Weismann’s theory of Heredity (1891).

We now proceed to examine Weismann’s theory of germ-plasm, and as this in its various developments has now become a highly complex theory, we had best begin by marking out the lines on which the examination will be conducted.

As I have already pointed out, the Weismannian system is not concerned only with the physiology of reproduction: it is concerned also—and in an even larger measure—with the doctrine of descent. The theory of germ-plasm as a whole is very much more than a theory of heredity; it is a new theory of evolution. The latter, indeed, is deduced from the former; but although the two are thus intimately related, they are nevertheless not mutually dependent. For the relationship is such that the new theory of evolution stands upon the basis supplied by the new theory of heredity, and although it follows from this that if the latter were disproved the former would collapse, it does not follow that if the former were to be found untenable the latter must necessarily be negatived. Hence, for the sake of clearness, and also for the sake of doing justice to both theories, we had best deal with them separately. The present chapter, then, will be devoted to examining Weismann’s theory of heredity, while the ensuing chapter will be concerned with his sequent theory of evolution.

Again, Weismann’s theory of heredity stands on his fundamental postulate—the continuity of germ-plasm; and also on a fact well recognized by all other theories of heredity, which he calls the stability of germ-plasm. But his sequent theory of evolution stands not only on this fundamental postulate, and on this well-recognized fact; it requires for its logical basis two further postulates—viz., that germ-plasm has been perpetually continuous “since the first origin of life,” and unalterably stable “since the first origin of sexual propagation.” That these things are so, a very few words will be sufficient to prove.

Any theory of heredity which supposes the material of heredity to occupy a more or less separate “sphere” of its own, is not obliged further to suppose that this material has always been thus isolated, or even that it is now invariably so. There have been one or two such theories prior to Weismann’s, and they were founded on the well-known fact of congenital characters being at any rate much more heritable than are acquired characters. But it has not been needful for these theories to assume that the “continuity” thus postulated has been perpetually unbroken. Even if it has been frequently to some extent interrupted, all the facts of heredity could be equally well comprised under such theories—and this even if it be supposed that acquired characters are but rarely, or never, transmitted to progeny. For, in as far as the continuity may have been interrupted, it does not follow that the acquired characters (body-changes), which by hypothesis caused the interruption, must be inherited by progeny exactly as they occurred in the parents. Or, in other words and adopting Weismann’s terminology, so far as the facts of heredity are concerned, there is no reason why germ-plasm should not frequently have had its hereditary qualities modified by some greater or less degree of commerce with somatic-tissues, and yet never have reproduced in progeny the identical acquired characters which caused the modification of germ-plasm in the parents: some other and totally different characters might with equal—or even more—likelihood have been the result, as we shall see more clearly a few pages further on. Why, then, does Weismann so insist upon this continuity of germ-plasm as perpetual “since the origin of life”? It appears to me that his only reason for doing so is to provide a basis, not for his theory of heredity, but for his additional theory of evolution. It is of no consequence to the former that germ-plasm should be regarded as thus perpetual, while it is of high importance to the latter that the fundamental postulate of continuity should be supplemented by this further postulate of the continuity as thus perpetual.

Similarly as regards the postulate of the stability of germ-plasm as absolute. It is enough for all the requirements of Weismann’s theory of heredity that the material basis of heredity should present a merely high degree of stability, such as the facts of atavism, degeneration, &c. abundantly prove that it possesses. For his sequent theory of evolution, however, it is necessary to postulate this stability as absolute “since the first origin of sexual reproduction.” Otherwise there would be no foundation for any of the distinctive doctrines which go to constitute this theory.

It may not be immediately apparent that Weismann’s theory of heredity is not per se concerned with either of these two additional postulates of the continuity of germ-plasm as perpetual, and the stability of germ-plasm as absolute; while both are logically necessary to his further theory of evolution. On this account, and also for the sake of clearness in all that is to follow, we had best begin by comparing his theory of heredity with those of his principal predecessors—Darwin and Galton.

For the purposes of this comparison we may start by again alluding to the fact, that even in the multicellular organisms reproduction is not confined to the sexual methods. Many kinds of invertebrated animals will reproduce entire organisms from the fragments into which a single organism has been chopped: plants of various kinds can be propagated indefinitely by cuttings, grafts, and buds, or even by leaves, as we have already observed in Chapter I. Now, when the whole organism is thus reproduced from a severed portion of somatic-tissue, it reproduces its sexual elements. Whence, then, in such cases are these elements derived? Obviously they are not derived immediately from the sexual organs—or even from the sexual cells—of their parents: they are derived from the somatic-cells of a single parent, if we choose to retain this term; and therefore, as Strasburger pointed out soon after Weismann’s theory was published, it seems as if such facts are in themselves destructive of the theory. How, then, does Weismann meet them? As we have already seen in Chapter II, he meets them in the only way they can be met on the lines of his theory—viz., by those newer amendments of his theory which suppose that in all these cases the germ-plasm is not confined to the specially sexual cells, but occurs also in the nuclear substance of those somatic-cells which thus prove themselves capable of developing into entire organisms. In other words, the sexual elements which develop during what I have previously called this “somatic reproduction” of multicellular organism, are supposed to be derived from the sexual cells of ancestors, not indeed immediately (for this they plainly are not), but mediately through the somatic-tissues of their a-sexual parent. Now, in view of this extension, the theory of germ-plasm becomes somewhat closely allied to that of pangenesis. For example, when the fragment of a leaf of Begonia is laid upon moist soil, there strikes root, and grows a new Begonia plant capable of sexual reproduction, Darwin supposes the explanation to be that what he calls “formative material” occurs in all cells of the leaf, while Weismann supposes the explanation to be that what he calls “germ-plasm “ occurs in all—or at any rate in most—of the cells of the leaf. So that, except as regards the terms employed, the two theories are identical in their mode of viewing this particular class of phenomena.

