It should be observed that not only is a different part of the ear affected in the progeny, but also a very much less quantity thereof. Naturally, therefore, the hypothesis of heredity seems less probable than that of mere coincidence on the one hand, or of transmitted microbes on the other. But I hope to have fairly excluded both these alternative explanations. For, as regards merely accidental coincidence, I have never seen this very peculiar morbid process in the ears, or in any other parts, of guinea-pigs which have neither themselves had their restiform bodies injured, nor been born of parents thus mutilated. As regards the hypothesis of microbes, I have tried to inoculate the corresponding parts of the ears of normal guinea-pigs, by first scarifying those parts and then rubbing them with the diseased surfaces of the ears of mutilated guinea-pigs; but have not been able in this way to communicate the disease.
It will be seen that the above results in large measure corroborate the statements of Brown-Séquard; and it is only fair to add that he told me they are the results which he had himself obtained most frequently, but that he had also met with many cases where the diseased condition of the ears in parents affected the same parts in their progeny, and also occurred in more equal degrees. Lastly, I should like to remark, with regard to these experiments on restiform bodies, and for the benefit of any one else who may hereafter repeat them, that it will be necessary for him to obtain precise information touching the modus operandi. For it is only one very localized spot in each restiform body which has to be injured in order to produce any of the results in question. I myself lost two years of work on account of not knowing this exact spot before going to Paris for the purpose of seeing Brown-Séquard himself perform the operation. I had in the preceding year seen one of his assistants do so, but this gentleman had a much more careless method, and one which in my hands yielded uniformly negative results. The exact spot in question in the restiform body is as far forwards as it is possible to reach, and as far down in depth as is compatible with not producing rotatory movements.
7th. Absence of two toes out of the three of the hind leg, and sometimes of the three, in animals whose parents had eaten up their hind-leg toes which had become anaesthetic from a section of the sciatic nerve alone, or of that nerve and also of the crural. Sometimes, instead of complete absence of the toes, only a part of one or two or three was missing in the young, although in the parent not only the toes but the whole foot were absent.
As I found that the results here described were usually given by division of the sciatic nerve alone—or, more correctly, by excision of a considerable portion of the nerve, in order to prevent regeneration—I did not also divide the crural. But, although I have bred numerous litters from parents thus injured, there has been no case of any inherited deficiency of toes. My experiments in this connexion were carried on through a series of six successive generations, so as to produce, if possible, a cumulative effect. Nevertheless, no effect of any kind was produced. On the other hand, Brown-Séquard informed me that he had observed this inherited absence of toes only in about one or two per cent. of cases. Hence it is possible enough, that my experiments have not been sufficiently numerous to furnish a case. It may be added that there is here no measurable possibility of accidental coincidence (seeing that normal guinea-pigs do not seem ever to produce young with any deficiency of toes), while the only possibility of mal-observation consists in some error with regard to the isolation (or the tabulation) of parents and progeny. Such an error, however, may easily arise. For gangrene of the toes does not set in till some considerable time after division of the sciatic nerve. Hence, if the wound be healed before the gangrene begins, and if any mistake has been made with regard to the isolation (or tabulation) of the animal, it becomes possible that the latter should be recorded as an uninjured, instead of an injured, individual. On this account one would like to be assured that Brown-Séquard took the precaution of examining the state of the sciatic nerve in those comparatively few specimens which he alleges to have displayed such exceedingly definite proof of the inheritance of a mutilation. For it is needless to remark, after what has been said in the preceding chapter on the analogous case of epilepsy, that the proof would not be regarded by any physiologist as displaced by the fact that there is no observable deficiency in the sciatic nerve of the toeless young.
8th. Appearance of various morbid states of the skin and hair of the neck and face in animals born of parents having had similar alterations in the same parts, as effects of an injury to the sciatic nerve.
I have not paid any attention to this paragraph, because the facts which it alleges did not seem of a sufficiently definite character to serve as a guide to further experiment.
On the whole, then, as regards Brown-Séquard's experiments, it will be seen that I have not been able to furnish any approach to a full corroboration. But I must repeat that my own experiments have not as yet been sufficiently numerous to justify me in repudiating those of his statements which I have not been able to verify.
The only other experimental results, where animals are concerned, which seemed to tell on the side of Lamarckianism, are those of Mr. Cunningham, already alluded to. But, as the research is still in progress, the school of Weismann may fairly say that it would be premature to discuss its theoretical bearings.
Passing now from experiments on animals to experiments on plants, I must again ask it to be borne in mind, that here also no researches have been published, which have had for their object the testing of the question on which we are engaged. As in the case of animals, therefore, so in that of plants, we are dependent for any experimental results bearing upon the subject to such as have been gained incidentally during the course of investigations in quite other directions.
Allusion has already been made, in my previous essay, to De Vries' observations on the chromatophores of algae passing from the ovum of the mother to the daughter organism; and we have seen that even Weismann admits, "It appears possible that a transmission of somatogenetic variation has here occurred[71]." It will now be my object to show that such variations appear to be sometimes transmitted in the case of higher plants, and this under circumstances which carry much less equivocal evidence of the inheritance of acquired characters, than can be rendered by the much more simple organization of an alga.
I have previously mentioned Hoffmann's experiments on transplantation, the result of which was to show that variations, directly induced by changed conditions of life, were reproduced by seed[72]. Weismann, however, as we have seen, questions the somatogenetic origin of these variations—attributing the facts to a blastogenetic change produced in the plants by a direct action of the changed conditions upon the germ-plasm itself[73]. And he points out that whether he is right or wrong in this interpretation can only be settled by ascertaining whether the observable somatic changes occur in the generation which is first exposed to the changed conditions of life. If they do occur in the first generation, they are somatogenetic changes, which afterwards react on the substance of heredity, so as to transmit the acquired peculiarities to progeny. But if they do not occur till the second (or any later) generation, they are presumably blastogenetic. Unfortunately Hoffmann does not appear to have attended to this point with sufficient care, but there are other experiments of the same kind where the point has been specially observed.
