Chapter II

CHARACTERS AS HEREDITARY AND ACQUIRED (PRELIMINARY).

We will proceed to consider, throughout Section I of the present work, the most important among those sundry questions which have come to the front since the death of Darwin. For it was in the year after this event that Weismann published the first of his numerous essays on the subject of Heredity, and, unquestionably, it has been these essays which have given such prominence to this subject during the last decade.

At the outset it is desirable to be clear upon certain points touching the history of the subject; the limits within which our discussion is to be confined; the relation in which the present essay stands to the one that I published last year under the title _An Examination of Weismannism_; and several other matters of a preliminary kind.

The problems presented by the phenomena of heredity are manifold; but chief among them is the hitherto unanswered question as to the transmission or non-transmission of acquired characters. This is the question to which the present Section will be confined.

Although it is usually supposed that this question was first raised by Weismann, such was not the case. Any attentive reader of the successive editions of Darwin's works may perceive that at least from the year 1859 he had the question clearly before his mind; and that during the rest of his life his opinion with regard to it underwent considerable modifications--becoming more and more Lamarckian the longer that he pondered it. But it was not till 1875 that the question was clearly presented to the general public by the independent thought of Mr. Galton, who was led to challenge the Lamarckian factors _in toto_ by way of deduction from his theory of Stirp--the close resemblance of which to Professor Weismann's theory of Germ-plasm has been shown in my _Examination of Weismannism_. Lastly, I was myself led to doubt the Lamarckian factors still further back in the seventies, by having found a reason for questioning the main evidence which Mr. Darwin had adduced in their favour. This doubt was greatly strengthened on reading, in the following year, Mr. Galton's _Theory of Heredity_ just alluded to; and thereupon I commenced a prolonged course of experiments upon the subject, the general nature of which will be stated in future chapters. Presumably many other persons must have entertained similar misgivings touching the inheritance of acquired characters long before the publication of Weismann's first essay upon the subject in 1883. The question as to the inheritance of acquired characters was therefore certainly not first raised by Weismann--although, of course, there is no doubt that it was conceived by him independently, and that he had the great merit of calling general attention to its existence and importance. On the other hand, it cannot be said that he has succeeded in doing very much towards its solution. It is for these reasons that any attempt at dealing with Weismann's fundamental postulate--i.e. that of the non-inheritance of acquired characters--was excluded from my _Examination of Weismannism_. As there stated, he is justified in assuming, for the purposes of his discussion, a negative answer to the question of such inheritance; but evidently the question itself ought not to be included within what we may properly understand by "Weismannism." Weismannism, properly so called, is an elaborate system of theories based on the fundamental postulate just mentioned--theories having reference to the mechanism of heredity on the one hand, and to the course of organic evolution on the other. Now it was the object of the foregoing _Examination_ to deal with this system of theories _per se_; and therefore we have here to take a new point of departure and to consider separately the question of fact as to the inheritance or non-inheritance of acquired characters. At first sight, no doubt, it will appear that in adopting this method I am putting the cart before the horse. For it may well appear that I ought first to have dealt with the validity of Weismann's postulate, and not till then to have considered the system of theories which he has raised upon it. But this criticism is not likely to be urged by any one who is well acquainted with the questions at issue. For, in the first place, it is notorious that the question of fact is still open to question; and therefore it ought to be considered separately, or apart from any theories which may have been formed with regard to it. In the second place, our judgement upon this question of fact must be largely influenced by the validity of general reasonings, such as those put forward in the interests of rival theories of heredity; and, as the theory of germ-plasm has been so thoughtfully elaborated by Professor Weismann, I have sought to give it the attention which it deserves as preliminary to our discussion of the question of fact which now lies before us. Thirdly and lastly, even if this question could be definitely answered by proving either that acquired characters are inherited or that they are not, it would by no means follow that Weismann's theory of heredity would be proved wholly false in the one case, or wholly true in the other. That it need not be wholly true, even were its fundamental postulate to be proved so, is evident, because, although the fact might be taken to prove the theory of Continuity, the theory of Germ-plasm is, as above stated, very much more than this. That the theory of Germ-plasm need not be wholly false, even if acquired characters should ever be proved heritable, a little thought may easily show, because, in this event, the further question would immediately arise as to the degrees and the comparative frequency of such inheritance. For my own part, as stated in the _Examination_, I have always been disposed to accept Mr. Galton's theory of Stirp in preference to that of Germ-plasm on this very ground--i. e. that it does not dogmatically exclude the possibility of an occasional inheritance of acquired characters in faint though cumulative degrees. And whatever our individual opinions may be touching the admissibility of such a _via media_ between the theories of Pangenesis and Germ-plasm, at least we may all agree on the desirability of fully considering the matter as a preliminary to the discussion of the question of fact.

