达尔文1868年在《家养动植物变异》一书讨论“泛生假说”（Hypothesis of Pangenesis）的一章。起因是我近年上《生物学思想和概念》课程时，和本科生一道读孟德尔1866年的遗传学论文。有趣的是，达尔文不仅做过类似实验，而且还脱离实验，提出过遗传理论，就是“泛生假设”。比较达尔文和孟德尔的文章，可以有多种讨论，特别是不同的科学思维方法。其中，有些问题并未过时，而很有现实意义。
1868年，达尔文把五类现象，代间遗传、发育、再生、植物嫁接、用进废退，全部联系在一起讨论，他提出的理论，把我们现在认识到的“细胞全能性”（全身很多细胞一直保持整套基因组）、和遗传规律混在一起。他的泛生论确实好像可以解释多个现象，但事后我们可以看到是不同性质和不同层次的现象，因为他求全，所以得出的反而是错误的理论。他提出生物体全身体细胞都产生泛子gemmules（后人亦称pangenes），进入性细胞中，这些gemmules的组合决定了性细胞内含，形成不同的性细胞，再产生不同的后代。在强调体细胞产生泛子的重要性时，达尔文说再生原生质要么不全在于生殖细胞，要么生殖细胞没有再生原生质，而是收集和选择泛子。他在讨论中甚至接受了拉马克（1744-1829）的“用进废退”，而认为泛生假说能解释用进废退（比如连续多代人工地切掉牛角），这是他的理论最可笑的部分，虽然他说很难相信，但自己也看过一个例子，当然还据其他人说。他说受外界影响的体细胞性状可以获得并通过gemmules进入性细胞而传代。达尔文在获得F2代重新出现F1代不见了的隐性性状时，仅看到现象，提出所谓“回复原理”（Principle of Reversion），这并非原理，而是以新名词复述现象。
达尔文当时认为体细胞的性状可以影响生殖细胞的遗传组成。后来德国生物学家魏思曼（August F. L. Weismann，1834-1914）提出germ plasma（种质）学说，种质只存在于生殖细胞中由亲代传给后代，生殖细胞可以产生体细胞，而体细胞不能产生生殖细胞，种质不受体细胞和环境影响而改变。完全摒弃了拉马克主义的基础。哈佛大学的Ernst Mayr将魏斯曼称为19世纪仅次于达尔文的进化论学者。
在有关孟德尔的文章《孤独的天才》一文 中， 更多讨论达尔文错失正确理解遗传学的机会。
Remy J-J (2010) Stable inheritance of an acquired behavior in Caenorhabditis elegans. Current Biology 20:R877-R878.
表观遗传学可参见：Richards EJ (2006) Inherited epigenetic variation — revisiting soft inheritance. Nature Reviews Genetics 7:395-401.
Weismann A (1893). The Germ-Plasm. A Theory of Heredity
Darwin C (1868). The variation of animals and plants under domestication. John Murray, London.