For instance, M. L. A. Carrière[74] gathered seed from the wild radish (Raphanus Raphanistrum) in France, and sowed one lot in the light dry soil near the Museum of Natural History in Paris, while another lot was sown by him at the same time in heavy soil elsewhere. His object was to ascertain whether he could produce a good cultivated radish by methodical selection; and this he did; in a wonderfully rapid manner, during the course of a very few generations. But the point for us is, that from the first the plants grown in the light soil of Paris presented sundry marked differences from those grown in the heavy soil of the country; and that these points of difference had nothing to do with the variations on which his artificial selection was brought to bear. For while his artificial selection was directed to increasing the size of the "root," the differences in question had reference to its form and colour. In Paris an elongated form prevailed, which presented either a white or a rose colour: in the country the form was more rounded, and the colour violet, dark brown, or "almost black." Now, as these differences were strongly apparent in the first generation, and were not afterwards made the subject of selection, both in origin and development they must have been due to "climatic" influences acting on the somatic tissues. And although the author does not appear to have tested their hereditary characters by afterwards sowing the seed from the Paris variety in the country, or vice versa, we may fairly conclude that these changes must have been hereditary—1st, from the fact of their intensification in the course of the five sequent generations over which the experiment extended, and, 2nd, from the very analogous results which were similarly obtained in the following case with another genus, where both the somatogenetic and the hereditary characters of the change were carefully and specially observed. This case is as follows.
The late Professor James Buckman, F.R.S., saved some seed from wild parsnips (P. sativa) in the summer of 1847, and sowed under changed conditions of life in the spring of 1848. The plants grown from these wild seeds were for the most part like wild plants; but some of them had "already (i.e. in the autumn of 1848) the light green and smooth aspect devoid of hairs which is peculiar to the cultivated plant; and among the latter there were a few with longer leaves and broader divisions of leaf-lobes than the rest—the leaves, too, all growing systematically round one central bud. The roots of the plant when taken up were observed to be for the most part more fleshy than those of wild examples[75]."
Professor Buckman then proceeds to describe how he selected the best samples for cultivation in succeeding generations, till eventually the variety which he called "The Student" was produced, and which Messrs. Sutton still regard as the best variety in their catalogue. That is to say, it has come true to seed for the last forty years; and although such great excellence and stability are doubtless in chief part due to the subsequent process of selection by Professor Buckman in the years 1848-1850, this does not affect the point with which we are here concerned—namely, that the somatogenetic changes of the plants in the first generation were transmitted by seed to the second generation, and thus furnished Professor Buckman with the material for his subsequent process of selection. And the changes in question were not merely of a very definite character, but also of what may be termed a very local character—affecting only particular tissues of the soma, and therefore expressive of a high degree of representation on the part of the subsequently developed seed, by which they were faithfully reproduced in the next generation.
Here is another case. M. Lesage examined the tissues of a large number of plants growing both near to, and remote from, the sea. He suspected that the characteristic fleshiness, &c. of seaside plants was due to the influence of sea-salt; and proved that such was the case by causing the characters to occur in inland plants as a result of watering them with salt-water. Then he adds:—
"J'ai réussi surtout pour le Lepidium sativum cultivé en 1888; j'ai obtenu pour la même plante des résultats plus nets encore dans la culture de 1889, entreprise en semant les graines récoltées avec soin des pots de l'année précédente et traitées exactement de la même façon[76]."
Here, it will be observed, there was no selection; and therefore the increased hereditary effect in the second generation must apparently be ascribed to a continuance of influence exercised by somatic tissues on germinal elements; for at the time when the changes were produced no seed had been formed. In other words, the accumulated change, like the initial change, would seem to have been exclusively of somatogenetic origin; and yet it so influenced the qualities of the seed (as this was afterwards formed), that the augmented changes were transmitted to the next generation, part for part, as the lesser changes had occurred in the preceding generation. "This experiment, therefore, like Professor Buckman's, shows that the alteration of the tissues was carried on in the second generation from the point gained in the first. In both cases no germ-plasm (in the germ-cells) existed at the time during which the alterations arose, as they were confined to the vegetative system; and in the case of the parsnips and carrots, being biennials no germ-cells are produced till the second year has arrived[77]."
Once more, Professor Bailey remarks:—
"Squashes often show remarkable differences when grown upon different soils; and these differences can sometimes be perpetuated for a time by seeds. The writer has produced, from the same parent, squashes so dissimilar, through the simple agency of a change of soil in one season, that they might readily be taken for distinct varieties. Peas are known to vary in the same manner. The seeds of a row of peas of the same kind, last year gave the writer marked variations due to differences of soil.... Pea-growers characterize soils as 'good' and 'viney.' Upon the latter sort the plants run to vine at the expense of the fruit, and their offspring for two or three generations have the same tendency[78]."
I think these several cases are enough to show that, while the Weismannian assumption as to the seeming transmission of somatogenetic characters being restricted to the lowest kinds of plants is purely gratuitous, there is no small amount of evidence to the contrary—or evidence which seems to prove that a similar transmission occurs likewise in the higher plants. And no doubt many additional cases might be advanced by any one who is well read in the literature of economic botany.
It appears to me that the only answer to such cases would be furnished by supposing that the hereditary changes are due to an alteration of the residual "germ-plasm" in the wild seed, when this is first exposed to the changed conditions of life, due to its growth in a strange kind of soil—e.g. while germinating in an unusual kind of earth for producing the first generation. But this would be going a long way to save an hypothesis. In case, however, it should now be suggested, I may remark that it would be negatived by the following facts.[79]
In the first place, an endless number of cases might be quoted where somatogenetic changes thus produced by changed conditions of life are not hereditary. Therefore, in all these cases it is certainly not the "germ-plasm" that is affected. In other words, there can be no question that somatogenetic changes of the kinds above mentioned do very readily admit of being produced in the first generation by changes of soil, altitude, &c. And that somatogenetic changes thus produced should not always—or even generally—prove themselves to be hereditary from the first moment of their occurrence, is no more than any theory of heredity would expect. Indeed, looking to the known potency of reversion, the wonder is that in any case such changes should become hereditary in a single generation. On the other hand, there is no reason to imagine that the hypothetical germ-plasm—howsoever unstable we may suppose it to be—can admit of being directly affected by a change of soil in a single generation. For, on this view, it must presumably be chiefly affected during the short time that the seed is germinating; and during that time the changed conditions can scarcely be conceived as having any points of attack, so to speak, upon the residual germ-plasm. There are no roots on which the change of soil can make itself perceptible, nor any stem and leaves on which the change of atmosphere can operate. Yet the changed condition's may produce hereditary modifications in any parts of the plant, which are not only precisely analogous to non-hereditary changes similarly produced in the somatic tissues of innumerable other plants, but are always of precisely the same kind in the same lot of plants that are affected. When all the radishes grown from wild seed in Paris, for instance, varied in the direction of rotundity and dark colour, while those grown in the country presented the opposite characters, we can well understand the facts as due to an entire season's action upon the whole of the growing plant, with the result that all the changes produced in each set of plants were similar—just as in the cases where similarly "climatic" modifications are not hereditary, and therefore unquestionably due to changed conditions acting on roots, stems, leaves, or flowers, as the case may be. On the other hand, it is not thus intelligible that during the short time of germination the changed conditions should effect a re-shuffling (or any other modification) of the "germ-plasm" in the seeds—and this in such a manner that the effect on the residual germ-plasm reserved for future generations is precisely similar to that produced on the somatic tissues of the developing embryo.