Moreover by thus allowing, in his second essay on Heredity, that germ-plasm need not be restricted to the specially sexual cells, but in some cases, at any rate[12], may occur distributed in full measure of reproductive efficiency throughout the general tissues of the organism, Weismann cannot refrain from taking the further step of supposing that the germ-plasm, like the gemmules of Darwin, is capable of any amount of multiplication in the general cellular tissues of plants—seeing that plants can be propagated by cuttings, buds, &c., indefinitely. And this, as we have seen, Professor Weismann, in his second essay, does not shrink from doing. Moreover, although I cannot remember that he has anywhere expressly said so, it is obvious that the allied phenomena of regeneration and repair admit of explanation by his hypothesis of “ontogenetic grades,” after the manner already stated in Chapter II. Indeed, it is evident that in no other way can these phenomena be brought within the range of his theory. But from this it follows that not only in the case of organisms which are capable of somatic reproduction is the formative nucleo-plasm (idio-plasm-B) diffused throughout the somatic-tissues: on the contrary, it must be universally diffused throughout all the somatic-cells of all living organisms; and whether as it there occurs it is capable of reproducing entire organisms, single organs, single tissues, or a mere cicatrix, depends only on the “ontogenetic grade” of differentiation which this diffused nucleo-plasm has (or has not) previously undergone. Moreover, as we have already seen, at whatever ontogenetic grade of differentiation it may be present in a given somatic-tissue, it must there be capable of indefinite self-multiplication. Therefore, in all these respects this “formative nucleo-plasm” (or idio-plasm-B) of Weismann precisely resembles the “formative material” (or gemmules) of Darwin.

Lastly, as De Vries has pointed out[13], there must be at least as many divisions and subdivisions in the substance of germ-plasm, as there are differences between the somatic organs, tissues, and even cells, to which germ-plasm eventually gives rise—no matter through how many ontogenetic grades of idio-plasm it may first have to pass. Or, in other words, we must accept, as the material basis of heredity, ultimate particles[14] of germ-plasm, which are already differentiated into as many diverse categories as there are differences between all the constituent parts of the resulting soma; for, as shown in the Appendix, no change in the facts of the case has been shown by simply changing the original term “germ-plasm” into “idio-plasm,” wherever the phenomena of ontogeny are concerned. It may be convenient, for the sake of presenting newer additions to the theory, to restrict the term “germ-plasm” to “idio-plasm of the first ontogenetic stage”; but as idio-plasms of all subsequent ontogenetic stages are supposed to be ultimately derived from this idio-plasm of the first stage, it is evident that the particulate differences in question must already have been present in the so-called “undifferentiated idio-plasm of the first ontogenetic stage.” Unless we are to have a mere juggling with words, we cannot put into our successive idio-plasms any particles of kinds differing from those which are contained in the original germ-plasm. Therefore I say that, notwithstanding this change of terminology, Weismann must continue to assume, as the material basis of heredity, ultimate particles of germ-plasm which are already differentiated into as many diverse categories as there are differences between the parts of the resulting soma—although, of course, these ultimate particles need not be nearly so numerous in each of their categories as they afterwards become by self-multiplication while forming each of the resulting tissues.

But this is precisely what the theory of pangenesis supposes; so that I see no reason why these ultimate particles of germ-plasm should not be regarded as “gemmules,” so far as their size, number, and function are concerned. In point of fact, they differ from gemmules only in respect to their origin: they are not particles derived from somatic-cells of the preceding generation, but particles derived from germ-plasm of the preceding generation. Or, to state the difference in another form, if we regard the sexual elements as constituting the physiological centre of the organism, then the theory of germ-plasm supposes these ultimate carriers of heredity to originate at this centre, and then to travel centrifugally; while the theory of pangenesis supposes them to originate at the periphery, and then to travel centripetally.

This point of difference, however, arises from the deeper ones, which—having now exhausted the points of agreement—we must next proceed to state.

If, as we have seen, “formative material” and “germ-plasm” agree in being particulate; in constituting the material basis of heredity; in being mainly lodged in highly specialized, or germinal, cells; in being nevertheless also distributed throughout the general cellular tissues, where they are alike concerned in all processes of regeneration, repair, and a-sexual reproduction; in having an enormously complex structure, so that every constituent part of the future organism is already represented in them by corresponding particles; in being everywhere capable of a virtually unlimited multiplication, without ever losing their hereditary endowments; in often carrying these endowments in a dormant state through a number of generations, until at last they reappear again in what we recognize as reversions to ancestral characters;—if in all these most important respects the two substances are supposed to be alike, it may well appear at first sight that there is not much room left for any difference between them. And, in point of fact, the only difference that does obtain between them admits of being stated in two words,—Continuity, and Stability. Nevertheless, although thus so few in number, these two points of difference are points of great importance, as I will now proceed briefly to show.

If the substance which constitutes the material basis of heredity has been perpetually continuous, in the sense of never having had any of its hereditary endowments in any way affected by the general body-tissues in which it resides, the following important consequences, it will be remembered, arise. The process of organic evolution must have been exclusively due to a natural selection of favourable variations occurring within the limits of this substance itself; and therefore the so-called Lamarckian factors can never have played any part at all in the evolution of any but the unicellular organisms. On the other hand, if this substance has not been thus perpetually continuous, but more or less formed anew at each ontogeny by the general body-tissues in which it resides, natural selection has probably been in some corresponding degree assisted in its work of organic evolution by the Lamarckian factors, with the result that the experiences of parents count for something in the congenital endowments of their offspring. So much for the first of the two differences between germ-plasm and gemmules, or the difference which arises from the perpetual continuity of germ-plasm.

