The Cell in Development and Inheritance

of the egg-mother-cell becomes the ^%g, appropriating to itself the entire mass of the yolk at the cost of the others which persist in rudimentary form as the polar bodies." ^ Up to this point the facts are clear and intelligible. Before com- ing to closer quarters with them it will be useful to make a digression in order to consider some of the theoretical aspects of reduction ; though the reader must be warned that this will lead us into very uncertain ground traversed by a labyrinth of conflicting hypotheses from which no exit has yet been discovered.

The process of reduction is very obviously a provision to hold con- stant the number of chromosomes characteristic of the species ; for if it did not occur, the number would be doubled in each succeeding generation through union of the germ-cells.^ A number of writers have contented themselves with this simple interpretation, Oscar Hertwig, for example, regarding reduction as '' merely a process to prevent a summation through fertilization of the nuclear mass and of the chromatic elements." ^ A moment's reflection reveals the entire inadequacy of such an explanation. As far as the chromatin-mass is concerned, it does not agree with the facts ; for in reduction with tetrad-formation the chromatin-mass is reduced not to one-half, but to one-fourth. That reduction must mean more than mere mass-reduc- tion is moreover proved by the fact that the bulk of the nucleus mav enormously increase or decrease at different periods in the same cell, irrespective of the number of chromosomes. The real problem is why the number of chromosomes should be held constant. The

1 Of the many earlier attempts to interpret the meaning of the polar bodies, we need only consider at this point the very interesting suggestion of Minot ('77), afterward adopted by Van Beneden ('83), that the ordinary cell is hermaphrodite, and that maturation is for the purpose of producing a unisexual germ-cell by dividing the mother-cell into its sexual con- stituents, or "genoblasts." Thus, the male element is removed from the egg in the polar bodies, leaving the mature egg a female. In like manner he believed the female element to be cast out during spermatogenesis (in the " Sertoli cells "), thus rendering the spermatozoa male. By the union of the germ-cells in fertilization, the male and female elements are brought together so that the fertilized egg or oosperm is again hermaphrodite or neuter. This ingenious view was independently advocated by Van Beneden in his great work on Ascaris <y"^'^. A fatal objection to it, on which both Strasburger and Weismann have insisted, lies in the fact that male as well as female qualities are transmitted by the egg-cell, while the sperm-cell also transmits female qualities. The germ-cells are therefore non-sexual. The researches of many observers show, moreover, that all of the four spermatids derived from a spermatocyte become functional spermatozoa. Minot's hypothesis must, therefore, in my opinion, be abandoned.

Balfour doubtless approximated more nearly to the truth when he said. " In the formation of the polar cells part of the constituents of the germinal vesicle, which are requisite for its functions as a complete and independent nucleus, is removed to make room for the supply of the necessary parts to it again by the spermatic nucleus" ('So, p. 62). He fell, however, into the same error as Minot and Van Beneden in characterizing the germ-nuclei as " male " and "female"; and, as shown at pages 194. 353. it has been found that a single germ- nucleus is able to carry out development of an embryo without union with another.

deeper meanin^^ of the phenomena was first seriously considered by Weismann in his essays of 1S83 and 1887 ; and, although his conclu- sions were of a highlv'speculative character, they nevertheless gave so Fig. 120. — Reduction in the spermatogenesis of Ascaris mcgalocephala,v7iX. bivalens. [Brauer.] 1 .^l-G. Successive stages in tlie division of the primary spermatocyte. The original reticulum undergoes a ver)' early division of the chromatin-granules which then form a doubly split spireme- tir'ad.Zf. This shortens (6), and breaks in two to form the two tetrads (Z) in profile, .fi" viewed endwise). F. G.H. Urst division to form two secondary spermatocytes, each receiving two dyads. /. Secondary sp<Tmatocyte. J. K. The same dividing. Z. Two resulting spermatids, each with two single chromosomes and a centrosome.