In the second place, as we have seen, in some of the foregoing cases the changes were produced months—and even years—before the seeds of the first germination were formed. Therefore the hereditary effect, if subsequent to the period of embryonic germination, must have been produced on germ-plasm as this occurs diffused through the somatic tissues. But, if so, we shall have to suppose that such germ-plasm is afterwards gathered in the seeds when these are subsequently formed. This supposition, however, would be radically opposed to Weismann's theory of heredity: nor do I know of any other theory with which it would be reconcilable, save such as entertain the possibility of the Lamarckian factors.
Lastly, in the third place, I deem the following considerations of the highest importance:—
"As other instances in which peculiar structures are now hereditary may be mentioned aquatic plants and those producing subterraneous stems. Whether they be dicotyledons or monocotyledons, there is a fundamental agreement in the anatomy of the roots and stem of aquatic plants, and, in many cases, of the leaves as well. Such has hitherto been attributed to the aquatic habit. The inference or deduction was, of course, based upon innumerable coincidences; the water being supposed to be the direct cause of the degenerate structures, which are hereditary and characteristic of such plants in the wild state. M. Costantin has, however, verified this deduction, by making terrestrial and aerial stems to grow underground and in water: the structures at once began to assume the subterranean or aquatic type, as the case might be; and, conversely, aquatic plants made to grow upon land at once began to assume the terrestrial type of structure, while analogous results followed changes from a subterranean to an aerial position, and vice versa."
This is also quoted from the Rev. Prof. Henslow's letters to me, and the important point in it is, that the great changes in question are proved to be of a purely "somatogenetic" kind; for they occurred "at once" in the ready-grown plant, when the organs concerned were exposed to the change from aquatic to terrestrial life, or vice versa—and also from a subterranean to an aerial position, or vice versa. Consequently, even the abstract possibility of the changed conditions of life having operated on the seed is here excluded. Yet the changes are of precisely the same kind as are now hereditary in the wild species. It thus appears undeniable that all these remarkable and uniform changes must originally have been somatogenetic changes; yet they have now become blastogenetic. This much, I say, seems undeniable; and therefore it goes a long way to prove that the non-blastogenetic character of the changes has been due to their originally somatogenetic character. For, if not, how did natural selection ever get an opportunity of making any of them blastogenetic, when every individual plant has always presented them as already given somatogenetically? This last consideration appears in no small measure to justify the opinion of Mr. Henslow, who concludes—"These experiments prove, not only that the influence of the environment is at once felt by the organ; but that it is indubitably the cause of the now specific and hereditary traits peculiar to normally aquatic, subterranean, and aerial stems, or roots[80]."
He continues to furnish other instances in the same line of proof—such as the distinctive "habits" of insectivorous, parasitic, and climbing plants; the difference in structure between the upper and under sides of horizontal leaves, &c. "For here, as in all organs, we discover by experiment how easily the anatomy of plants can be affected by their environment; and that, as long as the latter is constant, so are the characters of the plants constant and hereditary."
[The following letter, contributed by Dr. Hill to Nature, vol. I. p. 617, may here be quoted. C. Ll. M.
"It may be of interest to your readers to know that two guinea-pigs were born at Oxford a day or two before the death Dr. Romanes, both of which exhibited a well-marked droop of the left upper eyelid. These guinea-pigs were the offspring of a male and a female guinea-pig in both of which I had produced for Dr. Romanes, some months earlier, a droop of the left upper eyelid by division of the left cervical sympathetic nerve. This result is a corroboration of the series of Brown-Séquard's experiments on the inheritance of acquired characteristics. A very large series of such experiments are of course needed to eliminate all sources of error, but this I unfortunately cannot carry out at present, owing to the need of a special farm in the country, for the proper care and breeding of the animals.—Leonard Hill.
"Physiological Laboratory, Univ. Coll. London, Oct. 18, 1894."]
The strongest argument in favour of "continuity" is that based upon the immense difference between congenital and acquired characters in respect of heritability. For that there is a great difference in this respect is a matter of undeniable fact. And it is obvious that this difference, the importance of which must be allowed its full weight, is just what we should expect on the theory of the continuity of the germ-plasm, as opposed to that of pangenesis. Indeed it may be said that the difference in question, while it constitutes important evidence in favour of the former theory, is a difficulty in the way of the latter. But here two or three considerations must be borne in mind.
In the first place, this fact has long been one which has met with wide recognition and now constitutes the main ground on which the theory of continuity stands. That is to say, it was the previous knowledge of this contrast between congenital and acquired characters which led to the formulation of a theory of continuity by Mr. Galton, and to its subsequent development by Prof Weismann.
But, in the second place, there is a wide difference between the certainty of this fact and that of the theory based upon it. The certain fact is, that a great distinction in respect of heritability is observable between congenital and acquired characters. The theory, as formulated by Weismann, is that the distinction is not only great but absolute, or, in other words, that in no case and in no degree can any acquired character be ever inherited. This hypothesis, it will be observed, goes far beyond the observed fact, for it is obviously possible that, notwithstanding this great difference in regard to heritability between congenital and acquired characters, the latter may nevertheless, sometimes and in some degree, be inherited, however much difficulty we may experience in observing these lesser phenomena in presence of the greater. The Weismannian hypothesis of absolute continuity is one thing, while the observed fact of at least a high relative degree of continuity is quite another thing. And it is necessary to be emphatic on this point, since some of the reviewers of my Examination of Weismannism confound these two things. Being apparently under the impression that it was reserved for Weismann to perceive the fact of there being a great difference between the heritability of congenital and acquired characters, they deem it inconsistent in me to acknowledge this fact while at the same time questioning the hypothetical basis of his fundamental postulate touching the absolute continuity of germ-plasm. It is one merit of Galton's theory, as against Weismann's, that it does not dogmatically exclude the possible interruption of continuity on some occasions and in some degree. Herein, indeed, would seem to lie the central core of the whole question in dispute. For it is certain and has long been known that individually acquired characters are at all events much less heritable than are long-inherited or congenital ones. But Lamarckian theory supposes that congenital characters were in some cases originally acquired, and that what are now blastogenetic characters were in some cases at first somatogenetic and have become blastogenetic only in virtue of sufficiently long inheritance. Since Darwin's time, however, evolutionists (even of the so-called Lamarckian type) have supposed that natural selection greatly assists this process of determining which somatogenetic characters shall become congenital or blastogenetic. Hence all schools of evolutionists are, and have long been, agreed in regarding the continuity principle as true in the main. No evolutionist would at any time have propounded the view that one generation depends for all its characters on those acquired by its immediate ancestors, for this would merely be to unsay the theory of Evolution itself, as well as to deny the patent facts of heredity as shown, for example, in atavism. At most only some fraction of a per cent. could be supposed to do so. But Weismann's contention is that this principle is not only true in the main, but absolutely true; so that natural selection becomes all in all or not at all. Unless Weismannism be regarded as this doctrine of absolutism it permits no basis for his attempted theory of evolution.
