“Plagiostomorum spermatosomata, quæ magnitudine et peculiari quadam forma in primis ad evolutionis studium apta sunt, hac ratione iara plurium observations in se contraxerunt.”—LAVALETTE ST. GEORGE.

1. DURING the development of a metazoan embryo, after the differentiation of the generative cells from those of the general somatic “anlage,” the reproductive elements pursue a course of evolution peculiar to themselves. Instead of attaining to the high specialisation, decay, and final dissolution, characteristic of somatic tissues, their variation is of much less amplitude, but cyclical, and returns at length to the production of elements similar to those in which both series started.

In animals, the course of such a reproductive cycle appears at first sight tobe differentiated into two distinct periods of activity, the one extending from the earliest embryonic development of the generative elements to the commencement of the proper spermato- or ovo-genesis, the other beginning with the spermato- or ovo-genesis and ending with the formation of the mature reproductive cells.

In reality, however, the transition from the first of these periods to the second is much more expressive of changes incident to the surrounding parts, than any alteration in the structure of the generative cells themselves.

In fact, no definite change occurs in them till late in the second period, when .the advent of the so-called “reduction process” produces a sharp and definite alteration in the morphological value of all the elements affected. So characteristic is this change, that the time of its commencement can be used as a point of reckoning in all those reproductive cycles in which it is apparent.

For the sake of clearness, that part of the reproductive cycle in Elasmobranchs which comes before the proper spermato- or ovo-genesis will be called the primary or embryo nicperiod, while the cellular generations of spermato-or ovo-genesis before and after the numerical reduction of the chromosomes will be distinguished as those of the first and second spermato- or ovo-genetic series.

It is my purpose in the present paper to give a detailed exposition of the changes witnessed in the generative cells during the reproductive cycle of Elasmobranchs, after the completion of the embryonic period in the male, i.e. during the proper spermatogenesis in the fish, and then to draw such conclusions as may seem legitimate from a comparison with other forms.

2. The materials with which the present investigation was carried out were obtained while I occupied the British Association Table in the Naples Laboratory during the months of October, 1893, to July of the following year. They consist of a large number of Elasmobranch testes, including those of Scyllium canicula, Scyllium catulus, Pristiurus, Torpedo, Raja maciorhynchus, and Raja maculata. The testes were cut up into small pieces about the size of half a cubic centimetre, and fixed in various ways. I obtained the most successful preparations after the use of Flemming’s strong solution, Hermann’s fluid, osmic acid in various strengths, and corrosive sublimate, both with and without acetic acid. Valuable comparative material was also obtained by treating the testes with glacial acetic acid, and washing quickly in water, by tearing up the fresh material in acetic carmine, by fixing in a 2 or 3 per cent, solution of formic aldehyde, by the use of Carnoy’s fluid, and last but not least, by a formic acid method which I hit on quite by accident. This consists in placing small fragments of the living testis in a 50 per cent, solution of formic acid for a few seconds, and then transferring directly to 50 per cent, alcohol, after which they are treated for sectioning in paraffin, in the usual way. By this means the chromosomes were in some cases rendered admirably distinct, but the fixation, so far as I have yet tried it, renders the material difficult to stain.

Although the cells of these different kinds of fish examined necessarily differ somewhat inter se, the differences do not appear to be of any morphological importance, and the following description, although taken more especially from those of Scyllium canicula, is, so far as I am aware, applicable to them all.

3. The successive generations of the smaller and smaller cells which gradually fill up the testicular ampullæ during the first spermatogenetic period are all alike, and I find no essential difference between their structure and that of the single embryonic elements (fig. 1) budded off into the stroma from which they gradually arise.

The chromatin is dispersed as a coarse reticulum throughout the interior of the nucleus, and is not in the least specially related to its surface; in fact it is really denser towards the interior than at the periphery, and contains a large oval or flask-shaped nucleolus, generally attached to the nuclear wall (figs. 2, 11, 12, n.).

The chromatic threadwork itself is composed of innumerable staining granules, embedded in a scaffolding of clearer substance (linin) (fig. 12), and the unoccupied nuclear space is filled up with a thin nuclear sap.

4. The cytoplasm presents the usual fine reticulation in its substance (figs. 1, 11), and the nucleus is placed excentrically within its mass, so that there is more cell body on the one side than on the other.

The whole reticulation of the cytoplasm is disposed radially towards a point just outside that part of the nuclear wall which faces the larger mass of the cytoplasm (fig. 2, r.), and the point itself is occupied by two small centrosomes (c.), which can be stained bright red by treatment with fuchsin and orange G.

There is hardly any archoplasmic substance round the centrosomes, and they, together with their cytoplasmic radiation, which extends quite out to the periphery of the cell,1 constitute a good example of what I have elsewhere called2 a simple sphere.

There are one or two small chromatic bodies in the cytoplasm (fig. 2, b. c.).