great a stimulus to the study of the entire problem that his views deserve special attention. Weismann's interpretation was based on a remarkable paper published by Wilhelm Roux in 1883,^ in whicTi arej! 1 For division of the spermatogonia see Fig. 55; for the corresponding phenomena in var. univalens see Fig, 148. developed certain ideas which afterward formed the foundation of Weismann's whole theory of inheritance and development. Roux argued that the facts of mitosis are only explicable under the assump- tion that chromatin is not a uniform and homogeneous substance, but differs qualitatively in different regions of the nucleus ; that the col- lection of the chromatin into a thread and its accurate division into two halves is meaningless unless the chromatin in different regions of the thread represents different qtcalities which are to be divided and distributed to the daughter-cells according to some definite law. He urged that if the chromatin were qualitatively the same throughout the nucleus, direct division would be as efficacious as indirect, and the complicated apparatus of mitosis would be superfluous. Roux and Weismann, each in his own way, subsequently elaborated this con- ception to a complete theory of inheritance and development, but at this point we may confine our attention to the views of Weismann. The starting-point of his theory is the hypothesis of De Vries that the chromatin is a congeries or colony of invisible self-propagating vital units or biopJiores somewhat like Darwin's ** gemmules " (p. 12), each of which has the power of determining the development of a particu- lar quality. Weismann conceives these units as aggregated to form units of a higher order known as ''determinants," which in turn are grouped to form ** ids," each of which, for reasons that need not here be specified,^ is assumed to possess the complete architecture of the germ-plasm characteristic of the species. The "ids" finally, which are identified with the visible chromatin-granules, are arranged in linear series to form " idants " or chromosomes.

It is assumed further that the '* ids " differ slightly in a manner corresponding with the indi- vidual variations of the species, each chromosome therefore being a particular group of shghtly different germ-plasms and differing quali- tatively from all the others. We come now to the essence of W^eismann's interpretation. The end of fertilization is to produce new combinations of variations by the mixture of different ids. Since, however, their number, like that of the chromosomes which they form, is doubled by the union of two germ-nuclei, an infinite complexity of the chromatin would soon arise did not a periodic reduction occur. Assuming, then, that the " ances- tral germ-plasms" (ids) are arranged in a linear series in the spireme- thread or the chromosomes derived from it, Weismann ventured the prediction {'%j) that two kinds of mitosis would be found to occur. The first of these is characterized by a longitudinal sj)litting of the thread, as in ordinary cell-division, " by means of which all the ances- tral germ-plasms are equally distributed in each of the daughter-nuclei after having been divided into halves." This form of division, which

he called eqnal division ( Aequationsthcilung), was then a known fact. The second form, at that time a ])urely theoretical postulate, he as- sumed to be of such a character that each dau*;hter-nucleus should receive onlv half the number of ancestral germ-plasms possessed bv the mother-nucleus. This he termed a reducing division ( Re- duktionstheilung), and suggested that this might be effected either by a tninsvtrsf division of the chromosomes, or by the elimination of entire chromosomes without division.^ By either method the number of ** ids " would be reduced ; and W'eismann argued that such reduc- ing divisions must be involved in the formation of the polar bodies, and in the parallel phenomena of spermatogenesis.

The fulfilment of Weismann's prediction is one of the most inter- estiniT results of recent cvtological research. It has been demon- strated, in a manner which seems to be incontrovertible, that the reducing divisions postulated by Weismann actually occur, though not precisely in the manner conceived by him. Unfortunately for the general theory, however, transverse divisions have been cer- tainlv determined in only a few types, while in others, of which Ascaris is the best-known example, the facts thus far known seem clearly opposed to the assumption. On the whole, the evidence of reducing divisions, i.e. such as involve a transverse and not a longi- tudinal division of the chromatin-thread, has steadily increased ; but it remains quite an open question whether they have the significance attributed to them by Weismann.

In considering the origin of the tetrads or their equivalents, it should be borne in mind that true tetrad-formation, as described above, has only been certainly observed in a few groups (most clearly in the nematodes and arthropods). But even in cases where the chromatin does not condense into actual tetrads these bodies are represented by chromosomes in the form of rings, crosses, and the like, which arc clo.sely similar, and doubtless equivalent, to those from which actual tetrads .arise, and present us with the same prob- lems. With a few apparent exceptions, described hereafter, the tetrads of their equivalents alwaws arise by a double division of a single primary chromatin-rod or mass. Nearly all observers agree further that the number of primary rods at their first appearance in the germinal vesicle or in the spermatocyte-nucleus is one-half the usual number of chromosomes, and that this numerical reduction is due to the fact that the spireme-thread segments into one-half the

usual number of pieces. Apparently, however, there are two radi- cally different types of tetrad-formation as follows. In the first type the tetrad arises by one longitudinal and one trans- verse division of each primary cJironiatin-rod, the latter effecting the reduction demanded by Weismann's hypothesis(Fig. 121, I). To give the usual graphic representation, let us, for the sake of discus- sion, assume the somatic number of chromosomes to be four, designating the spireme-thread as a b c d, £ each letter representing a chromosome, each of which we may in turn assume to consist of a series of four granules or ** ids "(Fig. 121). 4 <^ S^^8