And, whatever may be said to the contrary by the more enthusiastic followers of Prof. Weismann, I must insist that there is the widest possible difference between the truly scientific question of fact which is assumed by Weismann as answered (the base-line of the diagram on p. 43), and the elaborate structure of deductive reasoning which he has reared on this assumption (the Y-like structure). Even if the assumption should ever admit of inductive proof, the almost bewildering edifice of deductive reasoning which he has built upon it would still appear to me to present extremely little value of a scientific kind. Interesting though it may be as a monument of ingenious speculation hitherto unique in the history of science, the mere flimsiness of its material must always prevent its far-reaching conclusions from being worthy of serious attention from a biological point of view. But having already attempted to show fully in my Examination this great distinction between the scientific importance of the question which lies at the base of "Weismannism," and that of the system which he has constructed on his assumed answer thereto, I need not now say anything further with regard to it.
Again, on the present occasion and in this connexion I should like to dissipate a misunderstanding into which some of the reviewers of the work just mentioned have fallen. They appear to have concluded that because I have criticized unfavourably a considerable number of Weismann's theories, I have shown myself hostile to his entire system. Such, however, is by no means the case; and the misunderstanding can only be accounted for by supposing that the strongly partisan spirit which these critics display on the side of neo-Darwinism has rendered them incapable of appreciating any attempt at impartial—or even so much as independent—criticism. At all events, it is a matter of fact that throughout the work in question I have been particularly careful to avoid this misunderstanding as to my own position. Over and over again it is there stated that, far from having any objection to the principle of "Continuity" as represented in the base-line of the above diagram, I have been convinced of its truth ever since reading Mr. Galton's Theory of Heredity in 1875. All the "hard words" which I have written against Weismann's system of theories have reference to those parts of it which go to constitute the Y-like structure of the diagram.
It is, however, desirable to recur to another point, and one which I hope will be borne in mind throughout the following discussion. It has already been stated, a few pages back, that the doctrine of continuity admits of being held in two very different significations. It may be held as absolute, or as relative. In the former case we have the Weismannian doctrine of germ-plasm: the substance of heredity is taken to be a substance per se, which has always occupied a separate "sphere" of its own, without any contact with that of somatoplasm further than is required for its lodgement and nutrition; hence it can never have been in any degree modified as to its hereditary qualities by use-inheritance or any other kind of somatogenetic change; it has been absolutely continuous "since the first origin of life." On the other hand, the doctrine of continuity may be held in the widely different sense in which it has been presented by Galton's theory of Stirp. Here the doctrine is, that while for the most part the phenomena of heredity are due to the continuity of the substance of heredity through numberless generations, this substance ("Stirp") is nevertheless not absolutely continuous, but may admit, in small though cumulative degrees, of modification by use-inheritance and other factors of the Lamarckian kind. Now this all-important distinction between these two theories of continuity has been fully explained and thoroughly discussed in my Examination; therefore I will not here repeat myself further than to make the following remarks.
The Weismannian doctrine of continuity as absolute (base-line of the diagram) is necessary for the vast edifice of theories which he has raised upon it (the Y), first as to the minute nature and exact composition of the substance of heredity itself ("Germ-plasm"), next as to the precise mechanism of its action in producing the visible phenomena of heredity, variation, and all allied phenomena, and, lastly, the elaborate and ever-changing theory of organic evolution which is either founded on or interwoven with this vast system of hypothetic speculation. Galton's doctrine of continuity, on the other hand, is a "Theory of Heredity," and a theory of heredity alone. It does not meddle with any other matters whatsoever, and rigidly avoids all speculation further than is necessary for the bare statement and inductive support of the doctrine in question. Hence, it would appear that this, the only important respect wherein the doctrine of continuity as held by Galton differs from the doctrine as held by Weismann, arisen from the necessity under which the latter finds himself of postulating absolute continuity as a logical basis for his deductive theory of the precise mechanism of heredity on the one hand, and of his similarly deductive theory of evolution on the other. So far as the doctrine of continuity is itself concerned (i.e. the question of the inheritance of acquired characters), there is certainly no more inductive reason for supposing the continuity absolute "since the first origin of life," than there is for supposing it to be more or less susceptible of interruption by the Lamarckian factors. In other words, but for the sake of constructing a speculative foundation for the support of his further theories as to "the architecture of germ-plasm" and the factors of organic evolution, there is no reason why Weismann should maintain the absolute separation of the "sphere" of germ-plasm from that of somatoplasm. On the contrary, he has no reason for concluding against even a considerable and a frequent amount of cutting, or overlapping, on the part of these two spheres.
But although this seems to me sufficiently obvious, as I have shown at greater length in the Examination of Weismannism, it must not be understood that I hold that there is room for any large amount of such overlapping. On the contrary, it appears to me as certain as anything can well be that the amount of such overlapping from one generation to another, if it ever occur at all, must be exceedingly small, so that, if we have regard to only a few sequent generations, the effects of use-inheritance, and Lamarckian factors are, at all events as a rule, demonstrably imperceptible. But this fact does not constitute any evidence—as Weismann and his followers seem to suppose—against a possibly important influence being exercised by the Lamarckian factors, in the way of gradual increments through a long series of generations. It has long been well known that acquired characters are at best far less fully and far less certainly inherited than are congenital ones. And this fact is of itself sufficient to prove the doctrine of continuity to the extent that even the Lamarckian is rationally bound to concede. But the fact yields no proof—scarcely indeed so much as a presumption—in favour of the doctrine of continuity as absolute. For it is sufficiently obvious that the adaptive work of heredity could not be carried on at all if there had to be a discontinuity in the substance of heredity at every generation, or even after any very large number of generations.