5. Just as the successive generations of resting cells in the first spermatogenetic period are all alike, so also are the divisional metamorphoses by which they are produced.

At the commencement of mitosis the nuclei become swollen, their smooth round contours appearing as if turgid with an excess of intra-nuclear sap (fig. 13), and at the same time the chromatic framework shortens up into a lesser number of stouter threads. But I have not seen any indication of early splitting in this threadwork or the granules which compose it, like that given by Brauer1 in the corresponding stage of the spermatogenesis of Ascaris. The individual granules (microsomes) become completely massed together into axial cores, from which delicate filamentous radii of linin (figs. 12, 13,l.) spread in all directions, and the continuation of this metamorphosis in the chromatin results eventually in the formation of one or two long chromatin threads coiled round the inner surface of the nuclear membrane (figs. 14, 15), the nucleolus lying a little more within.

These threads, as soon as they are found, break up apparently simultaneously into twenty-four bent rods, which form the twenty-four chromatic elements characteristic of the divisions of the first spermatogenetic period ; and at the same time the nucleolus, becoming smaller and smaller, breaks up and disappears.

6. Concomitantly with the above intra-nuclear changes, the centrosomes, originally occupying the focus of the cytoplasmic radiations, separate from one another and pass successively through the positions represented in c., figs. 11, 12, 13, 14.

Owing to the absence of the archoplasmic constituent of the sphere, there is in these cells no real archoplasmic spindle formation (as in the spermatocytes of Salamandra described by Hermann2). Each centrosome, with its crown of radiations, simply travels away from the other, until, at the period of the nuclear evolution reached in the last paragraph, they lie on opposite sides of the nucleus, with almost the whole diameter of the cell between them (cf. c., figs. 12, 13, 14).

7. As soon as the cytoplasmic conditions just described have been attained, the nuclear wall becomes irregular and disappears, while the chromosomes, collecting under the contraction of their connecting linin filaments, form a long oval mass stretched across the nuclear sap between the centrosomes (figs. 15, 16, 17).

At first the chromosomes are attached to the centrosomes by a few faint protoplasmic strands, which are apparently of cytoplasmic origin ; but as time goes on the chromatin assumes a more and more equatorial position, and the liuin filaments, being left stretched towards the centrosomes, help to form the central portion of an achromatic spindle figure, the equatorial moiety of which is nuclear, while its extreme ends appear to be cytoplasmic (figs. 17, 18, 19).

8. A portion of the astral radiation round the centrosomes becomes connected with the outer ends of the chromatic rods, clothing the inner achromatic spindle with a sheath of cytoplasmic fibres (hm., fig. 19), structurally equivalent to Hermann’s “outer mantle.”

9. The chromosomes to which these fibres are attached assume the form of short bent rods, and lie (fig. 19) at all angles on the equatorial plane, being by no means specially related to the surface of the spindle figure, and in surface view they consequently present the appearance of a somewhat irregular chromatic disc (fig. 21).

10. The achromatic spindle would thus appear to have a dual origin, its superficial portion and extreme ends originating in the cytoplasm, while its greater internal and equatorial mass arises from the nucleus—a state of things approximately coinciding with Flemming’s1 views respecting its complex origin in Amphibia, as opposed to the general acceptation of its wholly cytoplasmic nature among plants.

11. When the equatorial plate is fully formed, the chromosomes, after becoming extremely broad and flat, split longitudinally down the middle, each into two daughter-threads, which gradually separate from one another towards the spindle-poles (figs. 19, 20, 22, 23). During their transit those daughter-elements, which were at first internal, work outwards to the surface of the spindle in such a way that by the time they are halfway from the equator to the poles, the chromosomes of each daughter-nucleus have assumed the well-known open ring-form of the diastral figure represented in fig. 24.

12. In consequence of this outward motion of the inner chromosomes, the spindle (now intra-zonal) fibres with which they are connected become drawn out from their original axial position, and form a central fibrous tube (fig. 24, i. s.) enclosed by (i) the remains of the nuclear sap, (ii) the external spindle (intra-zonal) fibres (o. s.), and (iii) some linin filaments left stretched between the separate chromosomes, which last, in my opinion, represent the true1 “Verbindungsfäden.” The “Verbindungsfaden “and the intra-zonal fibres now become indistinguisbably fused, while the differentiation of the achromatic spindle into an inner and an outer sheath, becoming more distinct, gives to that structure the appearance of two concentric tubes, one of which (o. s., fig. 24) is stretched directly between the outer edge of the chromatic rings, the other (i. s., fig. 24), passing internally through them, to its termination in the centrosomes. The unstained fluid which separates the outer from the inner of these sheaths is that which previously filled the interspaces between the younger spindle-fibres, and it was once the parental nuclear sap. It contains a few irregular chromatic particles (fig. 24, b. c.), which appear to have been left as débris of the previous chromosome formation, and which sooner or later pass (fig. 26, b. c.) into the cytoplasm of the cell.

