesticated animals and cultivated plants which has been known since the time of Darwin. Something analogous to this must occur in natural conditions, though more slowly, when a species is subjected to a change of climatic conditions, but we shall discuss this later on in more detail.
We have thus arrived at the idea that the slight variations of the determinants may be counteracted whether they be directed upwards or downwards, and that in the case of so-called constant species they do frequently equalize themselves; but that more marked variations, produced by more pronounced nutritive fluctuations, may in a sense go on without limit, and then can only be restricted and controlled by personal selection, that is, by the removal of the ids concerned from the genealogical lineage of the species.
In one direction variation can be proved to go on without limit, and that is downwards, as is proved by the fact of the disappearance of disused organs, for here we have a variation-direction, which has been followed to its utmost limit, and which is completely independent of personal selection; it proceeds quite uninterfered with by personal selection, and is left entirely to itself. It is a significant fact that the disappearance of the individual parts of a larger organ, according to all the data that are as yet available, proceeds at a very unequal rate, so that it evidently depends to a great extent on chance whether a disused part begins to degenerate sooner or later. Thus in one of the Crustaceans living in the darkness of the caves of North America the optic lobes and optic nerves have disappeared, while the retina of the eye, the lens, and the pigment have been retained, and in others the reverse has taken place, and the nerve-centres have persisted while the parts of the eye have been lost (Packard). Variations of the relevant determinants towards the minus direction may thus occur, sometimes sooner, sometimes later; but when once they have started they proceed irresistibly, though with exceeding slowness.
But variation in an upward direction also, when it has once been set a-going, may in many cases go on unchecked until limits are set to it by personal selection, when the excess of the organ would disturb the harmony of the parts, or in any other way lessen the individual's chances of survival in the struggle for existence. This is proved especially by the phenomena of artificial selection, for almost all the parts of fowls and pigeons have been caused to vary to excess by breeding, and must thus have been, so to speak, capable of unlimited increase; and yet, as we have seen, personal selection cannot directly cause progress in any direction of variation; it can only secure a free course by excluding from breeding the bearers of variations with an opposite tendency. The beards of hens, the tail-feathers of the long-tailed domestic cocks, the long and short, straight and curved bills of pigeons, the enormously long ruffled feathers of the Jacobin, the multiplication of the tail-feathers in the fan-tail, and innumerable other breed-characters of these playthings of the breeder, prove that when variation-tendencies of any part are once present, that is, when they have arisen through germinal selection, they apparently go on unchecked until their further development would permanently and irretrievably destroy the harmony of the parts. As soon as this is threatened the breed loses its power of survival, and Darwin in his time cited the case of many extremely short-billed breeds of pigeon, which require the aid of the breeder before they can emerge from the hard-shelled egg, because their short and soft bills no longer allow them to break their way out. Here the correlation between the hardness of the egg-shell and that of the pigeon's bill has been disturbed, and the breed can now only be kept in existence by artificial aid.
There must be a possibility of something similar occurring in natural conditions, and when it does the species concerned must die out. But in the majority of cases the self-regulation which is afforded by personal selection will be enough to force back an organ which is in the act of increasing out of due proportion to within its proper limits. The bearers of such excessively increased determinants succumb in the struggle for existence, and the determinants are thus removed from the genealogical lineage of the species.
Having now established the fact that determinants can continue their direction of variation without limit because of internal, that is intra-germinal, reasons, we have come nearer an understanding of many secondary sexual characters, whose resemblance to the excessive developments artificially produced in our domestic poultry is so very striking. Here, too, we shall have to regard germinal selection as the root of the variations of plumage and other distinguishing characters, which have evolved by intra-germinal augmentation into the magnificently coloured crests, tufts, and collars, into the long or graduated, multiplied or erectile tail-feathers of the birds of Paradise, pheasants, and humming-birds. The conception of sexual selection formulated by Darwin will be so far modified, that we are no longer compelled to regard every minute step in this cumulative process as due to the selection of the males by the females. A preference of the finest males may still take place, and is probably general, since only thus could the distinguishing male characters become common property, that is, be transmitted to all or the majority of the ids of the germ-plasm, but the increase of the individual determinants which are in the act of varying goes on in each individual id, quite independently of this personal selection.
As it is not a single id with its determinant a in ascending variation that controls the organ A, but as it always requires a majority of the ids a, this must be secured here by personal selection just as it is in ordinary natural selection. If the handsomest males are the successful competitors, then a majority of the transformed ids a´ will be transmitted to a number of their descendants, and the oftener this happens the larger will the majority be, and the less becomes the danger that it will be dispersed again by reducing division and amphimixis. Personal selection is thus in no way rendered superfluous by germinal selection, only it does not produce the augmentation of the distinguishing characters, but is chiefly instrumental in fixing them in the germ-plasm; it collects, so to speak, only the favourably varying ids, and, where complex variations depending on the proper variation of many ids are concerned, it combines these. How very great the influence of personal selection may be in this case of secondary sexual characters we see clearly from the soberly coloured mates of the brilliant males, for here natural selection has been operative in conserving the coloration inherited from remote ancestry.