Touching the second difference, or that which arises from the absolute stability of germ-plasm, it will be remembered how from this character there arises another important chain of consequences. Namely, individual variations of the congenital kind can only be due to admixtures of different masses of germ-plasm in every act of sexual fertilization; natural selection is therefore dependent, for the possibility of its working, upon the sexual methods of propagation; hence, natural selection is without any jurisdiction among the unicellular organisms, where the Lamarckian factors hold exclusive sway; and hence, also, the multicellular organisms are ultimately dependent upon this absolute stability of their germ-plasm for all the progress which they have made in the past, as well as for any progress which they may be destined to make in the future.

Thus we see that the two points of difference between germ-plasm and gemmules are not merely of great importance as regards the particular problem which is presented by the phenomena of heredity: they are of still greater importance as regards the general theory of evolution. For if these two qualities of perpetual continuity and absolute stability can be proved to belong to the material basis of heredity, the entire theory of evolution will have to be reconstructed from its very foundation—and this quite apart from the more special question as to the transmission of acquired characters. Therefore we shall presently have to consider these two alleged qualities with the care that they demand, as having been seriously suggested by so eminent a naturalist as Professor Weismann. But, before proceeding to do so, I must briefly compare his theory with that of Mr. Galton.

“Stirp” resembles both “germ-plasm” and “gemmules” in all the respects which have above been named as common to the two latter. But it differs from gemmules and further resembles germ-plasm in all the following particulars. It is derived from the stirp of proceeding generations, and constitutes the sole basis of heredity. Only a part of it, however, is consumed in each ontogeny—the residue being handed over to “contribute to form the stirps of the offspring,” where it undergoes self-multiplication at the expense of the nutriment supplied to it from the somatic system of the offspring, and so on through successive generations. Again, stirp is concerned in all processes of regeneration and repair, in the same centrifugal manner as germ-plasm is so concerned. Furthermore, the influence of sexual propagation in the blending of hereditary qualities of the stirp is recognized, while the principle of panmixia, or the cessation of selection, is entertained, and shown to invalidate the evidence of pangenesis which Darwin derived from the apparently transmitted effects of use and disuse in our domesticated animals[15]. Lastly, it is clearly stated that on the basis supplied by this “theory of heredity,” it becomes logically possible to dispense with the Lamarckian principles in toto, leaving natural selection as the sole known cause of organic evolution through a perpetual continuity of stirp, together with individual variations of the same, whether by sexual admixture or otherwise.

So far, then, there is not merely resemblance, but virtual identity, between the theories of stirp and germ-plasm. Disregarding certain speculative details, the coincidence is as complete as that between a die and its impress. But although the two theories are thus similar in logical construction, they differ in their interpretations of biological fact. That is to say, although Galton anticipated by some ten years all the main features of Weismann’s theory of heredity[16], and showed that, as a matter of form, it was logically intact, he refrained from concluding on this account that it must be the true theory of heredity. He argued, indeed, that in the main it was probably the true theory; but he guarded his presentation of it by not undertaking to deny that there might still be some degree of intercommunication between the material basis of heredity in stirp, and the somatic tissues of successive organisms. The construction of a theory which, as a matter of theory, could dispense with the Lamarckian principles in toto, was seen to be a very different thing from proving, as a matter of fact, that these principles are non-existent—and this, even though it was seen that a recognition of the principle of panmixia must be taken to have considerably attenuated the degree of their operation as previously estimated by Darwin in the theory of pangenesis. In short, after pointing out that the doctrine of stirp might very well adopt the position which about a decade later was adopted by the doctrine of germ-plasm—namely, that of altogether supplanting the doctrine of gemmules,—Galton allowed that this could be done only as a matter of formal speculation; and that, as a matter of real interpretation of the facts of nature, it seemed more judicious to stop at modifying the doctrine of gemmules, by provisionally retaining the hypothesis of gemmules, but assigning to their agency a greatly subordinate rÔle. Or to quote his own words:—

The conclusion to be drawn from the foregoing arguments is, that we might almost reserve our belief that the structural [i. e., “somatic”] cells can react on the sexual elements at all, and we may be confident that at the most they do so in a very faint degree; in other words, that acquired modifications are barely, if at all, inherited, in the correct sense of that word. If they were not heritable, then the second group of cases [i.e., those of acquired as distinguished from congenital characters] would vanish, and we should be absolved from all further trouble; if they exist, in however faint a degree, a complete theory of heredity must account for them. I propose, as already stated, to accept the supposition of their being faintly heritable, and to account for them by a modification of Pangenesis[17].

Seeing, then, that Galton did not undertake to deny a possibly slight influence of somatic-tissues on the hereditary qualities of stirp, it follows that he did not have to proceed to those drastic modifications of the general theory of descent which Weismann has attempted. Stirp, like germ-plasm, is continuous; but, unlike germ-plasm, it is not necessarily or absolutely so. Again, stirp, like germ-plasm, is stable; yet, unlike germ-plasm, it is not perpetually or unalterably so. Hence we hear nothing from Galton about our having to explain the unlikeness of our children to ourselves by variations in our protozoan ancestors; nor do we meet with any of those other immense reaches of deductive speculation which, in my opinion, merely disfigure the republication of stirp under the name of germ-plasm.