While the above changes are in progress the centrosomes become gradually surrounded by a dusky zone (figs. 25,26, 27), which is caused by the shortening up and coalescence of the cytoplasmic fibres between them and the chromosomes, i. e, by those described above (§ 8) as structurally equivalent to Hermann’s mantle.

13. The chromosomes in the two daughter-rings (fig. 24) are at first quite distinct from one another, although lying closely side by side, but as time goes on they fuse together, until the chromatin eventually forms two solid chromatic rings, one in each daughter-cell (figs. 25, 26).

14. About this time the outer spindle-fibres begin to spread so widely in the equatorial plane (fig. 24) that they actually come in contact with the membrane of the cell, and at each of these rather angular connections there appear slight thickenings of the fibres (bi.), which stain, and thus constitute an interesting stepping-stone between the true cell-plate and Flemming’s1 intermediate bodies.

The chromatic rings now gradually lose their original connection with the outer spindle-fibres, which begin to bulge out, and pass round them to the poles (fig. 27′) ; the chromatic rings are thus left in a surrounding vacuole, but the core of fibres (o. i. s., fig. 27 a) still passes through them to the poles. The bulging out of the spindle-fibres round the nucleus increases, and is accompanied by a corresponding collapse of the same in the division plane of the daughter-cells (figs. 25, 26, 28). This brings the outer and the inner spindle-tube together in the division plane (figs. 26, 27, 28), and the whole spindle figure at last acquires the appearance of a sharply differentiated fusiform body between the daughter-cells (fig. 27 a). The terminal portions of this body (the remains of the spindle-fibres) as they pass round the daughter-nuclei (fig. 27 a, n.g.), are at first distinctly seen, but they become shortly indistinguishable from the surface of the vacuole and are consequently lost; but the conical extremities of the fusiform equatorial portion which remains are still prolonged as delicate protoplasmic filaments, which extend towards the nuclei in each daughter-cell (figs. 29, 30). These thread-like prolongations are the remains of that inner core of spindle-fibres above described (§ 12) as passing through the daughter-rings. In apolar view (fig. 27, l.) they perforate the daughter-nuclei very much to one side, and the little orifice (o.) is all that is left of the originally wide passage through the chromatin.

15. The closing up of the chromatic rings commences on the equatorial side, and is produced by the formation of a thin chromatic floor. In consequence of this, the spheres (fig. 27, a. b.) appear to occupy the hollow of a little nuclear cup, and by the continuance of this filling-up process the remnant of the inner spindle-core is pushed gradually to one side, and eventually out of the nucleus altogether, but it continues to pass round the nucleus (in a more or less deep furrow) (fig. 27 a, n.g.) towards the spheres.

16. The course of the terminal spindle-filaments becomes generally coincident with the surface of the vacuole about each nucleus, and they consequently take a curved course from each end of the fusiform remains of the original spindle-figure (between the cells) to the spheres in the polar faces of the nuclei. So that the whole arrangement of the daughter-nuclei and spindle-fibres at this time bears (see fig. 27 a) a curious resemblance to the figures seen in the divisions of the micronucleus1 during the conjugation of many infusoria.

17. The chromatin in the daughter-nuclei now blows up once more into a foam, and eventually completely fills the vacuoles originally surrounding them (figs. 28, 29, 30) while a nucleolus appears in the reticulum of each, generally at the base of the shallow depression (ng., fig. 29), which persists as the remains of the nuclear cup described above (§ 15). This depression, together with the spheres, is gradually rotated somewhat to the equatorial side (as in fig. 29), and the chromatic granules existing in the cytoplasm, becoming fewer in number and larger in size, assume the characters of the chromatic bodies described by Hermann in the spermatogenesis of Mammalia (fig. 30, b. c.).

18. The cells are now practically at rest once more, but the fusiform spindle remnant, with its equatorial band of intermediate bodies (fig. 30, b. i.), continues long after the daughter-cells have come to rest, and eventually degenerates and vanishes in the equatorial plane.

Mitoses of the above description are carried out with hardly any variation in their details, through all the cellular divisions of the first spermatogenetic period, and in the course of this the features which appear to be of primary comparative, importance may be summarised as follows :

  • The existence in the resting cells of a large round nucleolus lying near the nuclear periphery.

  • The evolution during the prophasis of division, of twenty-four bent chromosomes, which shorten up, and split longitudinally in half to form the same number of chromosomes, twenty-four, in each daughter-cell.

  • The existence of an extra-nuclear attraction sphere, which, during this period of the spermatogenesis, is practically destitute of archoplasm, being surrounded by a simple cytoplasmic radiation like that observed in many forms of tissue cells.

  • The consequent non-formation of an archoplasmic spindle figure, and the dual origin of this latter structure, partly from the simple cytoplasmic radiation, partly from the intra-nuclear substance.