In ordinary mitosis the spi- reme would segment into a — b — c — d, which then would divide lengthwise to form pairs of identical sister ^ abed To form the tetrad, on the other hand, the spireme first segments into two rods ab and cd, each of which, in view of its subsequent his- ^^ s^ks tory, may be regarded as bivalent, representing two chromosomes united end to end (Vom Rath, Riickert, Hacker). Each of type) ; II, witii two longitudinal divisions {Ascaris type).

divides once longitudinally, thegranulesof each constituent of the tetrad fuse to form dyads — — £^, and once transversely, giving the tetrads — — -' Inspection of Fig. 121, I, shows that through the second or transverse division, each member of the tetrad receives only half the number of ids contained in the original segment. This number, four, is the same as that assumed for a single chromosome ; and, since each of the two tetrads contributes one chromosome to the germ-cell, the latter receives

but half the usual number both of chromosomes and of ids. This mode of tetrad-formation has been most clearly demonstrated in insects and copepods. and an equivalent ])rocess occurs also in mollusks, annelids, turbellarians, and some other animals, as described beyond. In the second type, illustrated especially by Ascaris, the tetrad is apparently formed by t:^>o longitudinal divisions of each primary chromatin-rod, and no reducing division occurs. If, therefore, we adopt the same terminology as before, we have first ab and cd. then

it _ 12/, and hnall V — !— - — — , bv two longitudinal divisions. In ab id ' ab ! ab cd ^ cd this case, according to Brauer's careful studies, each chromatin-granule (** id") divides at each longitudinal division of the primary rod. The four chromosomes of the tetrad are therefore exactly equivalent, being derived from the same region of the spireme-thread, and containing the undiminished number of " ids " (Fig. 121, II). The contradiction may be stated in a different way. In the first type of tetrad formation, the number both of granules and of chro- mosomes is first doubled {i.e. in the assumed case, through the forma- tion of two tetrads, each consisting of four chromosomes, or eight in all), and then reduced to half that number by the two successive matu- ration-divisions. In the second type, on the other hand, the number of chromosomes is likewise doubled, but that of the granules is quad- rupled, so that, although in both types the two maturation-divisions reduce the number of elirouibsonies to one-half, only in the first type do they reduce the number of granules or **ids," as Weismann's hvpothesis demands. We must therefore distinguish sharply between the reduction of the chromosomes and that of the *'ids." The former is primarilv effected b\' the segmentation of the jDrimary s])ireme- thread, or the resolution of the nuclear reticulum, into one-half the usual number of segments (?>. the "pseudo-reduction" of Riickert); and here the real secret of the reduction of the chromosomes lies. The reduction of the "ids," if they have any real existence, is a distinct, and as yet unsolved, question.

We may now consider some of the phenomena in detail, though the limits of this work will only allow the consideration of a few typical cases. {a) Tetrad for)nation ivith one Longitudinal and one Trajisverse Division. — In many of the cases of this type the tetrads arise from ring-shaped bodies which are analogous to the ring-shaped chromo- somes occurring in heterotypical mitosis (p. %6). First observed by Henking('9i) in /^j';r//^rf7;7i-, tetrad-origin of this type has since been found in other insects by Vom Rath, Toyama, Paulmier, and others,

in copepods by Riickert, Hacker, and Vom Rath, in pteridophytcs by Calkins and Osterhout, in the onion, Alliiiui, by Ishikawa, and in various other forms where their history has been less clearly made out. The genesis of the ring was first determined by \k)w\ Rath in the mole cr'ickQt {Gryllotalpa, '92), and has been thoroughly elucidated by the later work of Riickert ('94), Hacker ('95, i), and Paulmier ('99). All these observers have reached the same conclusion; namely, that the ring arises by the longitudinal splitting of a primary chromatin-rod, the two halves remaining united by their ends, and opening out to form a ring. The ring-formation is, in fact, a form of