Now, I allude to these, the only important points of difference between stirp and germ-plasm, for the sake of drawing prominent attention to the fact that it makes a literally immeasurable difference whether we suppose the material basis of heredity to be perpetually continuous and unalterably stable, or whether we suppose that it is but largely continuous and highly stable. In the former case, all the far-reaching deductions which Weismann draws with reference to the general theory of descent—or apart from the more special problem of heredity—follow by way of logical consequence. In the latter case, there is no justification for any such deductions. For, no matter how faintly or how fitfully the hereditary qualities of the material in question may be modified by the somatic-tissues in which it resides, or by the external conditions of life to which it is exposed, these disturbances of its absolute stability, and these interruptions of its perpetual continuity, must cause more or less frequent changes on the part of its hereditary qualities—with the result that specific or other modifications of organic types need not have been solely due to the varying admixture of such material in sexual unions on the one hand, or to the unassisted power of natural selection on the other. Numberless additional causes of individual variation are admitted, while the Lamarckian principles are still allowed some degree of play. And although this is a lower degree than Darwin supposed, their influence in determining the course of organic evolution may still have been enormous; seeing that their action, in whatever measure it may be supposed to obtain, must always have been cumulative on the one hand, and directive of variations in adaptive lines on the other. Or, as Galton himself observes, in the passage already quoted, “if they exist, in however faint a degree, a complete theory of heredity must account for them.” He saw, indeed, that a most inviting logical system could be framed by denying that they can ever exist in any degree—or, in other words, by supposing that stirp was exactly the same as what was afterwards called germ-plasm, in that it always occupied a separate “sphere” of its own, where its continuity has been uninterrupted “since the first origin of life.” But Galton was not seduced by the temptation to construct an ideally logical system; and he had what I regard as the sound judgement to abstain from carrying his theory of stirp into any such transcendental “sphere” as that which is occupied by Weismann’s theory of germ-plasm, in relation to the general doctrine of descent.

There is, then, a vast distinction between any theory of heredity which postulates the material of heredity as highly stable and largely continuous, and Weismann’s theory, which postulates this material as absolutely stable and perpetually continuous. But we must next take notice that Weismann himself has not kept this distinction in view with the constancy which we should have expected from so forcible a thinker. On the contrary, although in the construction of his theory of evolution he never fails to press the postulates of absolute stability and perpetual continuity to their logical conclusions in the various doctrines above enumerated (pp. 57-58), when engaged on his more special theory of heredity he every now and then appears to lose sight of the distinction. Indeed, he occasionally makes such large concessions with regard to both these postulates, that, were they to be entertained, the occupation of his critics would be gone: his theory of heredity would become converted into Galton’s, while his theory of evolution would vanish altogether. It is therefore necessary to quote some of these concessions, if only to justify ourselves in subsequently ignoring them. I will give one instance of each; but it is necessary to preface the illustrations with a few words to mark emphatically three very distinct cases of congenital variation—leaving aside for the present the question whether or not they all occur in fact, as they are held to do by one or other of the theories of heredity.

1. The case where impoverished nutrition of the body has the effect of simply starving its germinal material. This is not a case where either the continuity or the stability of such material is affected. Its full efficiency as “formative material” may indeed be thus deteriorated to any extent, so that the progeny may be to any extent puny or malformed; but this will not necessarily cause any such re-shuffling of its “molecules” as will thereafter result in a permanent phylogenetic change. At most it will affect only the immediate offspring of poorly nourished parents; and natural selection will always be ready to eliminate such inefficient individuals. This case I will always hereafter call the case of nutritive congenital changes.

2. The case where germinal material is influenced by causes which do effect a re-shuffling of its “molecules,” so that a permanent phylogenetic change does result. Observe, in this case, it does not signify whether the causes arise from external conditions of life, from any action of the soma on its own germinal material, or from so-called “spontaneous” changes on the part of such material itself. But the one cause which has not been concerned in producing an hereditary modification of this class is the mixture of “germ-plasms” in an act of sexual union. In hereafter speaking of this case I will follow Weismann’s terminology, and call congenital changes thus produced specialized congenital changes.

3. Lastly, we have the case of the Lamarckian factors. This precisely resembles case 2, save that the congenital changes produced are still more “specialized.” For while in the preceding case the re-shuffling before mentioned may have produced a congenital change of any kind, in the present case the congenital change produced must be of one particular kind—viz., a reproduction by heredity of the very same modification which occurred in the parents. “The fathers have eaten sour grapes, and the children’s teeth are set on edge.” This would be an extreme example of “use-inheritance,” and so of case 3. But if the fathers had eaten sour grapes, and the children, instead of having their teeth set on edge, were to be born with a wryneck or a squint, then we should have a good example of case 2. In order, then, to mark the important distinction between these two cases, I will hereafter call the highly specialized changes due to the Lamarckian factors—supposing such changes to be possible—representative congenital changes.

These several distinctions being understood, I will proceed to furnish the two quotations from Weismann, which are respectively illustrative of his concessions touching his two fundamental postulates, as previously explained.

We may fairly attribute to the adult organism influences which determine the phyletic development of its descendants. For the germ-cells are contained in the organism, and the external influences which affect them are intimately connected with the state of the organism in which they lie hid. If it be well nourished, the germ-cells will have abundant nutriment; and, conversely, if it be weak and sickly, the germ-cells will be arrested in their growth. It is even possible that the effects of these influences may be more specialized; that is to say, they may act only upon certain parts of the germ-cells. But this is indeed very different from believing that the changes of the organism which result from external stimuli can be transmitted to the germ-cells, and will re-develop in the next generation at the same time as that at which they arose in the parent, and in the same part of the organism[18].

It will be perceived that Weismann himself here very clearly draws all the distinctions between cases 1, 2, and 3, as above explained. Therefore it becomes the more remarkable that he should not have perceived how radically inconsistent it is in him thus to entertain as “possible” congenital variations belonging to the case 2. For, as we have now so fully seen, the theory of germ-plasm (as distinguished from that of stirp) cannot entertain the possibility of an hereditary and specialized change of any kind as thus produced by external conditions of life: should such a possibility be entertained, there must obviously be an end to the absolute stability of germ-plasm, and a consequent collapse of Weismann’s theory of evolution. Either germ-plasm is absolutely stable, or else it is but highly stable. If it is absolutely stable, individual variations of an hereditary kind can occur only as results of sexual admixtures of germ-plasm, and Weismann’s theory of evolution is established. But if germ-plasm is not absolutely stable (no matter in how high a degree it may be so) hereditary individual variations may be produced by other causes, and Weismann’s theory of evolution collapses. Therefore, if we are to examine his theory of evolution, we can do so only by ignoring such a passage as the one just quoted, which surrenders the postulate of the absolute stability of germ-plasm.