  • The differentiation of the spindle during the dyastral figure into an outer and an inner fibrous sheath, which, after the escape of the parental nuclear sap, collapse and coalesce, forming a delicate connecting thread between the attraction spheres of both daughter-cells.

  • The formation of extra-nuclear chromatic bodies from the débris of the nuclear chromatin.

19. As I have already pointed out, the transition from the first into the second spermatogenetic period is completed in the cells during the rest which follows the last division of the first, and when the elements in the ampullæ are seven or eight rows deep (fig. 34). Such cells, although at first retaining the characteristics of those of previous generations, gradually acquire new ones, but so gradually that it is some time before we realise the profound nature of the changes wrought, and that, while yet apparently at rest, the cells have passed completely over from the first into the second spermatogenetic period. The commencement of this metamorphosis is marked by an increasing fineness of the reticulum in the nuclei, which continues to increase until cells with nuclear elements like that represented in fig. 35 are seen, and about the same time there appears a curious secondary nucleolus surrounded by a vacuole (fig. 31 n′), which, so far as I can ascertain, is in these fishes diagnostic of the change. After a while the nuclear threadwork again grows coarser and thicker, displaying at the same time a peculiar tendency to contract to one side of the nucleus, leaving a great clear space (fig. 39) across which stretch numerous linin filaments. The contraction is not so marked when the cells have been preserved with osmic acid, nor on the outside of sections which have been preserved with Flemming’s fluid, where the osmium has acted directly upon the cells. I have, however, seen it in elements of Torpedo which were simply immersed in dilute glycerine; and whether it exists in nature or not, the cells display at this period, and at no other, a remarkable tendency to have their chromatin contracted, in consequence of some internal change which renders these nuclear figures diagnostic of the particular period in question. Similar figures have been obtained at corresponding periods in the spermatogenesis of Amphibia, Mammals, Nematodes,1 and various Arthropods,2 and I do not think it probable that the contraction in many of these cases has anything to do with the reagents used.

20. In the cytoplasm the conversion from the first to the second spermatogenetic period is marked by a gradual increase in the small dark zone about the centrosomes, until it eventually attains the dimensions of a veritable spermatic “Nebenkern” or archoplasm (figs. 35, 36, 37), and from what has been said (§ 12) it follows that this body is here of an entirely cytoplasmic origin. The archoplasm, with its contained centrosomes, is at first closely applied to that part of the nuclear wall within which the curious lopsided condensation of the chromatin goes on.

21. The fine-meshed, tightly-coiled condition of the chromatin persists some time, but it gradually resolves itself into a coarse chromatic network on the nuclear periphery (figs. 37, 38). The strands of this network are sharply polarised towards the position occupied by the archoplasm and the centrosomes. The large oval nucleolus present in the resting cells of the first spermatogenetic period becomes now’ somewhat modified, both in position and character. Instead of being disposed casually along the nuclear circumference, it takes a position, generally, but not always, in line with the long axis of the archoplasm (fig. 37, n.). Along this line there is still to be seen the secondary nucleolus (fig. 36, 37, n.1) surrounded with a vacuole, which I described in § 19.

These two peculiar forms of nucleoli are always to be found after the transition from the first into the second spermatogenetic period, and throughout all the generations of the latter.

22. The archoplasm, which at first lies closely applied to the nuclear wall, during the early stages of the conversion of the first into the second spermatogenetic period, migrates away, quite into the cell body, while the two centrosomes which it contains, moving faster in the same direction, appear shortly on its exterior surface just beneath the membrane of the cell (fig. 37).

23. The advent of the first division in the second spermatogenetic period is characterised by the strong polarisation of the chromatin, represented in fig. 37. The chromatic strands are seen on close examination to be composed of a thick core of innumerable microsomes, which, collecting together into groups, give to the strands their curious monilated appearance, also described by Hermann1 in the prophasis of the great heterotype division of the spermatogenesis in Salamandra. These monilations in Elasmobranchs, however, do not consist of one large microsome, as Hermann’s figure would lead one to expect, but are each formed by a group of numerous chromatic granules,and these are embedded in a scaffolding of linin. Delicate connecting filaments of this substance spread from the monilations on the threads in all directions. The polarised threadwork is disposed throughout the nucleus in long parallel loops (figs. 37, 38), the free ends of which, if they exist, are difficult to discern. After a time the threads begin to show longitudinal splitting (figs. 38, 40), and the double ropes thus formed, dividing into equal segments, eventually give rise to twelve thick loops which (fig. 42) form the twelve ring chromosomes (fig. 43) typical of the divisions of the second spermatogenetic period.

