The Cell in Development and Inheritance

In this cell it breaks up into its constituent bio- phores, and gives the cell its inherited specific character."^ Devel- opment is, therefore, essentially evolutionary and not epigenetic ; ^ its point of departure is a substance in which all of the adult characters are represented by preformed, prearranged germs ; its course is the result of a predetermined har^iony in the succession of the qualitative divisions by which the hereditary substance is progressively disinte- grated. In order to account for heredity through successive genera- tions, Weismann is obliged to assume that, by means of quantitative or integral division, a certain part of the original germ-plasm is car- ried on unchanged, and is finally delivered, with its original architecture unaltered, to the germ-nuclei. The power of regeneration is explained, in like manner, as the result of a transmission of unmodified or sHghtly modified germ-plasm to those parts capable of regeneration.

It is impossible not to admire the thorou<(hness, candour, and lo^^ical skill with which Weismann has developed his theory, or to deny That, in its final form, it does afford up to a certain point a forma/ solution of the problems with which it deals. Its fundamental weakness is its ^//^?j-/-metaphysical character, which, indeed, almost places it outside Fig. 183. — Half and whole cleavage in the eggs of sea-urchins. A. Normal sixteen-cell stage, showing the four micromeres above (from Driesch, after Selenka). B. Half sixteen-cell stage developed from one blastomere of the two-cell stage after killing the other by shaking (Driesch). C. Half blastula resulting, the dead blastomere at the right (Driesch), D. Half-sized sixteen-cell stage of Toxopneustes, viewed from the micromere-pole (the «yght lower not shown). This embryo, developed from an isolated blastomere of the two-cell stage, segmented like an entire normal ovum.

the sphere of legitimate scientific hypothesis. Save in the maturation of the germ-cells (''reducing divisions"), none of the visible phenom- ena of cell-division give even a remote suggestion of qualitative divi- sion. All the facts of ordinary mitosis, on the contrary, indicate that the division of the chromatin is carried out with the most e.xact equality. The hypothesis mainly rests upon a quite different order of phenom- ena, namely, on facts indicating that isolated blastomeres, or other cells, have a certain power of self-determination, or ** self-differentia- tion" (Roux), pecuKar to themselves, and which is assumed to be pri- marily due to the specific quality of the nuclei. This assumption, which may or may not be true,^ is itself based upon the further assump- tion of quahtative nuclear division of which we actually know^ nothing whatever. The fundamental hypothesis is thus of purely a priori character; and every fact opposed to it has been met by subsidi-

A. Normal gastrula. B. Half-sized dwarf, from an isolated blastomere of ihe two-cell stage. C. Quarter-sized dwarf, from an isolated blastomere of the four-cell stage. ary hypotheses, which, like their principal, relate to matters beyond the reach of observation. Such an hypothesis cannot be actually overturned by a direct appeal to fact. We can, however, make an indirect appeal, the results of which show that the hypothesis of qualitative division is not only so improbable as to lose all semblance of reality, but is in fact quite superfluous. It is rather remarkable that Roux himself led the way in this direction. In the course of his observations on the development of a half-embryo from one of the blastomeres of the two-cell stage of the frog's ^gg, he determined the significant fact that the half-embryo in the end restores more or less completely

the missing half \iy a peculiar process, related to regeneration, which Roux designated as post-generation. Later studies showed that an isolated blastomere is able to give rise to a complete embryo in manv other animals, sometimes developing in its earlier stages as though Fig. 185. — Dwarf and double embryos of Amphioxus. A. Isolated blastomere of the two-cell stage segmenting like an entire egg Kef. Fig. 183. />>). B. Twin gastrulas from a" single egg. C. Double cleavage resulting from the partial separation, by shaking, of the blastomeres of the two-cell stage. D.E.F. Double gastrulas arising from such forms as the last.