Little more need be said concerning the arguments which fall under the headings A and B. The Indirect evidence is considered in Appendix I of the Examination of Weismannism; while the Direct evidence is considered in the text of that work in treating of Professor Weismann's researches on the Hydromedusae (pp. 71-76).
The facts of karyokinesis are generally claimed by the school of Weismann as making exclusively in favour of continuity as absolute. But this is a partisan view to take. In any impartial survey it should be seen that while the facts are fairly interpretable on Weismann's theory, they are by no means proof thereof. For any other theory of Heredity must suppose the material of heredity to be of a kind more or less specialized, and the mechanism of heredity extremely precise and well ordered. And this is all that the facts of karyokinesis prove. Granting that they prove continuity, they cannot be held to prove that continuity to be absolute. In other words, the facts are by no means incompatible with even a large amount of commerce between germ-plasm and somato-plasm, or a frequent transmission of acquired characters.
Again, Weismann's theory, that the somatic and the germ-plasm determinants may be similarly and simultaneously modified by external conditions may be extended much further than he has used it himself, so as to exclude, or at any rate invalidate, all evidence in favour of Lamarckianism, other than the inheritance of the effects of use and disuse. All evidence from apparently inherited effects produced by change of external conditions is thus virtually put out of court, leaving only evidence from the apparently inherited effects of functionally produced modifications. And this line of evidence is invalidated by Panmixia. Hence there remain only the arguments from selective value and co-adaptation. Weismann meets these by adducing the case of neuter insects, which have been already considered at sufficient length.
Let us now proceed to the experimental evidence which has been adduced on the side of Weismannism.
Taking this evidence in order of date, we have first to mention that on which the school of Weismann has hitherto been satisfied almost exclusively to rely. This is the line of negative evidence, or the seeming absence of any experimental demonstration of the inheritance of acquired characters. This kind of evidence, however, presents much less cogency than is usually supposed. And it has been shown in the last chapter that the amount of experimental evidence in favour of the transmission of acquired characters is more considerable than the school of Weismann seems to be aware—especially in the vegetable kingdom. I do not think that this negative line of evidence presents much weight; and, to show that I am not biassed in forming this judgement, I may here state that few have more reason than myself for appreciating the weight of such evidence. For, as already stated, when first led to doubt the Lamarckian factors, now more than twenty years ago, I undertook a research upon the whole question—only a part of which was devoted to testing the particular case of Brown-Séquard's statements, with the result recorded in the preceding chapter. As this research yielded negative results in all its divisions—and, not only in the matter of Brown-Séquard's statements—I have not hitherto published a word upon the subject. But it now seems worth while to do so, and for the following reasons.
First, as just observed, a brief account of my old experiences in this field will serve to show what good reason I have for feeling the weight of such negative evidence in favour of Continuity as arises from failure to produce any good experimental evidence to the contrary. In the second place, now that the question has become one of world-wide interest, it would seem that even negative results deserve to be published for whatever they may be worth on the side of Neo-Darwinism. Lastly, in the third place, although the research yielded negative results in my hands, it is perhaps not undesirable to state the nature of it, if only to furnish suggestions to other physiologists, in whose hands the experiments—especially in these days of antiseptics—may lead to a different termination. Altogether I made thousands of experiments in graft-hydridization (comprising bines, bulbs of various kinds, buds, and tubers); but with uniformly negative results. With animals I tried a number of experiments in grafting characteristic congenital tissues from one variety on another—such as the combs of Spanish cocks upon the heads of Hamburgs; also, in mice and rats, the grafting together of different varieties; and, in rabbits and bitches, the transplantation of ovaries of newly-born individuals belonging to different well-marked breeds. This latter experiment seems to be one which, if successfully performed (so that the transplanted ovaries would form their attachment in a young bitch puppy and subsequently yield progeny to a dog of the same breed as herself) would furnish a crucial test as to the inheritance or non-inheritance of acquired characters. Therefore I devoted to it a large share of my attention, and tried the experiment in several different ways. But I was never able to get the foreign ovary—or even any portion thereof—to graft. Eventually the passing of the Vivisection Act caused me to abandon the whole research as far as animals were concerned—a research, indeed, of which I had become heartily tired, since in no one instance did I obtain any adhesion. During the last few years, however, I have returned to these experiments under a licence, and with antiseptic precautions, but with a similar want of success. Perhaps this prolonged and uniformly fruitless experience may now have the effect of saving the time of other physiologists, by warning them off the roads where there seems to be no thoroughfare. On the other hand, it may possibly lead some one else to try some variation in the method, or in the material, which has not occurred to me. In particular, I am not without hope that the transplantation of ovaries in very young animals may eventually prove to be physiologically possible; and, if so, that the whole issue as between the rival theories of heredity will be settled by the result of a single experiment. Possibly some of the invertebrata will be found to furnish the suitable material, although I have been unable to think of any of these which present sufficiently well-marked varieties for the purpose. But, pending the successful accomplishment of this particular experiment in the grafting of any animal tissue, I think it would be clearly unjustifiable to conclude against the Lamarckian factors on the ground of any other experiments yielding negative results in but one generation or even in a large number of sequent generations.
For instance, the latter consideration applies to the negative results of Mr. Francis Galton's celebrated Experiments in Pangenesis.[82]. These consisted in transfusing the blood of one variety of rabbit into the veins of both sexes of another, and then allowing the latter to breed together: in no case was there any appearance in the progeny of characters distinctive of the variety from which the transfused blood was derived. But, as Mr. Galton himself subsequently allowed, this negative result constitutes no disproof of pangenesis, seeing that only a portion of the parents' blood was replaced; that this portion, even if charged with "gemmules," would contain but a very small number of these hypothetical bodies, compared with those contained in all the tissues of the parents; and that even this small proportional number would presumably be soon overwhelmed by those contained in blood newly-made by the parents. Nevertheless the experiment was unquestionably worth trying, on the chance of its yielding a positive result; for, in this event, the question at issue would have been closed. Accordingly I repeated these experiments (with the kind help of Professor Schäfer), but with slight differences in the method, designed to give pangenesis a better chance, so to speak.