Hypothesis of Pangenesis
Having reflected much on some of the subjects described in the several previous chapters, I have been driven to some hypothetical conclusions, which may perhaps be worth giving. I will in the first place enumerate the leading points;
True seminal generation passes by a not much broken series, through gemmation or multiplication by buds, through fissiparous generation and the renewal by growth of large portions of mutilated individuals, into simple continual growth. The concurrence of the two sexual organs and often of two individuals is necessary in all ordinary cases of seminal generation; but the now well-known cases of 'true parthenogenesis' show that the unimpregnated ovum, passing through the sexual embryonic stages, can be developed into a perfect individual. The unimpregnated ovum also has the power of imbuing every part of the being into which it is developed with its own characters, independently of those of the male, as we see in hybrids; especially in those hybrids sprung from two species of which the one is strongly prepotent in the transmission of character over the other. So conversely it is with the male element. It is admitted by Müller and other physiologists that there is no essential difference in nature between the germs of an ovum and a bud; and Huxley (Transact. Linn. Soc. xxii, pp. 199, 210) has submitted to a rigid comparison the pseudovum (which is of the nature of a bud) of the viviparous aphis, with the true ovum of the oviparous aphis, and can discover no difference in their minute structure. It has, however, sometimes been asserted that only individuals produced from impregnated germs acquire new characters, and that these can only be perfectly transmitted by bud-propagation; but there is no such constant difference, only one in degree and frequency. In the chapter on bud-variation we have seen that not rarely individuals (phytons) produced from buds do display quite new and strongly marked characters, which can sometimes be subsequently propagated by seed; the new varieties arising from buds cannot be distinguished by any characters from seminal varieties; that such new bud varieties, though generally capable of more faithful transmission by bud-propagation than seminal varieties are by seminal propagation, yet occasionally revert, even after a long series of bud generations, to their priestine or parental character. This tendency to reversion in buds is one of the most remarkable points of agreement between bud and seminal multiplication. There is another and still more remarkable point of agreement: we have seen that a bud of one variety inserted into the stock of another variety, in some rare cases has certainly affected, as if by a kind of hybridization, adjoining buds subsequently produced from the stock. If the evidence be thought sufficient, then the dingy purple Laburnam (C. adami), and those marvellous orange trees which produce pure oranges, lemons and citrons as well as fruit of a mixed nature, have thus arisen, then undoubtedly the buds of distinct species can blend together and subsequently produce hybrids by budding, like those produced from the union of the true male and female sexual elements of distinct species.
There is, I believe, one general but hardly invariable difference between seminal generation and gemmation. Beings propagated by the former method, usually pass in the course of their development from a lower to a higher grade, as we see in the metamorphosis of insects & in the concealed metamorphosis of the higher vertabrata; but this passage from a lower to a higher stage cannot be considered as a necessary accompaniment of seminal reproduction, where we look to the kind of development of aphis amongst insects, or to that of all the higher vascular plants. In beings propagated by buds there is, as far as I know, no metamorphosis of this kind; that is, they do not pass first to a lower and then to a higher stage of development; unless indeed the scales surrounding the buds and bulbs of plants may be looked at as indicating such a passage. But beings produced from buds often advance in organization either during the act of their production or subsequently to it, as we see in the many cases of alternate generation. For I follow those naturalists who look at alternate generation as a process of budding, or, as in the case of the strobila of the medusa, as one of fissiparous multiplication. In beings produced by true generation, their metamorphoses, starting from a low grade and advancing to a higher one, no doubt stand in the closest relation, either at present or in past times, to peculiar lines of life and lead to places unoccupied in the economy of nature. Now from several previous considerations we may conclude that there is the closest agreement in nature between a germ and a bud. The concurrence of the two sexes, in the case of the germ being only an accessory, though very general, contingent. We are naturally why the germ, which before impregnation undergoes a certain amount of development ceases to progress and perishes, unless it is acted on by the male element; and why conversely the male element, which can keep alive for even four or five years in the spermatheca of female insects, likewise perishes, unless it acts on or unites with the germ, are questions which notoriously cannot be answered.
But as in the ordinary cases, the male and female elements concur in so similar a manner and degree in giving characters to the embryo, produced from their union, it is just possible that both perish unless they unite, simply from including too little protoplasm or formative (matter) for separate existence and development.
Budding is said by M?ller to differ from fissiparous generation, whether spontaneous or artificial, in the new individual represented by the bud, not being at first perfectly organized, whilst the two or can hardly be said to be perfectly organized. A single cell of some of the lowest plants is capable of reproducing its kind. In the higher plants a bud or bulb with its scales has undergone development. When an animal capable of this kind of reproduction is divided into halves, the extremities are sometimes said at first to bud out; this apparently is a correct expression, for the papillae or projections at the cut extremities consist at first of undeveloped cellular matter. Hence there seems to be no fundamental distinction between gemmation and fissiparous generation; and the latter graduates from the conversion of the two halves of a being into two perfect individuals, through the renewal of a whole limb, into a mere cicatrice by which a wound is healed. We can hardly doubt that it is the same power which presides over the growth and increase of size of each part of the body during youth, and during the whole life of the animals which never cease growing: and we may believe it is the same power which presides over the continual repair of each part and tissue, as it undergoes incessant waste throughout life.