Again, if we are to examine Weismann’s theory of heredity, we must similarly ignore such a passage as the following, where he represents that he is similarly prepared to surrender his still more fundamental postulate of the perpetual continuity of germ-plasm.

After remarking that some of his own experiments on the climatic varieties of certain butterflies raise such difficulties against his whole theory of heredity that even now “he cannot explain the facts otherwise than by supposing the passive acquisition of characters produced by the direct influence of climate,” he goes on to remark more generally—“We cannot exclude the possibility of such a transmission occasionally occurring, for, even if the greater part of the effects must be attributed to natural selection, there might be a smaller part in certain cases which depends on this exceptional factor[19]”—i.e., the Lamarckian factor!

Now, it must be particularly noted that in this passage Weismann is speaking, not as in the previous passage, of specialized congenital characters, but of representative congenital characters. In other words, he here entertains the possibility which in the passage previously quoted he very properly rejects—namely, “that changes of the organism which result from external stimuli can be transmitted to the germ-cells, and will re-develop in the next generation at the same time as that at which they arose in the parent, and in the same part of the organism.” But it is evident that if the theory of germ-plasm is undermined by the concession made in the passage thus previously quoted, in the passage last quoted a match is put to the fuse. It does not signify whether the particular case of the butterflies in question will ever admit of any other explanation more in accordance with the theory of germ-plasm: the point is that in no case can this theory entertain the possibility of causes other than admixtures of germ-plasm in sexual unions producing hereditary changes, (A) of any kind, (B) still less of a specialized kind, and (C) least of all of a representative kind. For the distinguishing essence of this theory is, that germ-plasm must always have moved, so to speak, in a closed orbit of its own: its “sphere” must have been perpetually distinct from those of whatever other “plasms” there may be in the constellations of living things. So that, in such passages as those just quoted, Weismann is not only destroying the very foundations of his general theory of evolution, but at the same time he is identifying his more special theory of heredity with those which had been already published by his predecessors, and more particularly by Galton. Now, it is not Galton’s theory that we are considering; and therefore we must hereafter ignore those fundamental admissions, whereby Weismann every now and again appears ready to relinquish all that is most distinctive of, or original in, his own elaborate system of theories.

It is, indeed, impossible not to admire the candour of these admissions, or to avoid recognizing the truly scientific spirit which they betoken. But, at the same time, one is led to doubt whether in making them Professor Weismann has sufficiently considered their full import. He appears to deem it of comparatively little importance whether or not acquired characters can sometimes and in some degrees influence the hereditary qualities of germ-plasm, provided he can show that much the larger part of the phenomena of heredity must be ascribed to the continuity of germ-plasm. In other words, he seems to think that it matters but little whether in the course of organic evolution the Lamarckian factors have played but a very subordinate part, or whether they have not played any part at all. Moreover, I have heard one or two prominent followers of Weismann give public expression to the same opinion. Therefore I must repeat that it makes a literally immeasurable difference whether we suppose, with Galton, that the Lamarckian factors may sometimes and in some degrees assert themselves, or whether we suppose, with the great bulk of Weismann’s writings and in accordance with the logical requirements of his theory, that they can never possibly occur in any degree. The distinctive postulate of his theory of heredity, and one of the two fundamental doctrines on which he founds his further theory of evolution, is, that the physiology of sexual reproduction cannot admit of any inversion of the relations between “germ-plasm” and “somatic idio-plasm[20].” This is a perfectly intelligible postulate, but it is not one with which we may play fast and loose. Either there is such a physiological mechanism as it announces, in which case the relations in question can never be inverted “occasionally,” any more than rags may “occasionally” help to construct the mill which is to form them into paper;—or else there is no such mechanism, in which case we may have to do with gemmules, physiological units, stirp, micellae, pangenes, plastidules, or any of the other hypothetical “carriers of heredity” to which our predilections may happen to incline; but the one substance with which we certainly have not to do is germ-plasm[21].

After these tedious but necessary preambles, we may now proceed to examine Professor Weismann’s postulate as to the perpetual continuity of germ-plasm, with its superstructure in his theory of heredity—reserving for the next chapter our examination of his further postulate touching the absolute stability of germ-plasm, with its superstructure in his theory of evolution.

The evidence which Weismann has presented in favour of his fundamental postulate of the perpetual continuity of germ-plasm may be conveniently dealt with under two heads—namely, indirect evidence as derived from general reasoning, and direct evidence derived from particular facts.

The general reasoning is directed to show, (1) that there is no evidence of the transmission of acquired characters; (2) that the theory of pangenesis is “inconceivable”; and, (3) that the alternative theory of germ-plasm is amply conceivable. Now, to the best of my judgement, not one of these propositions is borne out by the general reasoning in question. But as the latter is almost entirely of an a priori character, and also of a somewhat abstruse construction, I think the patience of any ordinary reader will be saved by relegating this part of our subject to an Appendix. Therefore, remarking only that any one who cares to look at Appendix I ought, in my opinion, to perceive that there is no real evidence against the transmission of acquired characters to be derived from Weismann’s general reasoning in this connexion, I will at once proceed to consider the evidence which he has adduced in the way of particular facts.