Thus I chose wild rabbits to supply the blood, and Himalayan to receive it—the former being the ancestral type (and therefore giving reversion an opportunity of coming into play), while the latter, although a product of domestication, is a remarkably constant variety, and one which differs very much in size and colour from the parent species. Again, instead of a single transfusion, there were several transfusions performed at different times. Moreover, we did not merely allow the blood of one rabbit to flow into the veins of the other (whereby little more than half the blood could be substituted); but sacrificed three wild rabbits for refilling the vascular system of each tame one on each occasion. Even as thus improved, however, the experiment yielded only negative results, which, therefore, we never published.
Subsequently I found that all this labour, both on Mr. Galton's part and our own, was simply thrown away—not because it yielded only negative results, but because it did not serve as a crucial experiment at all. The material chosen was unserviceable for the purpose, inasmuch as rabbits, even when crossed in the ordinary way, never throw intermediate characters. Needless to say, had I been aware of this fact before, I should never have repeated Mr. Galton's experiments—nor, indeed, would he have originally performed them had he been aware of it. So all this work goes for nothing. The research must begin all over again with some other animals, the varieties of which when crossed do throw intermediate characters.
Therefore I have this year made arrangements for again repeating the experiments in question—only, instead of rabbits, using well-marked varieties of dogs. A renewed attack of illness, however, has necessitated the surrender of this research to other hands, with a consequent delay in its commencement.
My ignorance of the unfortunate peculiarity displayed by rabbits in not throwing intermediate characters has led to a further waste of time in another line of experiment. On finding that mammalian ovaries did not admit of being grafted, it seemed to me that the next best thing to try would be the transplantation of fertilized ova from one variety to another, for the purpose of ascertaining whether, if a parturition should take place under such circumstances, gestation by the uterine mother would affect the characters of the ovum derived from the ovarian mother—she, of course, having been fertilized by a male of her own variety. Of course it was necessary that both the mothers should be in season at about the same time, and therefore I again chose rabbits, seeing that in the breeding season they are virtually in a chronic state of "heat." I selected Himalayans and Belgian hares, because they are well-marked varieties, breed true, and in respect of colour are very different from one another. It so happened that while I was at work upon this experiment, it was also being tried, unknown to me, by Messrs. Heape and Buckley who, curiously enough, employed exactly the same material. They were the first to obtain a successful result. Two fertilized ova of the Angora breed having been introduced into the fallopian tube of a Belgian hare, developed there in due course, and gave rise to two Angora rabbits in no way modified by their Belgian hare gestation[83].
But, interesting and suggestive as this experiment is in other connexions, it is clearly without significance in the present one, for the reason already stated. It will have to be tried on well-marked varieties of other species of animals, which are known to throw intermediate characters. Even, however, if it should then yield a similarly negative result, the fact would not tell against the inheritance of acquired characters; seeing that an ovum by the time it is ripe is a finished product, and therefore not to be expected, on any theory of heredity, to be influenced as to its hereditary potentialities by the mere process of gestation. On the other hand, if it should prove that it does admit of being thus affected, so that against all reasonable expectation the young animal presents any of the hereditary characters of its uterine mother, the fact would terminate the question of the transmission of acquired characters—and this quite as effectually as would a similarly positive result in the case of progeny from an ingrafted ovary of a different variety. In point of fact, the only difference between the two cases would be, that in the former it might prove possible to close the question on the side of Lamarckianism, in the latter it would certainly close the question, either on this side or on the opposite as the event would determine.
The only additional fact that has hitherto been published by the school of Weismann is the result of Weismann's own experiment in cutting off the tails of mice through successive generations. But this experiment does not bear upon any question that is in debate; for no one who is acquainted with the literature of the subject would have expected any positive result to follow from such a line of inquiry. As shown further back in the text, Darwin had carefully considered the case of mutilations, and explained that their non-transmissibility constitutes no valid objection to his theory of pangenesis. Furthermore, it may now be added, he expressly alluded in this connexion to the cutting off of tails, as practised by horse-breeders and dog-fanciers, "through a number of generations, without any inherited effect." He also alluded to the still better evidence which is furnished by the practice of circumcision. Therefore it is difficult to understand the object of Weismann's experiment. Yet, other than the result of this experiment, no new fact bearing on the question at issue has been even so much as alleged.
In the foregoing chapters I have endeavoured to be, before all things, impartial; and if it seems that I have been arguing chiefly in favour of the Lamarckian principles, this has been because the only way of examining the question is to consider what has to be said on the affirmative side, and then to see what the negative side can say in reply. Before we are entitled to discard the Lamarckian factors in toto, we must be able to destroy all evidence of their action. This, indeed, is what the ultra-Darwinians profess to have done. But is not their profession premature? Is it not evident that they have not sufficiently considered certain general facts of nature, or certain particular results of experiment, which at all events appear inexplicable by the theory of natural selection alone? In any case the present discussion has been devoted mainly to indicating such general facts and particular results. If I have fallen into errors, either of statement or of reasoning, it is for the ultra-Darwinians to correct them; but it may be well to remark beforehand, that any criticism of a merely general kind touching the comparative paucity of the facts thus adduced in favour of Lamarckian doctrine, will not stand as a valid criticism. For, as we have seen in the opening part of the discussion, even if use-inheritance and direct action of the environment have been of high importance as factors of organic evolution, it must be in almost all cases impossible to dissociate their influence from that of natural selection—at any rate where plants and animals in a state of nature are concerned. On the other hand, experiments expressly devised to test the question have not hitherto been carried out. Besides, the facts and arguments here adduced are but comparatively few. For, unless it can be shown that what has been said of reflex action, instinct, so-called "self-adaptation" in plants, &c., is wrong in principle, the facts which tell in favour of Lamarckian theory are absolutely very numerous. Only when considered in relation to cases where we are unable to exclude the conceivable possibility of natural selection having been at work, can it be said that the facts in question are not numerous.