Fig. 122. — Origin of the tetrads by ring-formation in the spermatogenesis of the mole-cricket Gryllotalpa. [\'(m RATH.] A. Primary spermatocyte, containing six double rods, each of which represents two chromo- somes united' end to end and longitudinally split except at the free ends. B. C. Opening out of the double rods to form rings. D. Concentration of the rings. E. The rings broken up into tetrads. F. First division-figure established. heterotypical mitosis (p. S6). The breaking of the ring into four parts involves, first, the separation of these two halves (corresponding with the original longitudinal split), and second, the tniusrnsc division of each half, the latter being the reducing division of W'cismann. The number of primary rods, from which the rings arise, is one-half the somatic number. Hence each of them is conceived by Vom Rath, Hacker, and Riickert as bivalent or double ; i.e. as representing two chromosomes united end to end. This appears with the greatest clearness in the spermatogenesis of G)-yUotalpa (Fig. 122). Here

the spireme-thread splits lengthwise before its segmentation into rods. It then divides transversely to form six double rods (half the usual number of chromosomes), which open out to form six closed rings. These become small and thick, break each into four parts, and thus Fig. 123. — I-ormation of the tetrads and polar bodies in Cyclops, slightly schematic, full number of tetrads is not shown.) [RUCKKRT.] A. Germinal vesicle containing eight longitudinally split chromatin-rods (half the somatic number). li. Shortening of the rods; transverse division (to form the tetrads) in progress. C. Position of the tetrads in the first polar spindle, the longitudinal split horizontal. D. Ana- phase; longitudinal divisions of the tetrads. E. The first polar body formed; second polar spindle with the eight dyads in position for the ensuing division, which will be a transverse or reducinfT division.

ive rise to six typical tetrads. An essentially similar account of the ring-formation is given by Vom Rath in EucJicBta and Calanus, and by Riickert in Hctcrocopc and Diaptomns. That the foregoing interpretation of the rings is correct, is beauti- fully demonstrated by the observations of Hacker, and especially of Ruckert, on a number of other copepods {Cyclops, Ca7ithocamptiis\ in which rings are not formed, since the splitting of the primary chromatin-rods is complete. The origin of the tetrads has here been traced with especial care in Cyclops strciiims, by Ruckert ('94), whose observations, confirmed by Hacker, are quite as convincing as those

Fig. 124. — Diagrams of various modes of lelrad-formation. [Hacker.] a. Common starting-point, a double spireme-thread in the germinal vesicle; d. common re- sult, the typical tetrads ; b. c. intermediate stages : at the left the ring-formation (as in Diaptomiis, Gryllotalpa, Heterocope) ; middle series, complete splitting of the rods (as in Cyclops accordini: to Ruckert, and in Canthocaniptus) ; at the right by breaking of the V-shaped rods (as in Cy.i ; strenuus, according to Hacker.

of Brauer on Ascains, though they led to a diametrically opposite result. The normal number of chromosomes is here twenty-two. In the germinal vesicle arise eleven threads, which split lengthwise (Fig. 123), and finally shorten to form double rods, manifestly equivalent to the closed rings of Diaptomus. Each of these now segments transversely to form a tetrad group, and the eleven tetrads then place themselves in the equator of the spindle for the first polar body (Fig. 123, C\ in such a manner that the longitudnial split is transverse to the axis of the spindle. As the polar body is formed, the longitudinal halves of the tetrad separate, and the formation of the first polar body is thus demonstrated to be an "equal division" in Weismann's sense. The eleven dyads remaining in the eggs now rotate (as in Ascaris),

Fig. 125. — Germinal vesicles of various eggs, showing chromosomes, tetrads, and nucleoli. A. .\ copepod {Heterocope) showing eight of the sixteen ring-shaped tetrads and the nucleo- lus. [krcKKki.] /?. I^ter stage of the same, condensation and segmentation of the rings. [RUCKERT.] C. " Cyclops strenuus," illustrating Hacker's account of the tetrad-formation from elongate double rods; a group of " accessory nucleoli." [Hackkr.] D. Germinal vesicle of an annelid {^Op/nyotrocha) -^how-xng nucleolus and four chromosomes.