still forming part of a complete embryo (** partial development"), but in other cases developing directly into a complete dwarf embryo, as if it were an ^g% of diminished size. In 1891 Driesch w^is ab^e to follow out the development of isolated blastomeres of sea-nrchin eggs separated by shaking to pieces the two-cell and four-cell stages. Blastomeres thus isolated segment as if still forming part of an entire larva, and give rise to a half- (or quarter-) blastula (Fig. 183). The opening soon closes, however, to form a small complete blastula, and the resulting gastrula and Pluteus larva is a perfectly formed dwarf of only half (or quarter) the normal size. Incompletely separated blastomeres give rise to double embryos Uke the Siamese twins. Shortly afterward the writer obtained similar results in the case of AnipJiioxHS^ but here tJie isolated blastoinere behaves from the begin- 7iing like a complete ovum of half the usual sice, and gives rise to a complete blastula, gastrula, and larva. Complete embryos have also been obtained from a single blastomere in the teleost Fuiidulus (Morgan, '95, 2), in Triton (Herlitzka, '95), and in a number of hydromedusae (Zoja, '95, Bunting, '99); and nearly complete em- bryos in the tunicates Ascidiella (Chabry, "^'j\ PJiallusia (Driesch, '94), and Molgula (Crampton, '98).^ Perhaps the most striking of these cases is that of the hydroid Clytia, in which Zoja was able to obtain perfect embryos, not only from the blastomeres of the two- cell and four-cell stages, but from eight-cell and even from sixteen- cell stages, the dwarfs in the last case being but one-sixteenth the normal size.

These experiments render highly improbable the hypothesis of qualitative division in its strict form, for they demonstrate that the earlier cleavages, at least, do not in these cases sunder fundamentally different materials, either nuclear or cytoplasmic, but only split the ^&g up into a number of parts, each of which is capable of producing an entire body of diminished size, and hence m.ust contain all of the material essential to complete development. Both Roux and Weis- mann endeavour to meet this adverse evidence with the assumption of a " reserve idioplasm," containing all of the elements of the germ- plasm which is in these cases distributed equally to all the cells in addition to the specific chromatin conveyed to them by qualitative division. This subsidiary hypothesis renders the principal one {i.e. that of qualitative division) superfluous, and brings us back to the same problems that arise when the assumption of qualitative division is discarded.

The theory of qualitative nuclear division has been practically dis- proved in another way by Driesch, through the pressure-experiments already mentioned at page 375. Following the earlier experiments of Pfluger ('84) and Roux ('85) on the frog's Q,g^, Driesch subjected segmenting eggs of the sea-urchin to pressure, and thus obtained flat plates of cells in which the arrangement of the nuclei differed totally 1 The "partial" development in the earlier stages of some of these forms is considered at page 419.

from the normal (Fig. 186); yet such eggs when released from press- ure continue to segment, zvitJwiit rearrangeme^it of the nuclei, and give rise to perfectly normal larvae. I have repeated these experi- ments not only with sea-urchin eggs, but also with those of an annelid i^Nereis\ which yield a very convincing result, since in this case the histological differentiation of the cells appears very early. In the normal development of this animal the archenteron arises from four large cells or macromeres (entomeres), which remain after the suc- cessive formation of three quartets of micromeres (ectomeres) and the parent-cell of the mesoblast. After the primary differentiation of the germ-layers the four entomeres do not divide again until a very late period (free-swimming trochophore), and their substance always retains a characteristic appearance, differing from that of the other

Fig. 186. — Modification of cleavage in sea-urchin eggs by pressure. A. Normal eight-cell stage of Toxopneustes. B. Eight-cell stage of £'c/^///«j segmentino- under pressure. Both forms produce normal Plutei. blastomeres in its pale non-granular character and in the presence of large oil-drops. If unsegmented eggs be subjected to pressure, as in Driesch's echinoderm experiments, they segment in a flat plate, all of the cleavages being vertical. In this way are formed eight-celled plates in which all of the cells contain oil-drops (Fig. 187, D). If they are now released from the pressure, each of the cells divides in a plane approximately horizontal, a smaller granular micromere being formed above, leaving below a larger clear macromere in which the oil-drops remain. The sixteen-cell stage, therefore, consists of eight deutoplasm-laden macromeres and eight protoplasmic micromeres (instead of four macromeres and twelve micromeres, as in the usual development). These embryos developed into free-swimming trocho- phores containing eight instead of four macromeres, which have the typical clear protoplasm containing oil-drops. In this case there can