24. There are thus, after the rest of transformation, only half as many chromosomes, i.e. separate chromatic masses, as there were before, and the halving of their number, being brought about while the nuclei are still at rest, is to that extent comparable to what is now known to go forward during the maturation of the reproductive elements of plants. I therefore propose the term Synaptic phase2 to denote the period at which this most important change appears in the morphological character of reproductive cells.

25. Concomitantly with the formation of the twelve ring chromosomes, the centrosomes (figs. 40,41, c.) begin to separate, and their greatly enlarged archoplasmic envelope (Nebenkern) (fig. 41,a.) is drawn out between them into a little archoplasmic spindle (fig. 42), strictly comparable to that described by Hermann during the division of the spermatocytes of Salamandra. As in the divisions of the previous spermatogenetic period, the separation of the centrosomes occurs with great rapidity, the archoplasm being drawn asunder into two parts (figs. 43 and 44), although it sometimes presents the appearance of a fine achromatic line stretched round the nuclear membrane. My preparations indicate both these methods of procedure.

The protoplasmic contents of the cell become radially disposed, not directly to the centrosomes, as in the divisions of the previous spermatogenetic period, but towards the outer surface of the daughter-archoplasms (fig. 43, r.), and it consequently follows that the sphere of the first period is structurally less complex than that of the second. In cells which possess an archoplasm, any radiation in the cytoplasm external to this structure has not generally been considered a part of the attraction sphere ; neverthelesss, such external radiations are obviously similar to those directly related to the centrosomes in the cells of the first spermatogenetic period, where they would certainly he regarded as a portion of the sphere. To save confusion, therefore, I shall speak here, as I have done elsewhere, of spheres which possess an archoplasm, as compound, and those which do not, as simple, and thus avoid the necessity of determining whether any particular set of radiations should or should not be regarded as constituents of the sphere.

26. The ring chromosomes, which, when fully formed, become dispersed over the nuclear periphery, like those of the first spermatogenetic period, are in like manner connected to one another by numerous filamentous strands of linin (fig. 43, l.).

The nuclear membrane eventually becomes irregular, and, giving way at various points, leaves the chromosomes to collect, by the contraction of these connecting filaments, into a long oval mass stretched across the nuclear sap between the centrosomes (fig. 44). The nuclear sap is traversed from the first by numerous fine strands, putting the chromosomes into connection with the outer cytoplasmic network, and which are in all probability part of the latter, dragged inwards from without after the disruption of the nuclear wall. The central chromatic mass is somewhat stretched, and more firmly attached to the old nuclear surface in the direction of the spheres, appearing as if slung between the centrosomes (fig. 44).

As time goes on, the chromosomes assume a more and more equatorial position ; but their linin filaments remain stretched out towards the centrosomes, and form the greater portion of an achromatic spindle, the equatorial part of which is consequently nuclear.

27. The astral radiations which surround the centrosomes become connected with the chromosomes in such a way as to clothe the achromatic spindle with a fibrous sheath, structurally equivalent to that described by Hermann (ante, § 8), while even at this early period in the formation of the spindle figure, the centrosomes are sometimes divided at the poles (figs. 45, 46, c.).

28. The exact form of the chromosomes, when they appear in the monaster of this first heterotype of the second spermatogenetic series, varies a good deal from cell to cell ; but in the majority of cases the loops are at first bent up upon themselves, in the manner represented in figs. 45, 45′. The rodlike bodies thus produced at first stand stiffly out from the surface of the spindle (fig. 45′), but after a time they flatten down in the manner represented in fig. 45. In consequence of this, the two limbs of the loop appear in profile to have the form of two Greek Ω’s place, side by side, and the outer surface of the bends being greatly thickened, the original opening of the loop is reduced between them to the merest slit (fig. 45, s.). These thickenings on the outer curves of the Q’s would appear to correspond with the thickenings on one side of the heterotype loop of Salamandra, but in Elasmobranchs they developed equally on both limbs. I was consequently interested to find that in the great heterotype division of the spermatogenesis of newts, these thickenings sometimes occur on one, sometimes on both limbs of the elongated loops. From the drawings given by Hermann, Flemming, and vom Rath, who deal with this form of chromosome in the salamander, it would appear that the loops are often intentionally represented with the plane of their openings at right angles to the surface of the spindle, that is, with one limb on and the other off the spindle. However this may be, it is certainly rarely if ever the case in either Elasmobranchs or newts, in both of which the loops lie flat, with both limbs on the spindle surface.

As the loops lengthen out towards the poles, the outward bends are gradually reduced, but they never disappear, and at the time the chromosomes divide (fig. 47) they separate from one another in such a way that the original openings of the loops are clearly seen. The separation into daughter-elements is effected by a transverse splitting of the loop across the central thickening, at right angles to the then long axis of each chromosome (fig. 47), as is usual in Heterotype metores.