As it is not to be expected that even those who may have read my previous essay can now carry all these points in their memories, I will here re-state them in a somewhat fuller form.

The following diagram will serve to give a clearer view of the sundry parts of Professor Weismann's system of theories, as well as of their relations to one another.

Now, as just explained, the parts of this system which may be properly and distinctively called "Weismannism" are those which go to form the Y-like structure of deductions from the fundamental postulate. Therefore, it was the Y-like system of deductions which were dealt with in the _Examination of Weismannism_, while it is only his basal postulate which has to be dealt with in the following chapters.

So much, then, for the relations of Weismann's system of theories to one another. It is, however, of even more importance that we should gain a clear view of the relations between his theory of _heredity_ to those of Darwin and of Galton, as preliminary to considering the fundamental question of fact.

As we have already seen, the theory of germ-plasm is not only a theory of heredity: it is also, and more distinctively, a theory of evolution, &c. As a theory of heredity it is grounded on its author's fundamental postulate--the _continuity_ of germ-plasm. But as a theory of evolution, it requires for its support this additional postulate, that the continuity of germ-plasm has been _absolute_ "since the first origin of life." It is clear that this additional postulate is not needed for his theory of heredity, but only for his additional theory of evolution, &c. There have been one or two other theories of heredity, prior to this one, which, like it, have been founded on the postulate of Continuity of the substance of heredity; but it has not been needful for any of these theories to postulate further that this substance has been _always_ thus isolated, or even that it is now _invariably_ so. For even though the isolation be frequently invaded by influences of body-changes on the congenital characters of this substance, it does not follow that this principle of Continuity may not still be true _in the main_, even although it is supplemented in some degree by that of use-inheritance. Indeed, so far as the phenomena of heredity are concerned, it is conceivable that all congenital characters were originally acquired, and afterwards became congenital on account of their long inheritance. I do not myself advocate this view as biologically probable, but merely state it as logically possible, and in order to show that, so far as the phenomena of heredity are concerned, there appears to be no reason for Weismann's deduction that the principle of Continuity, if true at all, must be _absolute_. And it would further appear, the only reason why he makes this deduction (stem of the Y) is in order to provide a foundation for his further theories of evolution, &c. (arms of the Y). It is indeed necessary for these further theories that body-changes should never exercise any hereditary influence on the hereditary endowments of germ-plasm, and therefore it is that he posits the substance of heredity as, not only continuous, but uninterruptably so "since the first origin of life."

Now, this may be made more clear by briefly comparing Weismann's theory with those of Darwin and of Galton. Weismann's theory of heredity, then, agrees with its predecessors which we are considering in all the following respects. The substance of heredity is particulate; is mainly lodged in highly specialized cells; is nevertheless also distributed throughout the general cellular tissues, where it is concerned in all processes of regeneration, repair, and a-sexual reproduction; presents an enormously complex structure, in that every constituent part of a potentially future organism is represented in a fertilized ovum by corresponding particles; is everywhere capable of virtually unlimited multiplication, without ever losing its hereditary endowments; is often capable of carrying these endowments in a dormant state through a long series of generations until at last they reappear in what we recognize as recursions. Thus far all three theories are in agreement. In fact, the only matter of any great importance wherein they disagree has reference to the doctrine of Continuity[25]. For while Darwin's theory supposes the substance of heredity to be mainly formed anew in each ontogeny, and therefore that the continuity of this substance is for the most part interrupted in every generation[26], Weismann's theory supposes this substance to be formed only during the phylogeny of each species, and therefore to have been absolutely uninterrupted since the first origin of life.

[25] Originally, Weismann's further assumption as to the perpetual
stability of germ-plasm, "since the first origin of sexual
reproduction," was another very important point of difference,
but this has now been withdrawn.

[26] I say "_mainly_ formed anew," and "_for the most part_
interrupted," because even Darwin's theory does not, as is
generally supposed, exclude the doctrine of Continuity _in
toto_.