be no doubt whatever that four of the entoblastic nuclei were nor- mally destined for the first quartet of micromeres (Fig. 187, B\ from which arise the apical gangha and the prototroch. Under the condi- tions of the experiment, however, they have given rise to the nuclei of cells which differ in no wise from the other entoderm-cells. Even Fig. 187. — Modifications of cleavage by pressure in Nereis. A. B. Normal four- and eight-cell stages. C. Normal trochophore larva resulting, with four entoderm-cells. D. Eight-cell stage arising from an egg flattened by pressure ; such eggs give rise 10 trochophores with eight instead of four entoderm-cells. Numerals designate the successive cleavages.

in a highly differentiated type of cleavage, therefore, the nuclei of the segmenting Qgg are not specifically different, as the Roux-Weismann hypothesis demands, but contain the same materials even in the cells that undergo the most diverse subsequent fate. But there is, further- more, very strong reason for believing that this may be true in later stages as well, as Koiliker insisted in opposition to Weismann as early as 1886, and as has been urged by many subsequent writers. The strongest evidence in this direction is afforded by the facts of regeneration; and many cases are known — for instance, among the hydroids and the plants — in which even a small fragment of the body is able to reproduce the whole. It is true that the power of regeneration is always limited to a greater or less extent according to the species. But there is no evidence whatever that such hmita- tion arises through specification of the nuclei by quaHtative division, and, as will appear beyond, its explanation is probably to be sought in a very different direction.

We have now cleared the ground for a restatement of the problem of development and an examination of the views opposed to the Roux-Weismann theory. After discarding the hypothesis of quaH- tative division the problem confronts us in the following form. If chromatin be the idioplasm in which inheres the sum total of heredi- tary forces, and if it be equally distributed at every cell-division, how can its mode of action so vary in different cells as to cause diversity of structure, i.e. dijferentiatio7i ? It is perfectly certain that differen- tiation is an actual progressive transformation of the egg-substance involving both physical and chemical changes, occurring in a definite order, and showing a definite distribution in the regions of the Qgg. These changes are sooner or later accompanied by the cleavage of the Qgg into cells whose boundaries may sharply mark the areas of differentiation. What gives these cells their specific char- acter.? Why, in the four-cell stage of an annelid Qgg, should the four cells contribute equally to the formation of the ahmentary canal and the cephaUc nervous system, while only one of them (the left- hand posterior) gives rise to' the nervous system of the trunk-region and to the muscles, connective tissues, and the germ-cells.' (Figs. 171, 188, B.) There cannot be a fixed relation between the various regions of the &gg which these blastomeres represent and the adult pa'I-ts arising from them; for in some eggs these relations maybe artificially changed. A portion of the Qgg which under normal con- ditions would give rise to only a fragment of the body will, if split off from the rest, give rise to an entire body of diminished size. W hat then determines the history of such a portion.? What influence moulds it now into an entire body, now into a part of a body .?

De Vries, in his remarkable essay on Intracellular l\iuo;cmsis ('89), endeavoured to cut this Gordian knot by assuming that the character of each cell is determined by pangens that migrate from the nucleus into the cytoplasm, and, there becoming active, set up specific changes and determine the character of the cell, this way or that, according to their nature. But what influence guides the migrations of the pangens, and so correlates the operations of devel- opment? Both Driesch and Oscar Hertwig have attempted to

Fig. i88. — Diagrams illustrating the value of the quartets in a polyclade {Leptoplana), a lamel- libranch {Ufiio), and a gasteropod {Crepiduld). A. Leptoplana, showing mesoblast-formation in the second quartet. B. Crepidula, showing source of ectomesoblast (from a~, b-, c'^) and en- tomesoblast (from quadrant Z)). C. 6'>/2c, 'ectomesoblast formed only from u^. In all the figures the successive quartets are numbered with Arabic figures ; ectoblast unshaded, mesoblest dotted, entoblast vertically lined.