After their separation, the daughter-chromosomes form superficial chromatic rings (fig. 48—51), as did those in the divisions of the first spermatogenetic period (see § 13), and the spindle-fibres in like manner become differentiated into two concentric tubes, separated from one another by the nuclear sap. This differentiation of the spindle into fibrous tubes is carried further than in the divisions of the first spermatogenetic period, the whole structure appearing to be composed of two completely closed cylinders of fibres, one (fig. 49, o. s.) stretched directly between the outer edge of the chromatic rings, and the other (i. s.) passing internally through them to the centrosomes. The unstained fluid which separates the outer from the inner of these sheaths, is that which previously filled the interspaces between the fibres of the younger spindle, and it was once the parental nuclear sap. It contains irregular chromatic granules (5. c., fig. 50) which appear to have been left as débris of the chromosome formation, and which sooner or later pass into the cytoplasm of the cell (A c., fig. 54).

29. While the above changes are in progress, the centrosomes become surrounded by the dusky zone, created by the shortening up and coalescence of the cytoplasmic fibres between them and the chromosomes (figs. 48, 49, 50 a).

30. The chromosomes of each daughter-nucleus are at this time quite separate and distinct, although lying closely side by side ; but as time goes on they begin to fuse together, so that the chromatin eventually forms two solid chromatic rings, one in each daughter-cell (figs. 51, 52, 53).

31. About this time the intra-zonal fibrils spread out, until those from either pole meet at the circumference of the cell (fig. 49), and at these somewhat angular connections there appear beaded thickenings in the threads, which (fig. 50, b. i.) stain and thus form an interesting stepping-stone between Flemming’s intermediate bodies and a true cell-plate.

32. At the same time the chromatic rings gradually lose their original connection with the outer spindle-fibres (o. s., fig. 52), which begin to bulge out and to pass round the nuclei towards the poles (fig. 53). The chromatic rings are thus cast loose in a surrounding fluid vacuole, but the inner core of fibres (fig. 52, i. s.) continues to pass through them.

The bulging out of the intra-zonal fibres round the nuclei is marked by a collapse at the point of their original distension at the equator (figs. 51, 52, 53), which brings the outer and the inner sheath together in the median plane (figs. 52, 53). Thus the central portion of the residual spindle figure presents the appearance of a sharply differentiated fusiform body between the cells (figs. 53, 54, 55). Across its middle there is a chromatic band, produced by the fusion of the intermediate bodies (figs. 53, 54, 55, 56, 57, b. i.).

While contemplating the changes I have just described, it is impossible to avoid the impression that the rupture of the outer spindle-sheath and the consequent outflowing of the enclosed fluid to form the nuclear vacuoles (figs. 54, 55, 56, n. v.), are the primary causes by which the expanded equatorial spindlefigure is made to collapse, and that it may also have a direct mechanical connection with the formation of the primary constriction between the daughter-cells.

. 33. As in the divisions of the first spermatic period, the spheres (fig. 37, a) begin now to travel over the surface of the nucleus, generally along a groove (n. g., fig. 54) like that described in § 15, towards its equatorial face.1

34. The expanded central portion of the spindle remnant now lies between the daughter-cells (which are otherwise quite separate from one another), and inserts a conical termination into both (figs. 53, 54) ; but the delicate filaments into which these terminations are prolonged, after the translocation of the spheres from the polar to the equatorial surface of the reconstructed nuclei, disappear, the last function of the remains of the outer and inner spindle-tube being to form an open connection with an equatorial chromatic band (the intermediate bodies) between the daughter-cells (b. i., figs. 56, 57, 58, 59).

35. The spheres, during their passage from the polar to the equatorial nuclear faces, pass in a more pronounced manner through a similar metamorphosis to that described in § 22, and which, when rightly understood, appears to be of the most profoundly interesting nature. When the archoplasm (a, figs. 58, 59) has reached some point halfway between the pole and the equatorial nuclear side, it begins to move away from the nucleus, while the centrosomes, travelling faster in the same direction, pass from the centre to the surface of its mass (figs. 58, 59, c.). From this point (c.) there grows out a fine protoplasmic thread (fig. 59, f.), extending to the cell periphery. The cell membrane is indented slightly where the thread approaches it (fig. 59), but the thread itself is prolonged beyond it as a fine protoplasmic process, comparable to a short flagellum (figs. 59—64,.f.). By the time this structure has been formed the archoplasms of each daughter-cell are more or less facing each other, with the tubular remains of the spindle stretched between (figs. 59 and 61).

36. When the cells come perfectly to rest, there appears on each side of the archoplasm, or in its immediate vicinity, a marked condensation of the cytoplasmic substance (fig. 62, x.), which, in the absence of the attraction sphere, might readily be taken for an enlarged representative of that body ; and as this mass is of some importance in understanding the process of conversion of the next generation of cells into the spermatozoa, I shall speak of it as the Nebenkern.

37. Before the prophasis of the next division, the remains of the spindle become no longer visible between the cells, and the rudimentary flagellum is withdrawn, but the centrosomes remain immediately beneath the membrane of the cell.