We have seen in a former chapter that some organic effects, which have been included by the older physiologists under the term of nisus formativus together with the renewal of amputated limbs and the healing of wounds, differ in so far that new structures are thus produced, such as false membranes after inflammation or new bones in hydrocephalic skulls; but these structures though in one sense new, are formed of membrane, blood vessels, nerves and bones all resembling those proper to the species.
Turning now to Inheritance. According to the view just given, namely that even seminal generation does not differ fundamentally from the other methods of reproduction or from continual growth, we might have expected that every character belonging to every form would have been as regularly transmitted by all the methods of reproduction as by continual growth. On this view that some organic effects, which have been included by the older physiologists under the term of nisus formativus together with the renewal of amputated limbs and the healing of wounds, differ in so far that new structures are thus produced it is not inheritance, but non-inheritance, which is the anomaly. We have seen in the chapter devoted to the subject, that a multitude of newly acquired characters, whether injurious or beneficial, whether of the lowest or highest vital importance, are often faithfully transmitted, frequently even when one parent alone is characterized.
We have seen that the very same characters will in one case be transmitted by inheritance and not in another case. Reason has been assigned for the belief that some characters are not transmitted, not from a failure of the power of inheritance, but from the conditions of life incessantly inducing fresh variability, as with grafted fruit trees and highly cultivated flowers. Other, perhaps all other, cases of non-inheritance, may be included under the principle of Reversion, by which the child tends to resemble its grandfather or more remote progenitor rather than its parents.
This principle of Reversion is the most wonderful of all the attributes of Inheritance. It frequently comes into action. What can be more wonderful than that characters, completely lost during scores or hundreds or even thousands of generations, should suddenly reappear perfectly developed, as we have seen with differently coloured pigeons and fowls when purely bred, but more especially when crossed, and as with the zebrine stripes on dunn coloured horses and in several such cases. Many monstrosities come under this same head, as when rudimentary organs are redeveloped, as in three toed horses, or when organs, which we may believe were possessed by an early progenitor, but of which no rudiment is visible in its descendants suddenly reappears as with additional mammae in a woman and a fifth stamen in some Scrophulariaceae. We have seen that reversion acts in bud-reproduction; and we occasionally see it during the growth even of individual animals, especially, but not exclusively when of crossed parentage, as in the rare cases described of fowls, pigeons, cattle and rabbits, which revert as they grow older to the colours of either parent or of some ancestor. We are led to believe, as explained in the chapter on inheritance, that every character which occasionally reappears through reversion, is present, though latent in each generation, in nearly the same way as in most or all females the secondary male characters lie latent and ready to be evolved, when the female reproductive organs are affected.
In every living creature we may feel assured that a host of lost characters lie latent and ready to be evolved under the proper conditions. How can we make intelligible and connect with other facts this wonderful and common capacity of reversion—this power of apparently calling back to life long lost characters?
We have seen in the latter chapters of this volume, that changed conditions of life, at first often causes sterility and when prolonged during several generations, induce much fluctuating variability throughout the whole organization.
Sometimes changed conditions directly cause modifications in the beings propagated under them; by which I mean a considerable number of individuals are similarly modified. Again certain conditions, as intemperate living, causes an inheritable diseased condition, as with gout. In all such cases, as the germs and male elements are parts of the individuals which have been exposed to the new or injurious conditions, one may suppose that they have been in some way directly, though slightly, affected. But what shall we say about the inherited results of the use and disuse of particular organs? The domesticated duck flies less and walks more than the wild duck, and the limb bones of the tame duck have become in a corresponding manner diminished and increased in comparison with those of the wild duck: a horse is trained to certain paces, and the colt inherits similar consensual movements: the domesticated rabbit becomes tame from close confinement; the dog intelligent from associating with man; the retriever is taught to fetch and carry, and these mental endowments are inherited. Nothing in the whole circuit of natural history is more wonderful. How can the use or disuse of a particular limb or of the brain affect a small aggregate of cells in the reproductive organs, in such a manner that the being developed from these organs inherits these newly acquired characters of either one or both parents?