But now, Galton's theory of heredity stands much nearer to Weismann's in this matter of Continuity; for it is, as he says, a theory of "modified pangenesis," and the modification consists in allowing very much more for the principle of Continuity than is allowed by Darwin's theory; in fact he expresses himself as quite willing to adopt (on adequate grounds being shown) the doctrine of Continuity as absolute, and therefore propounded, as logically possible, the identical theory which was afterwards and independently announced by Weismann. Or, to quote his own words--

"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 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[27]."

[27] _Theory of Heredity_ (Journ. Anthrop. Inst. 1875, p. 346).

So far Mr. Galton; but for Weismann's further theory of evolution, &c., it is necessary to postulate the additional doctrine in question; and it makes a literally immeasurable difference to any theory of evolution whether or not we entertain this additional postulate. For no matter how faintly or how fitfully the substance of heredity may be modified by somatic tissues, the Lamarckian principles are hypothetically 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 any degree must always have been _directive_ of variation on the one hand, and _cumulative_ on the other.

Thus, by merely laying this theory side by side with Weismann's we can perceive at a glance how a _pure_ theory of _heredity_ admits of being based on the postulate of Continuity alone, without cumbering itself by any further postulate as to this Continuity being _absolute_. And this, in my opinion is the truly scientific attitude of mind for us to adopt as preliminary to the following investigation. For the whole investigation will be concerned--and concerned only--with this question of Continuity as absolute, or as admitting of degrees. There is, without any question, abundant evidence to prove that the substance of heredity is at least partly continuous (Gemmules). It may be that there is also abundant evidence to prove this substance much more _largely_ continuous than Darwin supposed (Stirp); but be this as it may, it is certain that any such question as to the _degree_ of continuity differs, _toto caelo_, from that as to whether there can ever be any continuity at all.

How, then, we may well ask, is it that so able a naturalist and so clear a thinker as Weismann can have so far departed from the inductive methods as to have not merely propounded the question touching Continuity and its degrees, or even of Continuity as absolute; but to have straightway assumed the latter possibility as a basis on which to run a system of branching and ever-changing speculations concerning evolution, variation, the ultimate structure of living material, the intimate mechanism of heredity, or, in short, such a system of deductive conjectures as has never been approached in the history of science? The answer to this question is surely not far to seek. Must it not be the answer already given? Must it not have been for the sake of rearing this enormous structure of speculation that Weismann has adopted the assumption of Continuity as absolute? As we have just seen, Galton had well shown how a theory of heredity could be founded on the general doctrine of Continuity, without anywhere departing from the inductive methods--even while fully recognizing the possibility of such continuity as absolute. But Galton's theory was a "_Theory of Heredity_," and nothing more. Therefore, while clearly perceiving that the Continuity in question _may_ be absolute, he saw no reason, either in fact or in theory, for concluding that it _must_ be. On the contrary, he saw that this question is, for the present, necessarily unripe for profitable discussion--and, _a fortiori_, for the shedding of clouds of seed in all the directions of "Weismannism."

Hence, what I desire to be borne in mind throughout the following discussion is, that it will have exclusive reference to the question of fact already stated, without regard to any superjacent theories; and, still more, that there is a vast distinction between any question touching the degrees in which acquired characters are transmitted to progeny, and the question as to whether they are ever transmitted in any degree at all. Now, the latter question, being of much greater importance than the former, is the one which will mainly occupy our attention throughout the rest of this Section.