38. The last division in the second spermatogenetic period is a heterotype, like the first, but the elements are scarcely more than half the size. The number of the chromosomes is again twelve, and, like those of the earlier divisions, they become eventually grouped together as a globular mass in the centre of the nuclear sap (figs. 66, 67, and 68).

When the spindle has been formed, the ring chromosomes are not altogether on its surface, and, owing to their small size, a polar view of the monastral figure often presents the curious appearance represented in fig. 70, on account of the upper and lower edges being in focus at the same time. The chromatic loops eventually split transversely, like those of the previous division, the daughter V’s travelling to the pole in the manner represented in figs. 69, 71, and 72.

Thus the last division in the spermatogenesis of these fishes is, as I pointed out so early as in 1893,1 a perfectly normal affair, each mother-chromosome splitting into two daughterelements, so that the two cells produced contain each the same number (twelve) of residual chromosomes. Consequently, like the last division in Mammalia, it presents nothing in common with the “Reductionstheilung “described by Hertwig, and upon the assumed universality of which so much of Weismann’s latest theory of heredity is built.

39. The process of return to rest of the daughter-cells (spermatids) is in all respects essentially the same as that in the previous generation. The distension of the outer spindle-sheath (fig. 73) and the formation of intermediate bodies (b. i.) (cell plate) is perhaps more marked, the inner tube (i.s.) appearing consequently more isolated and alone. But there is the same detachment of the chromatin from the fibres of the outer sheath (figs. 75 and 76), the same formation of a nuclear vacuole in each daughter-cell (fig. 76, n.v.), the same gradually diminishing connecting spindle filament between the nuclei (figs. 76 and 77), and lastly, the same formation of nuclear grooves (fig. 77, n.g.) along which the spheres travel to the equatorial side. Further, when the spheres have reached some point halfway between the polar and equatorial nuclear faces, the archoplasm leaves the nuclear wall. The centrosomes (fig. 79, c.) pass on to the outer archoplasmic surface, and from this there passes a fine protoplasmic strand (fig. 80, f.) to the cell periphery, and the cell membrane is indented where this strand perforates it as the whiplash-like spermatozoon tail. It thus becomes evident that the metamorphoses described in § § 22, 35 are nothing more nor less than abortive attempts at tail-formation, and it consequently follows that the synaptic phase in these fishes marks the assumption by the cells, during spermatogenesis, of a flagellate condition.

40. Besides the nucleus, there appears a dusky condensation of the cytoplasm (x.,figs. 80—83), which at first sight gives the cells the appearance of possessing more than one attraction sphere, and is obviously similar to the nebenkern of the preceding generation (of § 36). This body when it first appears is closer to the nucleus than the archoplasm (fig. 80), and in the latter there is seen at the point of origin of the flagellum a clear round vesicle (fig. 82, a. v.), which enlarges and eventually moves, with its archoplasmic surroundings and the centrosomes, into close apposition with the nucleus (fig. 83, a. v.).

41. The nuclear chromatin rises up into a shallow collar round the base of the archoplasmic vesicle, while the rest of the chromatic substance, contracting from the nuclear membrane, becomes condensed into a flask-shaped mass below the collar (figs. 84 and 85). This contraction increases rapidly while the collar, elongating, spreads into the nuclear membrane at the base of the archoplasmic vesicle, to form a small chromatic flange round the neck of a bottle-like structure (figs. 85, 86, 87), the body of which is filled with nuclear chromatin, and the neck of which is stopped with the archoplasmic vesicle (fig. 85, a.v.). Beyond the chromatic flange the nuclear membrane encases the whole, much in the same way as the basket-work used to protect an Italian wine-flask, the nuclear sap between it and the chromatin representing the glass.

42. The base of the intra-cellular part of the flagellum, with the centrosomes, now lies between the archoplasmic vesicle and the chromatic flange, but the point of attachment of the flagellum moves round the surface of the nucleus, the archoplasmic substance penetrating the nuclear membrane, and resting with the base of the flagellum on the chromatin, in a funnel-shaped mass (fig. 85, a.). The centrosomes are no longer visible, being either lost or becoming indistinguishable among the rest of the chromatic substance at the base of the tail. The translocation of the point of attachment of the flagellum continues (figs. 86, 87) until it finally comes to rest at the side of the nucleus opposite to the neck and the archoplasmic vesicle where it started (fig. 88). The “nebenkern “is implicated in this motion, and its substance is eventually mixed up with that of the true archoplasm, both structures forming a distinctly differentiated protoplasmic mass extending along the intra-cellular part of the flagellum, from its base in the nuclear chromatin to its exit through the nuclearwall (figs. 85—89, x. a.). The whole of this mass (composed of the “nebenkern” and the archoplasm, together with the intra-cellular part of the flagellum) eventually forms the long Mittelstiick of the mature spermatozoon (figs. 90, m.). The origin of the Mittelstiick in these fishes will thus be seen to coincide with what I have related respecting this structure in Mammalia, and probably with Hermann’s description of its formation in Salamandra together with what occurs in a number of Invertebrate spermatogeneses.