I have now enumerated the chief leading points which we naturally wish to connect together by some intelligible bond. It will, I presume, be admitted that the protoplasm or formative matter, included within the germ and male element, and endowed with vital force, cause in seminal generation the development of each new being whose germs and buds agree, as we have seen, in structure as far as this is visible, in many remarkable attributes, as in varying inheritance, reversion, and hybridization, and lastly in their fully developed product. Hence it seems by far the simplest belief that protoplasm, identical in nature with that within the germ, collects at certain points to form buds. If this view be admitted it must certainly be extended to fissiparous generation, to the renewal of an amputated limb, to the healing of a wound and probably to continuous growth. We are thus led to believe that protoplasm of the same nature, must be diffused throughout the whole of each organic being, ready when super-abundant to form by budding new beings, both at the period of maturity and in the cases of alternate generation during youth; and ready to form new structures as after inflammation, and ready to repair lost or wasted structures. On this view we must believe that the reproductive organs do not by any means exclusively form the generative protoplasm, if indeed they form any of it, but only select and accumulate it in the proper quantity, and make it ready for separate existence.
We can thus understand the antagonism that has long been observed in plants between increase by buds, rhizomes, suckers and seminal generation (and indeed between the latter and active growth during youth); for in both cases the same protoplasmic matter is consumed; and there is not enough for both methods of propagation.
It is surprising that this antagonism should be as general as it is; but it does not invariably hold good; for the young males of the salmon, whilst very small, have their reproductive organs active; and Ernst Haeckel has recently (Monatsbericht Akad. Wiss. Berlin. Feb. 2, 1865) described the wonderful case of a medusa, with its reproductive organs active, which at the same time produces by budding a widely different form of medusa, which likewise has the power of seminal reproduction.
Furthermore, I am led to believe from analogies immediately to be given that the protoplasm or formative matter which is diffused throughout the whole organization, is generated by each different tissue and cell or aggregate of similar cells;—that as each tissue or cell becomes developed, a superabundant atom or gemmule as it may be called of the formative matter is thrown off;—that these almost infinitely numerous and infinitely minute gemmules unite together in due proportion to form the true germ;—that they have the power of self-increase or propagation; and that they here run through the same course of development, as that which the true germ, of which they are to constitute elements, has to run through, before they can be developed into their parent tissue or cells. This may be called the hypothesis of Pangenesis.
On this hypothesis the many different parts of the structures of each individual may be compared to so many distinct organic beings, united together, but each of which propagates its own proper form. The union is far more intimate than that of flower buds or leaf-buds on the same tree, or of the polypi on the same coral; but even in these cases we have some differentiation in the so-called individuals, and some parts in common; for plants have trunks and roots in common, and some kinds habitually produce two kinds of flowers; and the polypi of some corals have certain parts and the power of movement in common.
It is known that all the species in the same great class start in their course of embryonic development from the same point; and running for a time along the same path and therefore resembling each other in their earlier embryonic stages diverge in structure more and more, according as they are more and more different when grown to maturity. So we may believe it to be with the development of the gemmules thrown off from each different tissue and cell. There is hardly any greater difficulty in believing that these many gemmules may unite or cohere, each retaining its own power and qualities, into a single true germ, than in the well-known union of two species into a hybrid, and of the hybrid with another hybrid until several species are commingled in a single individual. Most of those who have closely studied hybrids and mongrels, especially M. Naudin, believe that all the characters of both parent-species are commingled, often in very unequal degrees in the unified product, but are not fused together or changed in nature like two elements in a chemical union.