answer this question, though the first-named author does not commit himself to the pangen-hypothesis. These writers have maintained that the particular mode of development in a given region or blasto- mere of the tgg is a result of its relation to the remainder of tJie 7nass^ i.e. a product of what may be called the intra-embryonic environ- ment. Hertwig insisted that the organism develops as a whole as the result of a physiological interaction of equivalent blastomeres the transformation of each being due not to an inherent specific power of self-differentiation, as Roux's mosaic-theory assumed but to the action upon it of the whole system of which it is a part ''According to my conception," said Hertwig, - each of the first two blastomeres contains the formative and differentiating forces not simply for the production of a half-body, but for the entire organism ; the left blastomere develops into the left half of the bodv only because it is placed in relation to a right blastomere." 1 Again, in a later paper : " The ^gg is a specifically organized elementary organism that develops epigenetically by breaking up into cells and their subsequent differentiation. Since every elementary part {i.e. cell) arises through the division of the germ, or fertilized ^gg, it contains also the ger'm of the w^hole, but during the process of development it becomes ever more precisely differentiated and determined by the formation of cytoplasmic products according to its position with reference to the entire organism (blastula, gastrula, etc.)." 2

An essentially similar view was advocated by the writer ('93, '94) nearly at the same time, and the same general conception was ex- pressed with great clearness and precision by Driesch shortlv after Hertwig: "The fragments {i.e. cells) produced by cleavage are com- pletely equivalent or indifferent." "The blastomeres of the sea- urchin are to be regarded as forming a uniform material, and they may be thrown about, like balls in a pile, without in the least degree impairing thereby the normal power of development." '"^ '' The rehi- tive positio7i of a blastomere iji tJie zi'/io/e determines in general what develops from it ; if its position be e hanged, it gives rise to something different; in other words, its prospective valne is a function of its position. "^

Jn/this last aphorism the whole problem of development is brought to a focus. It is clearly not a solution of the j^roblem, but (Mily a highly suggestive restatement of it; for everything turns upon how the relation of the part to the whole is conceived. W^-y little con- sideration is required to show that this relation cannot be a merelv geometrical or rudely mechanical one, for in the eggs of different ^ '93» P- 793- It should be pointed out that Roux himself in several papers expressl\ recognizes the fact that development cannot be regarded as a ]iure mosaic-work, and that besides the power of self-differentiation postulated by his hypothesis we must assume a "correlative differentiation" or differentiating interaction of parts in the embryo. Cf. Koux,

animals blastomeres may almost exactly correspond in origin and relative position, yet differ widely in their relation to the resulting embryo. Thus we find that the cleavage of polyclades, annelids, and gasteropods (Fig. i88) shows a really wonderful agreement in form, yet the individual cells differ markedly in prospective value. In all of these forms three quartets of micromeres are successively formed according to exactly the same remarkable law of the alternation of the spirals ; ^ and, in all, the posterior cell of a fourth quartet lies at the hinder end of the embryo in precisely the same geometrical relation to the remainder of the embryo ; yet in the gasteropods and annelids this cell gives rise to the mesoblast-bands and their products, in the polyclade to a part of the archenteron, while important differences also exist in the value of the other quartets. The relation of the part to the whole is therefore of a highly subtle character, the pro- spective value of a blastomere depending not merely upon its geomet- rical position, but upon its relation to the whole complex inherited organization of which it forms a part. The apparently simple con- clusion stated in Driesch's clever aphorism thus leads to further prob- lems of the highest complexity. It should be here pointed out that Driesch does not accept Hertwig's theory of the interaction of blasto- meres as such, but, hke Whitman, Morgan, and others, has brought forward effective arguments against that too simple and mechanical conception.