In the first place, as one result of his brilliant researches on the Hydromedusae, he has found that the generative cells occur only in certain localized situations, which, however, vary greatly in different species, though they are always constant for the same species. He has also found that the varying situations in different species of the localized or generative areas correspond, place for place, with successive stages in a process of gradual transposition which has occurred in the phylogeny of the Hydromedusae. Lastly, he has found that in each ontogeny these successive stages of transposition are repeated, with the result that during the individual lifetime of one of these animals the germ-cells migrate through the body, from what used to be their ancestral situation to what is now the normal situation for that particular species. Such being the facts, Weismann argues from them that the germ-cells of the Hydromedusae are thus proved to present properties of a peculiar kind, which cannot be supplied by any of the other cells of the organism; for, if they could, whence the necessity for this migration of these particular cells? Of course it follows that these peculiar properties must depend on the presence of some peculiar substance, and that this is none other than the “germ-plasm,” which here exhibits a demonstrable “continuity” throughout the entire phylogeny of these unquestionably very ancient Metazoa.

The second line of direct evidence in favour of the continuity of germ-plasm which Weismann has adduced is, that in the case of some invertebrated animals the sexual apparatus is demonstrably separated as reproductive cells (or cells which afterwards give rise to the reproductive glands) at a very early period of ontogeny—so early indeed, in certain cases, that this separation constitutes actually the first stage in the process of ontogeny. Therefore, it is argued, we may regard it as antecedently improbable that the after-life of the individual can in any way affect the congenital endowments of its ova, seeing that the ova have been thus from the first anatomically isolated from all the other tissues of the organism.

The third and only other line of direct evidence is, that organisms which have been produced parthenogenetically, or without admixture of germ-plasms in any previous act of sexual fertilization, do not exhibit congenital variations.

Taking, then, these three lines of verification separately, none of them need detain us long. For although the fact of the migration of germ-cells becomes one of great interest in relation to Weismann’s theory after the theory has been accepted, the fact in itself does not furnish any evidence in support of the theory. In the first place, it tends equally well to support Galton’s theory of stirp; and therefore does not lend any special countenance to the theory of germ-plasm—or the theory that there cannot now be, and never can have been, any communication at all between the plasm of the germ and that of the soma. In the second place, the fact of such migration is not incompatible even with the theory of pangenesis, or the theory which supposes such a communication to be extremely intimate. There may be many other reasons for this migration of germ-cells besides the one which Weismann’s theory supposes. For example, the principle of physiological economy may very well have determined that it is better to continue for reproductive purposes the use of cells which have already been specialized and set apart for the execution of those purposes, than to discard these cells and transform others into a kind fitted to replace them. Even the theory of pangenesis requires to assume a very high degree of specialization on the part of germ-cells; and as it is the fact of such specialization alone which is proved by Weismann’s observations, I do not see that it constitutes any criterion between his theory of heredity and that of Darwin—still less, of course, between his theory and that of Galton. Lastly, in this connexion we ought to remember that the Hydromedusae are organisms in which the specialization in question happens to be least, as is shown by the fact that entire individuals admit of being reproduced from fragments of somatic-tissues; so that these are organisms where we would least expect to meet with the migration of germ-cells, were the purpose of such migration that which Weismann suggests. This line of evidence therefore seems valueless.

Nor does it appear to me that the second line of evidence is of any more value. In the first place, there is no shadow of a reason for supposing that an apparently anatomical isolation of germ-cells necessarily entails a physiological isolation as regards their special function—all “physiological analogy,” indeed, being opposed to such a view, as is shown in Appendix I. In the second place, there is no proof of any anatomical isolation, as we may likewise see in that Appendix. In the third place, the fact relied upon to indicate such an isolation—viz., the early formation of germ-cells—is not a fact of any general occurrence. On the contrary, it obtains only in a comparatively small number of animals, while it does not obtain in any plants. In the Vertebrates, for example, the reproductive cells are not differentiated from the somatic cells till after the embryo has been fully formed; while in plants their development constitutes the very last stage of ontogeny. In the fourth place, the argument, even for what it is worth, is purely deductive; and deductive reasoning in such a case as this—where the phenomena are enormously complex and our ignorance unusually profound—is always precarious. Lastly, in the fifth place, Weismann has now himself abandoned this argument. For in one of his later essays he says:—

Those instances of early separation of sexual from somatic cells, upon which I have often insisted as indicating the continuity of the germ-plasm, do not now appear to be of such conclusive importance as at the time when we were not sure about the localization of the plasm in the nuclei. In the great majority of cases the germ-cells are not separated at the beginning of embryonic development, but only in some of the later stages.... It therefore follows that cases of early separation of the germ-cells afford no proof of a direct persistence of the parent germ-cells in those of the offspring.

The last line of direct evidence, or that derived from the alleged non-variability of parthenogenetic organisms, is, as Professor Vines has shown, opposed to fact. Therefore, in his later writings, Weismann has abandoned this line of evidence also.

Upon the whole,then, we must conclude with regard to the fundamental postulate of perpetual continuity, that there is actually no evidence of a direct kind in its favour. And, as Weismann’s arguments of an indirect kind are dealt with in Appendix I, it remains only to state such evidence per contra as, to the best of my judgement, appears valid.

The fundamental proposition which we have been considering, and to the further consideration of which we have now to proceed, is, in effect, that germ-plasm differs from stirp in having been perpetually restricted to a “sphere” of its own, “since the first origin of life.” Criticism, therefore, must be directed to show that the “sphere” in question has not been proved so entirely independent as this fundamental proposition sets forth; but that, on the contrary, there appears to be a certain amount of reciprocal action between this sphere and that of the somatic-tissues—even though we may agree (as I myself agree) with Galton in holding that the degree of such reciprocal action is neither so intimate nor so constant as it was held to be by Darwin. This, indeed, is the direction which the course of our criticism has taken already. For it has just been shown that Weismann has failed to adduce any facts (preceding text) or considerations (Appendix I) in support of his fundamental proposition as above stated, save such as proceed on a prior acceptance of the proposition itself. The facts and considerations which he has adduced are therefore useless as evidence in support of this proposition, although they would admit of being explained by it supposing it to have been already substantiated by any facts or considerations of an independent kind. Which is merely another way of saying, as already said, that there is no evidence in favour of the proposition.