We sometimes see evidence of this in the manner in which the petals of mongrel hybrid plants are finely streaked or blotched with the pure colours of the two parent-forms; and still more plainly in their reversion during successive generations, when not crossed by either parent species to the perfect character in every respect of such parents.
We must further believe that the many gemmules which together form a germ or a bud possess a marvellous mutually elective power by which they are all brought together in proper place and in due proportions. But we know that some such power resides in the male and female elements of every species; for when the germ is placed in contact with the male element of any number of closely allied species and with its own male element, the latter alone in all ordinary cases has any effect. But under certain peculiar conditions as we have seen, the male element of a distinct species is in the most marked manner elected by the germ.
We apparently see the same elective power in the law so strongly insisted on by Geoffroy St. Hilaire, de l'affinité de soi pour soi, which is displayed in so marvellous a manner in double monsters, with their corresponding and adjoining parts perfectly fused together; for if the modern view (see Carpenter and others) be correct, these double monsters originate from the early division of a single embryo, yet the complete fusion for instance of the adjoining arms of the two monsters, into one, but not of the two other arms, seems to require Geoffrey's law. But there are many other cases in which we see the same mutual attraction of homologous parts in monsters which are not double, as when the two eyes or the two legs, are united into one, with the almost perfect fusion of the corresponding bones, muscles and nerves. So again, some normal structures, which at an early embryonic age consist of two distinct organs are perfectly fused during the subsequent development into a single organ.
According to our hypothesis, this same mutual elective power probably comes into play in all such cases as that described by Spallanzani, who many times cut off the leg of a newt, sometimes at one joint or place and sometimes at another, and the new limb was always perfectly formed, neither too much nor too little being added. Hence we must suppose that each part or cell of the severed surface of bone, muscle and nerve, elected or attracted the different gemmules of those cells which ought to come next in due order, and these elected others, till the whole limb was exactly reproduced. To the same elective power acting under abnormal circumstances in our
During the indefinite multiplication of each species by seminal generation, no one believes that the protoplasm included within the germ and male elements is handed down or distributed from parents to offspring from one large primordial stock; on the contrary, it must be increased or multiplied during each generation. So it must be with the diffused protoplasm in those organisms which bud into an indefinite number of new individuals, and likewise in all such cases, as that, described by Bonnet, of a fresh water worm which from a mere fragment eight times successively reproduced its whole head and tail. So it must be with the separate constituent elements of the germ, that is the many gemmules or atoms of protoplasm thrown off from each tissue and cell during its development. By the multiplication and preservation in an undeveloped state of these gemmules, I account for all latent characters — those of the male latent in the female, and those which reappear in all the wonderful cases of Reversion. There is hardly any greater a priori improbability in gemmules remaining for a long time undeveloped than in fertilized seeds lying dormant for hundreds or thousands of years; nor is there much more improbability in our gemmules, in an undeveloped condition, being held down from generation to generation, being added to or growing at each generation, than is the transmission and partial development of numerous rudimentary organs; or than in the transmission of only a tendency to the production of such rudimentary organs. When a hornless breed of cattle, for instance, is greatly multiplied, it must be believed according to my hypothesis that gemmules thrown off from the cells which formed the horns in the parent-race, have been propagated through each succeeding generation, and occasionally that these gemmules from unknown causes run through the proper course of development and form horns and thus a horned beast appears by what is called Reversion. It is hardly more wonderful that these supposed latent gemmules, when undergoing development, should find their proper place and form horns, that thus the horns of a calf from a short-horned cow by a long-horned bull, should have its horns and not its hoofs affected by the male element of the bull; or that in crossing two birds with coloured and uncoloured tails, the tail of the offspring and not all the plumage [should] be affected. We have seen that it is most difficult to determine whether mutilations are ever inherited: I know of one case where an organ has been removed during several generations before its development and consequently before it can have thrown off gemmules afterwards capable of self propagation: but if mutilations are ever inherited, as has so often been stated to be the case, we could on this view in some degree understand the cause. Lastly, we can in a rough manner understand the most perplexing of all the cases of inheritance, namely how the effects of use and disuse of parts can be inherited; for the tissues and cells increased or decreased by use or disuse (but why actually existing parts should be thus affected by use and disuse I believe is not clear to physiologists) are supposed at the period of their formation to throw off gemmules endowed with all the qualities which they have acquired.