We have already seen that before the subject was taken up by Weismann the difference between acquired and congenital characters in respect to transmissibility was generally taken to be one of degree; not one of kind. It was usually supposed that acquired characters, although not so fully and not so certainly inherited as congenital characters, nevertheless were inherited in some lesser degree; so that if the same acquired character continued to be successively acquired in a number of sequent generations, what was at first only a slight tendency to be inherited would become by summation a more and more pronounced tendency, till eventually the acquired character might become as strongly inherited as a congenital one. Or, more precisely, it was supposed that an acquired character, in virtue of such a summation of hereditary influence, would in time become congenital. Now, if this supposition be true, it is evident that more or less assistance must be lent to natural selection in its work of evolving adaptive modifications[28]. And inasmuch as we know to what a wonderful extent adaptive modifications are secured during individual life-times--by the direct action of the environment on the one hand, and by increased or diminished use of special organs and mental faculties on the other--it becomes obvious of what importance even a small measure of transmissibility on their part would be in furnishing to natural selection ready-made variations in required directions, as distinguished from promiscuous variations in all directions. Contrariwise, if functionally-produced adaptations and adaptations produced by the direct action of the environment are never transmitted in any degree, not only would there be an incalculable waste, so to speak, of adaptive modifications--these being all laboriously and often most delicately built up during life-times of individuals only to be thrown down again as regards the interest of species--but so large an additional burden would be thrown upon the shoulders of natural selection that it becomes difficult to conceive how even this gigantic principle could sustain it, as I shall endeavour to show more fully in future chapters. On the other hand, however, Weismann and his followers not only feel no difficulty in throwing overboard all this ready-made machinery for turning out adaptive modifications when and as required; but they even represent that by so doing they are following the logical maxim, _Entia non sunt multiplicanda praeter necessitatem_--which means, in its relation to causality, that we must not needlessly multiply hypothetical principles to explain given results. But when appeal is here made to this logical principle--the so-called Law of Parsimony--two things are forgotten.

[28] Mr. Platt Ball has, indeed, argued that "use-inheritance would
often be an evil," since, for example, "the condyle of the
human jaw would become larger than the body of the jaw, because
as the fulcrum of the lever it receives more pressure"; and
similarly as regards many other hypothetical cases which he
mentions. (_The Effects of Use and Disuse_, pp. 128-9 _et
seq._) But it is evident that this argument proves too much.
For if the effects of use and disuse as transmitted to progeny
would be an evil, it could only be because these effects as
they occur in the parents are an evil--and this they most
certainly are not, being, on the contrary and as a general
rule, of a high order of adaptive value. Moreover, in the race,
there is a superadded agency always at work, which must
effectually prevent any undue accumulation of these
effects--namely, natural selection, which every Darwinist
accepts as a controlling principle of all or any other
principles of change. Therefore, if, as first produced in the
life-time of individuals, the effects of use and disuse are not
injurious, much less can they become so if transmitted through
the life-time of species. Again, Mr. Wallace argues that, even
supposing use-inheritance to occur, its adapting work in the
individual can never extend to the race, seeing that the
natural selection of fortuitous variations in the directions
required must always produce the adaptations _more quickly_
than would be possible by use-inheritance. This argument, being
one of more weight, will be dealt with in a future chapter.

In the first place, it is forgotten that the very question in debate is whether causes of the Lamarckian order _are_ unnecessary to explain all the phenomena of organic nature. Of course if it could be proved that the theory of natural selection alone is competent to explain all these phenomena, appeal to the logical principle in question would be justifiable. But this is precisely the point which the followers of Darwin refuse to accept; and so long as it remains the very point at issue, it is a mere begging the question to represent that a class of causes which have hitherto been regarded as necessary are, in fact, unnecessary. Or, in other words, when Darwin himself so decidedly held that these causes are necessary as supplements to natural selection, the burden of proof is quite as much on the side of Weismann and his followers to show that Darwin's opinion was wrong, as it is on the side of Darwin's followers to show that it was right. Yet, notwithstanding the elaborate structure of theory which Weismann has raised, there is nowhere one single fact or one single consideration of much importance to the question in debate which was not perfectly well known to Darwin. Therefore I say that all this challenging of Darwinists to justify their "Lamarckian assumptions" really amounts to nothing more than a pitting of opinion against opinion, where there is at least as much call for justification on the one side as on the other.

Again, when these challenges are thrown down by Weismann and his followers, it appears to be forgotten that the conditions of their own theory are such as to render acceptance of the gauge a matter of great difficulty. The case is very much like that of a doughty knight pitching his glove into the sea, and then defying any antagonist to take it up. That this is the case a very little explanation will suffice to show.