Comparatively few, then, though the facts may be of which I have given some examples, in my opinion they are amply sufficient for the purpose in hand. This purpose is to show that the question which we are now considering is very far from being a closed question; and, therefore, that the school of Weismann is much too precipitate in alleging that there is neither any necessity for, nor evidence of, the so-called Lamarckian factors[85]. And this opinion, whatever it may be worth, is at all events both deliberate and impartial. As one of the first to doubt the transmission of acquired characters, and as one who has spent many years in experimental inquiries upon the subject, any bias that I may have is assuredly against the Lamarckian principles—seeing that nearly all my experiments have yielded negative results. It was Darwin himself who checked this bias. But if the ultra-Darwinians of the last ten years had succeeded in showing that Darwin was mistaken, I should be extremely glad to fall into line with them. As already shown, however, they have in no way affected this question as it was left by Galton in 1875. And if it be supposed a matter of but little importance whether we agree with Galton in largely diminishing the comparative potency of the Lamarckian principles, or whether we agree with Weismann in abolishing them together, it cannot be too often repeated that such is an entirely erroneous view. No matter how faintly or how fitfully acquired characters may be transmitted, in so far as they are likewise adaptive characters, their transmission (and therefore their development) must be cumulative. Hence, the only effect of attenuating our estimate of their intensity, is that of increasing our estimate of their duration—i.e. of the time over which they have to operate in order to produce important results. And, even so, it is to be remembered that the importance of such results is not to be estimated by the magnitude of modification. Far more is it to be estimated by the character of modification as adaptive. For if functionally produced changes, and changes produced in adaptive response to the environment, are ever transmitted in a cumulative manner, a time must sooner or later arrive when they will reach a selective value in the struggle for existence—when, of course, they will be rapidly augmented by natural selection. Thus, if in any degree operative at all, the great function of these principles must be that of supplying to natural selection those incipient stages of adaptive modifications in all cases where, but for their agency, there would have been nothing of the kind to select. Themselves in no way dependent on adaptive modifications having already attained a selective value, these Lamarckian principles are (under the Darwinian theory) direct causes of determinate variation in adaptive lines; and variation in those lines being cumulative, the result is that natural selection is in large part presented with the raw material of its manufacture—special material of the particular kinds required, as distinguished from promiscuous material of all kinds. And the more complex the manufacture the more important will be the work of this subordinate factory. We can well imagine how the shell of a nut, for instance, or even the protective colouring of an insect, may have been gradually built up by natural selection alone. But just in proportion as structures or organs are not merely thus of passive use (where, of course, the Lamarckian principles cannot obtain), but require to be actively used, in that proportion does it become difficult to understand the incipient construction of them by natural selection alone. Therefore, in many such cases, if the incipient construction is not to be explained by the Lamarckian principles, it is difficult to see how it is to be explained at all.
Furthermore, since the question as to the transmission of acquired characters stands now exactly as it did after the publication of Mr. Galton's Theory of Heredity twenty years ago, it would seem that our judgement with regard to it should remain exactly what it was then. Although we must "out-Darwin Darwin" to the extent of holding that he assigned too large a measure of intensity to the Lamarckian factors, no sufficient reason has been shown for denying the existence of these factors in toto; while, on the other hand, there are certain general considerations, and certain particular facts, which appear to render it probable that they have played a highly important part in the process of organic evolution as a whole. At the same time, and in the present state of our information, this judgement must be deemed provisional, or liable eventually to be overturned by experimental proof of the non-inheritance of acquired characters. But, even if this should ever be finally accomplished, the question would still remain whether the principle of natural selection alone is capable of explaining all the facts of adaptation; and, for my own part, I should then be disposed to believe that there must be some other, though hitherto undiscovered, principle at work, which co-operates with natural selection, by playing the subordinate role which was assigned by Darwin to the principles of Lamarck.
Finally, let it be noted that no part of the foregoing argument is to be regarded as directed against the principle of what Professor Weismann calls "continuity." On the contrary, it appears to be self-evident that this principle must be accepted in some degree or another by every one, whether Darwinians, Neo-Darwinians, Lamarckians, Neo-Lamarckians, or even the advocates of special creation. Yet, to hear or to read some of the followers of Weismann, one can only conclude that, prior to his publications on the subject, they had never thought about it at all. These naturalists appear to suppose that until then the belief of Darwinians was, that there could be no hereditary "continuity" between any one organic type and another (such, for instance, as between Ape and Man), but that the whole structure of any given generation must be due to "gemmules" or "somato-plasm," derived exclusively from the preceding generation. Nothing can show more ignorance, or more thoughtlessness, with regard to the whole subject. The very basis of the general theory of evolution is that there must always have been a continuity in the material substance of heredity since the time when the process of evolution began; and it was not reserved for our generation, or even for our century, to perceive the special nature of this material substance in the case of sexual organisms. No, the real and the sole question, where Weismann's theory of heredity is concerned, is simply this—Are we to hold that this material substance has been absolutely continuous "since the first origin of sexual propagation," always occupying a separate "sphere" of its own, at all events to the extent of never having been modified by the body substance in which it resides (Lamarckian factors); or, are we to hold that this "germ-plasm," "stirp," or "formative-material," has been but relatively continuous, so as to admit of some amount of commerce with body-substance, and therefore to admit of acquired characters, when sufficiently long continued as such, eventually becoming congenital? If this question be answered in the latter sense, of course the further question arises as to the degree of such commerce, or the time during which acquired characters must continue to be acquired in successive generations before they can sufficiently impress themselves on the substance of heredity to become congenital. But this is a subordinate question, and one which, in the present state of our information, it seems to me almost useless to speculate upon. My own opinion has always been the same as that of Mr. Galton; and my belief is that eventually both Weismann and his followers will gravitate into it. It was in order to precipitate this result as far as possible that I wrote the Examination. If it ever should be accomplished, Professor Weismann's elaborate theory of evolution will have had its bases removed.