I have not as yet alluded to one most remarkable physiological fact, namely the influence of a first impregnation on the subsequent offspring of the female. The case of Lord Morton's mare has often been quoted: after producing a hybrid to a quagga, she bore two foals to a purely-bred horse. Yet these foals in their striped legs and character of hair plainly showed the quagga influence. Several analogous cases have been observed in crossed domestic animals. Now by our hypothesis the male element includes gemmules of every part of the parent-structures; and these gemmules have the power of self-increase and are properly diffused throughout the whole organization. We can thus see how the adult female may occasionally and under unknown favourable circumstances become, as it were, impregnated with characters from the male, and her subsequent offspring be thus affected. If the tissues of the female herself were plastic or undergoing their first development, they might, also, thus become affected. I make this remark because in the vegetable kingdom we meet with this very case: several undoubted instances of the pollen of one species or variety affecting parts of the female flower of another species or variety have been given: thus when Gallesio fertilized an orange flower with pollen from a lemon, the fruit bore stripes of lemon peel: when pale-coloured varieties of the pea and of the stock have been fertilized by pollen from darker varieties, the coats of the seeds have become coloured. Now the peel of the orange and the seed coats of the pea or stock are as much parts of the female, as the skin, hair and womb of a female quadruped. There is this difference in the two cases, that these parts in the plant have undergone a large amount of development since the act of impregnation, and have thus been affected by the differentiation and growth of certain gemmules which were contained within the pollen grains of the male.
The hypothesis, as now given, is in truth extremely complex, but so assuredly are the facts whether or no we invent some hypothesis by which they may be embraced. The gemmules thrown off from each different tissue and cell in one of the higher animals, which together form a true germ or bud, must be inconceivably numerous and minute. But reflect how minute and numerous the organic particles must be, with which the wind is tainted by certain offensive animals over miles of space, yet these particles can affect the olfactory nerves of other animals:
Not only each developed tissue as it is continually renewed is assumed by our hypothesis to throw off gemmules, but in every animal and plant there must be innumerable latent, self-propagating gemmules, ready under fitting circumstances to be developed. More than this we know that a moth during its development passes through several different moults in its caterpillar state, and through the cocoon state, and some animals pass through many more stages; and all the characters of each stage have to be transmitted and consequently at each stage gemmules have to be thrown off, preserved and multiplied, so as to be developed in the next generation into the parent cell or tissue.
In animals which undergo the process of alternate generation, at certain stages of development, all the tissues of the body break up and pass into a higher state of development, like a caterpillar when passing into the cocoon stage; and at this period the gemmules of each organ must multiply greatly so that uniting together into germ-like bodies, several individuals are produced instead of one; but each of these new individuals must include gemmules of all the earlier stages which have been passed through and of all the later stages which have to be passed through. Notwithstanding the astounding complexity of the processes implied by this hypothesis of pangenesis, yet it seems to me to comprehend the several leading facts better than any other view. On this hypothesis we may fancifully look at each animal and plant as being compounded of many beings, in the same manner as a tree or coral is compounded of many similar beings; but in neither case have these so-called beings had a separate existence. Each of these beings, or parts, is supposed to be capable of throwing off gemmules, which whilst within the organism are capable of self-increase, and which can be separately developed at the part or organ whence they were derived, and can be united, as in the case of hybrids, with other gemmules into a single germ or bud, which reproduces the complete parent form. On this view, each organic being may be looked at as a little universe, formed of a host of different self-propagating organisms, almost as numerous as the stars in heaven, and as minute as they are immense.