The question to be settled is whether acquired characters are ever transmitted by heredity. Now suppose, for the sake of argument, that acquired characters are transmitted by heredity--though not so fully and not so certainly as congenital characters--how is this fact to be proved to the satisfaction of Weismann and his followers? First of all they answer,--Assuredly by adducing experimental proof of the inheritance of injuries, or mutilations. But in making this answer they appear to forget that Darwin has already shown its inefficiency. That the self-styled Neo-Lamarckians have been much more unguarded in this respect, I fully admit; but it is obviously unfair to identify Darwin's views with those of a small section of evolutionists, who are really as much opposed to Darwin's teaching on one side as is the school of Weismann on the other. Yet, on reading the essays of Weismann himself--and still more those of his followers--one would almost be led to gather that it is claimed by him to have enunciated the distinction between congenital and acquired characters in respect of transmissibility; and therefore also to have first raised the objection which lies against the theory of Pangenesis in respect of the non-transmissibility of mutilations. In point of fact, however, Darwin is as clear and decided on these points as Weismann. And his answer to the obvious difficulty touching the non-transmissibility of mutilations is, to quote his own words, "the long-continued inheritance of a part which has been removed during many generations is no real anomaly, for gemmules formerly derived from the part are multiplied and transmitted from generation to generation[29]." Therefore, so far as Darwin's theory is concerned, the challenge to produce evidence of the transmission of injuries is irrelevant: it is no more a part of Darwin's theory than it is of Weismann's to maintain that injuries _are_ transmitted.

[29] _Variation under Domestication_, ii. 392.

There is, however, one point in this connexion to which allusion must here be made. Although Darwin did not believe in the transmissibility of mutilations when these consist merely in the amputation of parts of an organism, he did believe in a probable tendency to transmission when removal of the part is followed by gangrene. For, as he says, in that case, all the gemmules of the mutilated or amputated part, as they are gradually attracted to that part (in accordance with the law of affinity which the theory assumes), will be successively destroyed by the morbid process. Now it is of importance to note that Darwin made this exception to the general rule of the non-transmissibility of mutilations, not because his theory of pangenesis required it, but because there appeared to be certain very definite observations and experiments--which will be mentioned later on--proving that when mutilations are followed by gangrene they are apt to be inherited: his object, therefore, was to reconcile these alleged facts with his theory, quite as much as to sustain his theory by such facts.

So much, then, for the challenge to produce direct evidence of the transmissibility of acquired characters, so far as mutilations are concerned: believers in Darwin's theory, as distinguished from Weismann's, are under no obligation to take up such a challenge. But the challenge does not end here. Show us, say the school of Weismann, a single instance where an acquired character _of any kind_ (be it a mutilation or otherwise) has been inherited: this is all that we require: this is all that we wait for: and surely, unless it be acknowledged that the Lamarckian doctrine reposes on mere assumption, at least one such case ought to be forthcoming. Well, nothing can sound more reasonable than this in the first instance; but as soon as we begin to cast about for cases which will satisfy the Neo-Darwinians, we find that the structure of their theory is such as to preclude, in almost every conceivable instance, the possibility of meeting their demand. For their theory begins by assuming that natural selection is the one and only cause of organic evolution. Consequently, what their demand amounts to is throwing upon the other side the burden of disproving this assumption--or, in other words, of proving the negative that in any given case of transmitted adaptation natural selection has _not_ been the sole agent at work. Now, it must obviously be in almost all cases impossible to prove this negative among species in a state of nature. For, even supposing that among such species Lamarckian principles have had a large share in the formation of hereditary and adaptive characters, how would Weismann himself propose that we should set about the proof of such a fact, where the proof demanded by his assumption is, that the _abstract possibility_ of natural selection having had anything to do with the matter must be excluded? Obviously this is impossible in the case of inherited characters which are also _adaptive_ characters. How then does it fare with the case of inherited characters which are not also adaptive? Merely that this case is met by another and sequent assumption, which constitutes an integral part of the Neo-Darwinian creed--namely, that in nature there _can be no such characters_. Seeing that natural selection is taken to be the only possible cause of change in species, it follows that all changes occurring in species must necessarily be adaptive, whether or not we are able to perceive the adaptations. In this way apparently useless characters, as well as obviously useful ones, are ruled out of the question: that is to say, _all_ hereditary characters of species in a state of nature are _assumed_ to be due to natural selection, and then it is demanded that the validity of this assumption should be disproved by anybody who doubts it. Yet Weismann himself would be unable to suggest any conceivable method by which it can be disproved among species in a state of nature--and this even supposing that the assumption is entirely false[30].

[30] In subsequent chapters, especially devoted to the question
(i.e. Section II), the validity of this assumption will be
considered on its own merits.