That theory is, in fact, merely Schwann's cell-composite theory of the organism applied to the developing embryo, and the general arguments against that theory find some of their strongest \ support in the facts of growth and development.^ Th]s has been ' forcibly urged by Whitman ('93), who almost simultaneously with the statements of Driesch and Hertwig, cited above, expressed the con- viction that the morphogenic process cannot be conceived as merely the sum total or resultant of the individual cell-activities, but operates as a unit without respect to cell-boundaries, precisely as De Bary con- cludes in the case of growing plant-tissues (p. 393), and the nature of that process is due to the organization of the ^gg as a whole. ^

While recognizing fully the great value of the results attained during the past few years in the field of experimental and specula- tive embryology, we are constrained to admit that as far as the essence of the problem is concerned we have not gone very far beyond the conclusions stated above ; for beyond the fact that the inherited organization is involved in that of the germ-cells we remain quite ignorant of its essential nature. This has been recognized by no one more clearly than by Driesch himself, to whose critical researches we owe so much in this field. At the cUmax of a recent elaborate analysis, the high interest of which is somewhat obscured by 1 Cf. p. 368. ^ Cf. pp. 388-394.

its too abstruse form, Driesch can only reiterate his former aphorism,^ finally taking refuge in an avowed theory of vitalism which assumes the localization of morphogenic phenomena to be determined bv "a wholly unknown principle of correlation,"^ and forms a problem sni generis^^ This conclusion recognizes the fact that the fundamental problem of development remains wholly unsolved, thus confirming from a new point of view a conclusion which it is only fair to point out has been reached by many others.

- But while the fundamental nature of the morphogenic process thus remains unknown, we have learned some very interesting facts regard- ing the conditions under which it ta'kes place, and which show that Driesch's aphorism loses its meaning unless carefully qualified. The experiments referred to at pages 353, 410, show that up to a certain stage of development the blastomeres of the early echinoderm, AnipJii- oxiis or medusa-embryo, are "totipotent" (Roux), or "equipotential " (Driesch), i.e. capable of producing any or all parts of the body. Even in these cases, however, we cannot accept the early conclusion of Pfliiger i^'^Z), applied by him to the frog's ^gg, and afterward accepted by Hertwig, that the material of the ^gg,, or of the blasto- meres into which it splits up, is absolutely " isotropic," i.e. consists of quite uniform indifferent material, devoid of preestablished axes. Whitman and Morgan, and Driesch himself, showed that this cannot be the case in the echinoderm Qgg\ for the ovum possesses a polarity predetermined before cleavage begins, as proved by the fact that at the fourth cleavage a group of small cells or micromeres always arises at a certain point, which may be precisely located before cleavage by reference to the eccentricity of the first cleavage-nucleus,'* and which, as Morgan showed,^ is indicated before the third, and sometimes before the second cleavage, by a migration of pigment away from the micromere-pole.

These observers are thus led to the assumption of a primary polarity of the egg-protoplasm, to which Driesch, in the course of further analysis of the phenomena, is compelled to add the assumption of a secondary polarity at right angles to the first.'^ The.se polarities, inherent not only in the entire ^gg, but also in each of the blastomeres into which it divides, form the primary conditions under which the bilaterally symmetrical organism develops by epigenesis. To this extent, therefore, the material of the blastomeres, though ''totipotent," shows a certain predetermination with respect to the adult body.

2 This phrase is cited by Driesch from an earher work (^'92, p. 59O) as ^ivm- a cncct though " unanalytical " statement of his view. It may be questioned whether many readers will regard as an improvement the '' analytical " form it assumes in his last work. ~\Ve now proceed to the consideration of experiments which show- that in some animal egK'^ ^^»ch ]-)redetermination may go much farther, so that the development does, in fact, show many of the features of a mosaic-work, as maintained ]:>y Roux. The best-determined of these cases is that of the ctenoph()re-e«;K. ^^^ shown by the work of Chun,

Fig. 189. — Partial larvcne of the ctenophore Beroc. [Dkiesch and Morgan.] A. Half sixteen-cell stage, from an isolated blastomere. li. Resulting larva, with four rows of swimming-plates and three gastric pouches. C. One-fourth sixtecn-cell stage, from an isolated blastomere. D. Resulting larva, with two rows of plates and two gastric pouches. E. Defective larva, with six rows of plates and three gastric pouches, from a nucleated fragment of an unseg- mented egg. F. Similar larva with five rows of plates, from above.