SO that the transverse division lies in the equatorial plane, and are halved during the formation of the second polar body. The division is accordingly a "reducing division," which leaves eleven single chro- mosomes in the egg. Paulmier's work on Auasa and other Hemip- tera ('99) gives the same result as the above in regard to the origin of the tetrads (Figs. 126, 127). The process is, however, slightly complicated by the fact that no continuous spireme-thread is formed, while the rings are often bent or twisted and never open out to a

circular form. They finally condense into true tetrads which are successively divided into dyads and monads by the two divisions; but it is an interesting fact that the order of division occurring in the copepods appears here to be reversed, the first division being the transverse and the second the longitudinal one — a result agreeinc^ with Henking's earlier conclusion in the case of Pyrrochoris. '^Oster'- hout ('97) and Calkins ('97) independently discovered tetrads in the vascular cryptogams (Equisetum, Ptcris), and the last-named observer finds that in Ptej'is they may arise either from rings, as in (iryllotalpa or Heterocope, or from double rods as in Cyclops, the halves in the latter case being either parallel or forming a cross. This longitu- dinal spHt, occurring in the spireme, is followed by a trans\x'rse division by which the tetrad is formed. Tetrads having an essentially similar mode of origin are also described by Atkinson ('99) in Ari- scEma, and tetrad-formation is nearly approached in Allium according to Ishikawa ('99).^ These cases are considered at page 263.

Resume. In all the foregoing cases the tetrads arise from a spi- reme which splits lengthwise, segments into one-half the somatic number of rods (each longitudinally divided) and each of the latter divides transversely to form the tetrad. When the ends of the daughter-chromosomes resulting from the longitudinal split remain united (as in insects) ring-forms result, and the earlier phases of tetrad- formation are thus identical with those of heterotypical mitosis. When the split is complete, so that the ends remain free, double rods result; while, if the daughter-chromosomes remain temporarily united at the middle or at the end, X-, Y-, and V-shaped figures may arise. In all these forms tetrad-formation is completed by the com- plete separation of the daughter-rods, the transverse division of each in the middle, and the condensation of the four resulting bodies into a quadruple mass. As will be shown in Section C (p. 258) the transverse division is in many forms delayed until after sepa- ration of the longitudinal halves. In such cases no actual tetrads are formed, though the result is the same.

{I)) Second Type. Tetrad-formation ivith two Longitudinal Divi- sions. — The only accurately known case of this type is Ascaris, the object in which tetrads were first discovered by Van Beneden in 1883. Carnoy ('S6, 2) reached the conclusion that the tetrads in some other nematodes {Ophiostomum, Ascaris clavata, .i. lumbricoidcs) arose by a double longitudinal splitting of the primary chromatin-rods. 1 Vom Rath ('93, '59) has endeavoured to show that a process involving the formation of true tetrads occurs in the salamander and the frog, but the later and more accurate studies of Meves ('96) seem to leave little doubt that this was an error, and that the tetrads observed in these forms are not of normal occurrence, as Flemming ('87) had earlier concluded. Cf. p. 259.

In the first of his classical cell-studies Boveri ('87, i) reached the same result through a careful study of Ascaris mcgaloccphala, showing that each tetrad appears in the germinal vesicle in the form of four j)arallel rods, each consisting of a row of chromatin-granules (Fig. 1 17, A-C ). He believed these rods to arise by the double longitudinal splitting of a single primary chromatin-rod, each cleavage being a A. Resting spermatogonium with single plasmosome and two chromatin-nucleoli. B. Equa- torial plate of dividing spermatogonium ; twenty large and two small chromosomes, 6'. Final spcrmatogonium-division. D-I. Prophases of first maturation-division. D. E. Synapsis, with single chromatin-nucleolus. /'. Segmented split spireme. G. H. Formation of the tetrad-rings. //. /. Concentration of the rings to form tetrads.

preparation for one of the polar bodies. In his opinion, therefore, the formation of the polar bodies differs from ordinary mitosis only in the fact that the chromosomes split very early, and not once, but twice, in preparation for two rapidly succeeding divisions without an intervening resting period. He supported this view by further obser- vations in 1890 on the polar bodies of Sagitta and several gastero- pods, in which he again determined, as he believed, that the tetrads

arose by double longitudinal splitting. An essentially similar view of the tetrads was taken by Hertwig in 1890, in the spermatogenesis of Ascaris, though he could not support this conclusion by very con- vincing evidence. In 1893, finally, Brauer made a most thorough and apparently exhaustive study of their origin in the spermatogene- sis of Ascaris, which seemed to leave no doubt of the correctness of Boveri's result. Every step in the origin of the tetrads from the retic- ulum of the resting spermatocytes was traced with the most pains- taking care. In the early prophases of the first division the nuclear reticulum breaks up more or less completely into granules, which