But I am now about to argue that there is evidence against the proposition. For I am about to argue, not only as heretofore that for anything Weismann has shown to the contrary there may be a certain amount of reciprocal action between the sphere of germinal-substance and the sphere of body-substance; but that, as a matter of fact, there is a certain amount of such reciprocal action.

Without laying undue stress on the intimate “correlation” that subsists between the reproductive organs and all other parts of the organism, I nevertheless think that the fact ought here to be noted. For the changes which occur at puberty and after the reproductive functions have ceased, as well as those which may be artificially produced by castration, &c., prove at any rate some extremely important association between the soma as a whole and its reproductive apparatus as a whole. No doubt it may properly enough be answered that this proof does not extend to the vital point of showing the association to be between the soma as a whole, and that particular part of the reproductive apparatus in which the “carriers of heredity” reside—namely, the ova and spermatozoa; and, therefore, that the facts in question may be due only to some changed conditions of nutrition on the part of the somatic-tissues which these alterations on the part of the reproductive glands entail. On this account we must fully allow that the facts in question are not in themselves of any conclusive weight; but I think they are worth mentioning, because they certainly seem to countenance the theory which supposes some reciprocal influence as exercised by the germinal elements on the somatic-tissues and vice versa, rather than they do the theory which supposes the germinal elements and the somatic-tissues to have always occupied totally different “spheres.”

Here, however, is a stronger class of facts. It has not unfrequently been observed, at any rate in mammals, that when a female has borne progeny to a male of one variety, and subsequently bears progeny to a male of another variety, the younger progeny presents a more or less unmistakable resemblance to the father of the older one. Now, this is a fact to which Weismann has nowhere alluded; and therefore I do not know how he would meet it. But, as far as I can see, it can be explained only in one or other of two ways. Either there must be some action of the spermatic element on the hitherto unripe ovum, or else this element must exercise some influence on the somatic-tissues of the female, which in their turn act upon the ovum[22]. Now, I do not deny that the first of these possibilities might be reconcilable with the hypothesis of an absolute continuity of germ-plasm; for it is conceivable that the life of germ-plasm is not coterminous with that of the spermatozoa which convey it, and hence that, if the carriers of heredity, after the disintegration of their containing spermatozoa, should ever penetrate an unripe ovum, the germ-plasm thus introduced might remain dormant in the ovum until the latter becomes mature, and is then fertilized by another sire. In this way it is conceivable that the hitherto dormant germ-plasm of the previous sire might exercise some influence on the progeny of a subsequent one. But it seems clear that the second of the two possibilities above named could not be thus brought within the hypothesis of an absolute continuity of germ-plasm. Therefore it seems that the school of Weismann must adopt the first, to the exclusion of the second. Unfortunately for them, however, there is another (and clearly analogous) fact, which goes to exclude the first possibility, and most definitely to substantiate the second. For, in the case of plants, where there can be no second progeny borne by the same “ovary,” but where we happen to be able to see that a marked effect is sometimes produced on the somatic-tissues of the mother by the pollen of the father, there can be no question as to the male element being able to exercise a direct influence on the soma of the female. Consequently, whatever we may think with regard to the case of animals, the facts with regard to plants are in themselves enough to sustain the only position with which we are concerned—viz., that the male element is capable of directly modifying the female soma.

The facts with regard to plants are these. When one variety fertilizes the ovules of another, not unfrequently the influence extends beyond the ovules to the ovarium, and even to the calyx and flower-stalk, of the mother plant. This influence, which may affect the shape, size, colour, and texture of the somatic-tissues of the mother, has been observed in a large number of plants belonging to many different orders. The details of the matter have already been dealt with by Darwin, in the eleventh chapter of his work on Variation, &c.; and this is what he says. The italics are mine.

The proofs of the action of foreign pollen on the mother-plant have been given in considerable detail, because this action is of the highest theoretical importance, and because it is in itself a remarkable and apparently anomalous circumstance. That it is remarkable under a physiological point of view is clear, for the male element not only affects, in accordance with its proper function, the germ, but at the same time various parts of the mother-plant, in the same manner as it affects the same parts in the seminal offspring from the same two parents. We thus learn that an ovule is not indispensable for the reception of the influence of the male element.

Darwin then proceeds to show that this direct action of the male element on the somatic tissues of another organism is not so rare or anomalous as it at first sight appears; for in the case of not a few flowers it comes into play as a needful preliminary to fertilization. Thus, for instance:—

GÄrtner gradually increased the number of pollen grains until he succeeded in fertilizing a Malva, and has proved that many grains are first expended in the development, or, as he expresses it, in the satiation, of the pistil and ovarium. Again, when one plant is fertilized by a widely distinct species, it often happens that the ovarium is fully and quickly developed without any seeds being formed; or the coats of the seeds are formed without any embryo being developed therein.

So much, then, in proof of the direct action of the male element on the somatic-tissues of another organism. It remains to show that a similar action may be exercised by this element on the somatic tissues of its own organism. This has been proved by Hildebrand, who found “that in the normal fertilization of several Orchideae, the action of the plant’s own pollen is necessary for the development of the ovarium; and that this development takes place not only long before the pollen tubes have reached the ovules, but even before the placentae and ovules have been formed”; so that with these orchids the pollen acts directly on their own ovaria, as a preliminary to the formation of the ovules which are subsequently to be fertilized.

It is to be regretted that Professor Weismann has not given us his opinion upon this whole class of facts, for assuredly they appear directly to contradict his theory. The theory is, “that the germ-plasm and the somato-plasm have always occupied different spheres”: the fact is, that the germ-plasm may directly act upon the somato-plasm, both within and beyond the limits of the same organism.