Driesch, and Morgan ('95), and Fischel ('98). These observers have demonstrated that isolated blastomeres of the two-, four-, or eight-cell stage undergo a cleavage which, through the earliest stages, is exactly like that which it would have undergone if forming part of a com- plete embryo, and gives rise to a defective larva, having only four two, or one row of swimming-plates (Fig. 189); and Fischel's obser' vations give strong reason to believe that each of the eight micromeres of the sixteen-cell stage is definitely specified for the formation of one of the rows of plates. In like manner Crampton (96) found that in case of the marine gasteropod Ilyanassa isolated blastomeres of two- cell or four-cell stages segmented exactly as if forming part of an entire embryo, and gave rise to fmgmaits of a larva, not to complete dwarfs, as in the echinoderm (Fig. 190). Further, in embryos from which the ''yolk-lobe" (a region of that macromere from which the primary mesoblast normally arises) had been removed, no mesoblast- bands were formed. Most interesting of all, Driesch and Morgan discovered that if a part of the cytoplasm of an imscgjnaitcd cte^'no- phore-egg were removed, the remainder gave rise to an incomplete larva, showing definite defects (Fig. 189, E, F).

In none of these cases is the embryo able to complete itself, though it should be remarked that neither in the ctenophore nor in the snail is the partial embryo identical with a fragment of a whole embryo, since the micromeres finally enclose the macromeres, leaving no sur- face of fracture. This extreme is, however, connected by a series of forms with such cases as those of Amphioxns or the medusa, where the fragment develops nearly or quite as if it were a whole. In the tunicates the researches of Chabry i^^y), Driesch ('94), and Crampton ('97) show that an isolated blastomere of the two-cell stage undergoes a typical half-cleavage (Crampton), but finally gives rise to a nearly perfect tadpole larva lacking only one of the asymmetrically placed sense-organs (Driesch). Next in the series may be placed the frog, where, as Roux, Endres, and Walter have shown, a blastomere of the two-cell stage may give rise to a typical half-morula, half-gastrula, and half-embryo 1 (Fig. 182), yet finally produces a perfect larva. A further stage is given by the echinoderm-egg, which, as Driesch showed, undergoes a half-cleavage and produces a half-blastula, which, however, closes to form a whole before the gastrula-stage (Fig. 183). Perfectly formed though dwarf larvae result. Finally, we reach Auiphi- oxus and the hydromasae in which a perfect ** whole develoj)mcnt " usually takes place from the beginning, though it is a verv interest- ing fact that the isolated blastomeres of Aiiiphioxus sometimes show, in the early stages of cleavage, peculiarities of development that recall their behaviour when forming part of an entire embryo.-

We see throughout this series an effort, as it were, on the part of the isolated blastomere to assume the mode of development character- istic of a complete ^^^, but one that is striving against conditions that tend to confine its operations to the role it would have played if still forming part of an entire developing ^g^. In Amphioxits or Clytia this tendency is successful almost from the beginning. In other forms the Umiting conditions are only overcome at a later period, while in the ctenophore or snail they seem to afford an insurmount-

Fig. 190. — Partial development of isolated blastomeres of the gasteropod ^gg, Ilyanassa. [Crampton.] A. Normal eight-cell stage. B. Normal sixteen-cell stage. C. Half eight-cell stage, from isolated blastomere of the two-cell stage. D. Half twelve-cell stage succeeding. E. Two stages in the cleavage of an isolated blastomere of the four-cell stage ; above a one-fourth eight-cell stage, below a one-fourth sixteen-cell stage. able barrier to complete development. What determines the limita- tions of development in these various cases ? They cannot be due to nuclear specification ; for in the ctenophore the fragment of an iniseg- mented ^%^, containing the normal egg-nucleus, gives rise to a defec- tive larva ; and my experiments on Nereis show that even in a highly