Consequently, the only way in which these speciously-sounding challenges can be adequately met is by removing some individuals of a species from a state of nature, and so from all known influences of natural selection; then, while carefully avoiding artificial selection, causing these individuals and their progeny through many generations unduly to exercise some parts of their bodies, or unduly to fail in the exercise of others. But, clearly, such an experiment is one that must take years to perform, and therefore it is now too early in the day to reproach the followers of Darwin with not having met the challenges which are thrown down by the followers of Weismann[31].

[31] I say "the followers of Weismann," because Weismann himself,
with his clear perception of the requirements of experimental
research, expressly states the above considerations, with the
conclusions to which they lead. Nevertheless, he is not
consistent in his utterances upon this matter; for he
frequently expresses himself to the effect, "that the _onus
probandi_ rests with my opponents, and therefore they ought to
bring forward actual proofs" (_Essays_, i. p. 390). But, as
above shown, the _onus_ rests as much with him as with his
opponents; while, even if his opponents are right, he elsewhere
recognizes that they can bring "actual proofs" of the fact only
as a result of experiments which must take many years to
perform.

Probably enough has now been said to show that the Neo-Darwinian assumption precludes the possibility of its own disproof from any of the facts of nature (as distinguished from domestication)--and this even supposing that the assumption be false. On the other hand, of course, it equally precludes the possibility of its own proof; and therefore it is as idle in Darwinists to challenge Weismann for proof of his negative (i. e. that acquired characters are not transmitted), as it is in Weismann to challenge Darwinists for proof of the opposite negative (i. e. that all seeming cases of such transmission are not due to natural selection). This dead-lock arises from the fact that in nature it is beyond the power of the followers of Darwin to exclude the abstract possibility of natural selection in any given case, while it is equally beyond the power of the followers of Weismann to exclude the abstract possibility of Lamarckian principles. Therefore at present the question must remain for the most part a matter of opinion, based upon general reasoning as distinguished from special facts or crucial experiments. The evidence available on either side is presumptive, not demonstrative[32]. But it is to be hoped that in the future, when time shall have been allowed for the performance of definite experiments on a number of generations of domesticated plants or animals, intentionally shielded from the influences of natural selection while exposed to those of the Lamarckian principles, results will be gained which will finally settle the question one way or the other.

[32] Note A.

Meanwhile, however, we must be content with the evidence as it stands; and this will lead us to the second division of our subject. That is to say, having now dealt with the antecedent, or merely logical, state of the question, we have next to consider what actual, or biological, evidence there is at present available on either side of it. Thus far, neither side in the debate has any advantage over the other. On grounds of general reasoning alone they both have to rely on more or less dogmatic assumptions. For it is equally an unreasoned statement of opinion whether we allege that all the phenomena of organic evolution can be, or can not be, explained by the theory of natural selection alone. We are at present much too ignorant touching the causes of organic evolution to indulge in dogmatism of this kind; and if the question is to be referred for its answer to authority, it would appear that, both in respect of number and weight, opinions on the side of having provisionally to retain the Lamarckian factors are more authoritative than those _per contra_[33].

[33] For a fair and careful statement of the present balance of
authoritative opinion upon the question, see H. F. Osborn,
_American Naturalist_, 1892, pp. 537-67.

* * * * *

Turning then to the question of fact, with which the following chapters are concerned, I will conclude this preliminary one with a few words on the method of discussion to be adopted.

First I will give the evidence in favour of Lamarckianism; this will occupy the next two chapters. Then, in Chapter V, I will similarly give the evidence _per contra_, or in favour of Continuity as absolute. Lastly, I will sum up the evidence on both sides, and give my own judgement on the whole case. But on whichever side I am thus acting as special pleader for the time being, I will adduce only such arguments as seem to me valid--excluding alike from both the many irrelevant or otherwise invalid reasonings which have been but too abundantly published. Moreover, I think it will be convenient to consider all that has been said--or may be said--in the way of criticism to each argument by the opposite side while such argument is under discussion--i. e. not to wait till all the special pleading on one side shall have been exhausted before considering the exceptions which have been (or admit of being) taken to the arguments adduced, but to deal with such exceptions at the time when each of these arguments shall have been severally stated. Again, and lastly, I will arrange the evidence in each case--i. e. on both sides--under three headings, viz. (A) Indirect, (B) Direct, and (C) Experimental[34].

[34] [The above paragraph is allowed to remain exactly as Mr.
Romanes left it. Chapters V and VI were however not completed.
_See_ note appended to Preface. C. Ll. M.]