One of the great changes which has been wrought in biological science by the Darwinian theory of natural selection, consists in its having furnished an intelligible explanation of the phenomena of adaptation. Indeed, in my opinion, this is the most important function which this theory has had to perform; and although we still find systematic zoologists and systematic botanists who hold that the chief merit of Darwin's work consists in its having furnished an explanation of the origin of species, a very little consideration is enough to show that such an idea is but a survival, or a vestige, of an archaic system of thought. So long as species were regarded as due to separate acts of creation, any theory which could explain their production by a process of natural evolution became of such commanding importance in this respect, that we cannot wonder if in those days the principal function of Darwin's work was held to be what the title of that work—The Origin of Species by means of Natural Selection—itself serves to convey. And, indeed, in those days this actually was the principal function of Darwin's work, seeing that in those days the fact of evolution itself, as distinguished from its method, had to be proved; and that the whole proof had to stand or fall with the evidence which could be adduced touching the mutability of species. Therefore, without question, Darwin was right in placing this issue as to the stability or instability of species in the forefront of his generalizations, and hence in constituting it the title of his epoch-making book. But nowadays, when the fact of evolution has been sufficiently established, one would suppose it self-evident that the theory of natural selection should be recognized as covering a very much larger field than that of explaining the origin of species—that it should be recognized as embracing the whole area of organic nature in respect of adaptations, whether these happen to be distinctive of species only, or of genera, families, orders, classes, and sub-kingdoms. For it follows from the general fact of evolution that species are merely arbitrary divisions, which present no deeper significance from a philosophical point of view than is presented by well-marked varieties, out of which they are in all cases believed to have arisen, and from which it is often a matter of mere individual taste whether they shall be separated by receiving the baptism of a specific name. Yet, although naturalists are now unanimously agreed that what they classify as species are nothing more than pronounced—and in some greater or less degree permanent—varieties, so forcible is the influence of traditional modes of thought, that many zoologists and botanists still continue to regard the origin of species as a matter of more importance than the origin of adaptations. Consequently, they continue to represent the theory of natural selection as concerned, primarily, with explaining the origin of species, and denounce as a "heretic" any one who regards the theory as primarily a theory of the origin and cumulative development of adaptations—whether structural or instinctive, and whether the adaptations are severally characteristic of species only or of any of the higher taxonomic divisions. Indeed, these naturalists appear to deem it in some way a disparagement of the theory to state that it is, primarily, a theory of adaptations, and only becomes secondarily a theory of species in those comparatively insignificant cases where the adaptations happen to be distinctive of the lowest order of taxonomic division—a view of the matter which may fitly be compared to that of an astronomer who should define the nebular hypothesis as a theory of the origin of Saturn's rings. It is indeed a theory of the origin of Saturn's rings; but only because it is a theory of the origin of the entire solar system, of which Saturn's rings form a part. Similarly, the theory of natural selection is a theory of the entire system of organic nature in respect of adaptations, whether these happen to be distinctive of particular species only, or are common to any number of species.
Now the outcry which has been raised over this definition of the theory of natural selection is a curious proof of the opposition which may be furnished by habitual modes of thought to an exceedingly plain matter of definition. For, I submit, that no one can deny any of the following propositions; nor can it be denied that from these propositions the foregoing definition of the theory in question follows by way of necessity. The propositions are, first, that natural selection is taken to be the agency which is mainly, if not exclusively, concerned in the evolution of adaptive characters: secondly, that these characters, when evolved, are in some cases peculiar to single species only, while in other cases, and in process of time, they become the common property of many species: thirdly, that in cases where they are peculiar to single species only, they constitute at all events one of the reasons (or even, as the ultra-Darwinians believe, the only reason) why the particular species presenting them have come to be species at all. Now, these being the propositions on which we are all agreed, it obviously follows, of logical necessity, that the theory in question is primarily one which explains the existence of adaptive characters wherever these occur; and, therefore, whether they happen to be restricted to single species, or are common to a whole group of species. Of course in cases where they are restricted to single species, the theory which explains the origin of these particular adaptations becomes also a theory which explains the origin of these particular species; seeing that, as we are all agreed, it is in virtue of such particular adaptations that such particular species exist. Yet even in these cases the theory is, primarily, a theory of the adaptations in virtue of which the particular species exists; for, ex hypothesi, it is the adaptations which condition the species, not the species the adaptations. But, as just observed, adaptations may be the common property of whole groups of species; and thus the theory of natural selection becomes a theory of the origin of genera, of families, of orders, and of classes, quite as much as it is a theory of the origin of species. In other words, it is everywhere a theory of adaptations; and it is only where the adaptations happen to be restricted to single species that the theory therefore and incidentally becomes also a theory of the particular species which presents them. Hence it is by no means the same proposition to affirm that the theory of natural selection is a theory of the origin of species, and that it is a theory of the origin of adaptations, as some of my critics have represented it to be; for these two things are by no means conterminous. And in as far as the two propositions differ, it is perfectly obvious that the latter is the true one.
Possibly, however, it may be said—Assuredly natural selection is a theory of the origin (i.e. cumulative development) of adaptations; and, no less assuredly, although species owe their origin to such adaptations, there is now no common measure between these two things, seeing that in numberless cases the same adaptations are the common property of numberless species. But, allowing all this, we must still remember that in their first beginnings all these adaptations must have been distinctive of, or peculiar to, some one particular species, which afterwards gave rise to a whole genus, family, order, or class of species, all of which inherited the particular adaptations derived from this common ancestor, while progressively gaining additional adaptive characters severally distinctive of their subsequently diverging lines of descent. So that really all adaptive characters must originally have been specific characters; and therefore there is no real distinction to draw between natural selection as a theory of species and as a theory of adaptations.
Well, if this objection were to be advanced, the answer would be obvious. Although it is true that every adaptive character which is now common to a group of species must originally have been distinctive of a single parent species, it by no means follows that in its first beginning as a specific character it appeared in the fully developed form which it now presents as a generic, family, ordinal, or yet higher character. On the contrary, it is perfectly certain that in the great majority of instances such cannot possibly have been the case; and the larger the group of species over which any particular adaptive character now extends, the more evidently do we perceive that this character must itself have been the product of a gradual evolution by natural selection through an innumerable succession of species in branching lines. The wing of a bird, for example, is an adaptive structure which cannot possibly have ever appeared suddenly as a merely specific character: it must have been slowly elaborated through an incalculable number of successive species, as these branched into genera, families, and orders of the existing class. So it is with other class distinctions of an adaptive kind; and so, in progressively lessening degrees, is it with adaptive characters of an ordinal, a family, or a generic value. That is to say, in all cases where an adaptive structure is common to any considerable group of species, we meet with clear evidence that the structure has been the product of evolution through the ancestry of those species; and this evidence becomes increasingly cogent the higher the taxonomic value of the structure. Indeed, it may be laid down as a general rule, that the greater the degree of adaptation the greater is its diffusion—both as regards the number of species which present it now, and the number of extinct species through which it has been handed down, in an ever ramifying extension and in an ever improving form. Species, therefore, may be likened to leaves: successive and transient crops are necessary for the gradual building up of adaptations, which, like the woody and permanent branches, grow continuously in importance and efficiency through all the tree of life. Now, in my view, it is the great office of natural selection to see to the growth of these permanent branches; and although natural selection has likewise had an enormously large share in the origination of each successive crop of leaves—nay, let it be granted to the ultra-Darwinians for the sake of argument, an exclusive prerogative in this respect—still, in my view, this is really the least important part of its work. Not as an explanation of those merely permanent varieties which we call species, but as an explanation of the adaptive machinery of organic nature, which has led to the construction both of the animal and vegetable kingdoms in all their divisions do I regard the Darwinian theory as one of the greatest generalizations in the history of science.