Hitherto we have been considering certain very definite facts, which seem to prove that the germinal elements are able directly to affect the somatic-tissues. We have next to consider such facts as seem to prove the opposite side of a reciprocal relationship—viz., that the somatic-tissues are able directly to affect the germinal elements.

And here there are two distinct lines of evidence to be distinguished.

Firstly, in certain cases—exceptional it is true, but this does not signify—somatic-tissues have been found capable of modifying the hereditary endowments of germinal elements by means of simple grafting. This line of evidence has also been disregarded both by Weismann and his followers; but it is nevertheless an important one to consider. For, if it be the case that the somatic-tissues of an organism A, by being merely grafted on-those of organism B, can so affect the germinal elements of B as to cause their offspring to resemble A—or, contrariwise, if the somatic-tissues of A can thus act on B—then, although it may not be properly said that any “acquired characters” have been transmitted from A to the progeny of B, (or vice versa,) such an a-sexual transmission of alien characters, in its relation to the theory of germ-plasm, is scarcely less awkward than are certain facts which they appear to prove.

Secondly, that acquired characters may be transmitted to progeny by the more ordinary methods of sexual propagation (Lamarckian factors). This second line of evidence will be fully and independently dealt with in future chapters, specially devoted to the subject. Therefore we have here to consider only the first.

Now, the force of this first line of evidence will become apparent, if we reflect that the only way in which the facts can be met by Weismann’s theory, would be by supposing that the somatic germ-plasms which are respectively diffused through the cellular tissues of the scion and the graft become mixed in some such way as they might have been, had the hybrid been due to seminal propagation instead of to simple grafting. But against this, the only interpretation of the facts which is open to the theory, there lies the following objection, which to me appears insuperable.

Where sexual cells are concerned there is always a definite arrangement to secure penetration of the one by the other, and we can see the necessity for such an arrangement in order to effect an admixture of their nuclear contents, where alone germ-plasm is supposed by Weismann’s theory to reside. But in tissue-cells, which have not been thus specialized, it would be difficult to believe that nuclear contents can admit of being intimately fused by a mere apposition of cell-walls. For not only are the nuclear contents of any two such cells thus separated from one another by two cell-walls and two masses of “cytoplasm”; but it is not enough to suppose that in order to produce a graft-hybrid only two of these somatic-cells need mix their nuclear contents, as we know is all that is required in order to produce a seminal hybrid by means of sexual cells. On the contrary, in the former case most, if not all, the somatic-cells which are brought into apposition by the graft must be supposed thus to mix their nuclear contents at the plane of the graft; for otherwise the hybrid would not afterwards present equally the characters of stock and scion. Now, there may be hundreds of thousands of such cells, and therefore it seems impossible that the facts of graft-hybridization can be reconciled with the theory of germ-plasms[23].

The third line of evidence against this theory—i.e., the evidence in favour of the transmission of acquired characters—is to constitute the subject-matter of future chapters. Therefore it will here be sufficient to adduce only one fact of this kind. And I select it because it is one that has been dealt with by Weismann himself. In one of his more recent statements he says:—

The distinguished botanist De Vries has proved that certain constituents of the cell body—e. g., the chromatophores of Algae—pass directly from the maternal ovum to the daughter organism, while the male germ-cells generally contain no chromatophores. Here it appears possible that a transmission of somatogenetic variation has occurred[24].

Now although, as Weismann goes on to observe, “in these lower plants, the separation between somatic and reproductive cells is slight,” in the facts to which he alludes we appear to have good evidence of an influence exercised by somatic cells upon the germinal contents of reproductive cells. And if such an influence is capable of being exercised in the case of “these lower plants,” it follows that there is no such absolute separation between somatic tissues and germ-plasm as Weismann’s theory requires. Moreover it follows that, if the essential distinction between germ-plasm and somato-plasm (or “somatic idio-plasm”) is thus violated at the very foundation of the multicellular organisms, there ceases to be any a priori reason for drawing arbitrary limits, either as to the level of organization at which such “transmission of somatogenetic variation has occurred,” or as to the degree of detail into which it may extend. Both these matters then stand to be tested by observation; and the burden of proof lies with the school of Weismann to show at what level of organization, and at what degree of representation, somatogenetic changes cease to reproduce themselves by heredity.

Passing on, then, to higher levels of organization, and therefore to higher degrees of representation, I shall endeavour to show that this burden of proof cannot be discharged. For I shall endeavour to show, not merely, as just shown, that there ceases to be any a priori reason for drawing arbitrary limits with respect either to levels of organization or to degrees of representation, but that, as a matter of fact, there are no such limits as the passage above quoted assigns. On the contrary, I believe there is as good evidence to prove the not unfrequent transmission of acquired (“somatogenetic”) characters among the higher plants—and even among the higher animals—as there is of the occurrence of this phenomenon in the case of the Alga just mentioned. But in order to do this evidence justice, I shall have to take a new point of departure and consider as a separate question the transmissibility of acquired characters. Meanwhile, and as far as Weismann’s theory of heredity is concerned, it is enough to have shown,—if I have been successful in doing so,—that not only is there no evidence to sustain his fundamental postulate touching the material of heredity having always occupied a separate “sphere” of its own “since the first origin of life”; but that there is good evidence to prove the contrary. For whether or not the reciprocal action of “somato-plasm” and “germ-plasm” can ever proceed to the extent of causing acquired characters to be inherited (so as to produce “representative congenital changes”), all that is distinctive in this theory must be regarded as barren speculation, unless it can be shown that the foregoing facts have failed to prove such a reciprocal action as ever occurring in any lower degree (so as to produce “specialized congenital changes”).