43. At the opposite end of the nucleus the archoplasmic vesicle (a. v., figs. 88,89, 90) becomes first flattened, and then elongates out, together with the nuclear chromatin, forming a definite cephalic point to the spermatazoon head. The nuclear jacket (figs. 88, 89, 90, n.v.), formed by the sap separating the nuclear chromatin from the nuclear wall, continues well marked even at maturity, and the swelling on the cell membrane, where the flagellum originally passed out, remains (fig. 90, mb.) as the little bead at the hinder end of the Mittelstiick.

The spermatogenesis is now practically complete, and the facts of the second spermatogenetic period which appear to be of primary comparative importance are :

  • The transformation of the cells of the first spermatogenetic period into those of the second, which I have termed the synapsis, is accomplished while the cells are in complete repose, and is marked by a peculiar evolution in the chromatin with the formation of peculiar nucleoli (which are repeatedly characteristic of the succeeding cellular generations) and by the formation of an archoplasmic constituent round the centrosomes.

  • The evolution during the prophases of the first and second divisions of the second spermatogenetic period of twelve ring chromosomes, which split transversely to form the same number, twelve, in each daughter-cell.

  • The differentiation of the spindle during the diastral figure into an outer and an inner fibrous sheath, which coalesce, forming a delicate connecting thread between the attraction spheres of both daughter-cells.

  • The existence during the synapsis of a peculiar evolution among the constituents of the attraction sphere, whereby the centrosomes are brought to its exterior surface, beneath the membrane of the cell.

  • The repetition of the process in a more pronounced manner, after the first heterotype division in the second spermatogenetic period, so that a short flagellum is protruded from the centrosomes through the membrane of the cell.

  • The origin of the long whiplash tail of the spermatozoon in a similar manner, after a corresponding metamorphosis of the sphere during the formation of the final cellular generation.

44. The whole course of the spermatogenesis may now be diagrammatically represented as follows : In Diagram I the rings under a represent a succession of resting cells in the first spermatogenetic period, while the signs of division n1 (24), n2 (24), &c., stand for the successive divisions by which they are produced. The number 24 represents the constant number of chromosomes in each.

Diagram I.

illustrating the course of Elasmobranch spermatogenesis, (α) First spermatogenetic period. (β) Synapsis, (γ) Second spermatogenetic period. n1 (24), &c., number of chromosomes in each division of first period, where n represents an indefinite number of previous divisions. 1 (12), &c., same in second.

Diagram I.

illustrating the course of Elasmobranch spermatogenesis, (α) First spermatogenetic period. (β) Synapsis, (γ) Second spermatogenetic period. n1 (24), &c., number of chromosomes in each division of first period, where n represents an indefinite number of previous divisions. 1 (12), &c., same in second.

The cone under β represents the synaptic change, while under γ are represented (by black dots) the cellular generations of the second spermatogenetic period, up to the formation of the final spermatozoa.

45. As the majority of the operations performed by living protoplasm are inexplicable on any structural arrangement in the parts of cells which has hitherto been observed, it is obvious that the structural relationships which, if known, would render the actions of protoplasm self-explanatory, lie somewhere below the present range of vision, and it consequently follows that the theoretical explanations of this or that property which protoplasm exhibits are, at bottom, nothing more nor less than hypothetical forecasts of the ultimate structure on which this or that manifestation of vitality depends. The probability of any forecast being true is proportionate to the capacity which its premises exhibit of being logically worked up into harmony with what has been actually observed. Thus, according to Weismann,1 the phenomena of heredity depend ultimately on the existence of innumerable little unities in the “germplasm,” or “ids,” and these are in reality the hypothetical doers of everything that is done. They are capable of influencing the protoplasm which surrounds them in different ways, and by coming into action successively during development, they produce the structural differentiation of a complex form. Representatives of all the different kinds of “ids,” actual or potential, which exist in any given animal or plant, are continually being locked up for future use in every ovum or spermatozoon formed, and in consequence of the indefinite multiplication of the “ids,” which must occur after every act of fertilisation, it appears, according to Weismann, a logical necessity from the premises of his theory, that the reproductive cells, before fertilisation, must each get rid of half their hereditary substance (i. e. “must each get rid of half their nuclear rods “). This is supposed to be accomplished by there being two kinds of division among cells. In the first of these (the ordinary somatic division) the chromosomes split in half, there being consequently the same number in each daughter-cell, and this method of division has consequently been termed “Equationstheilung,” to distinguish it from the second or “Reductionstheilung,” which is apparently introduced only during the final stages of the development of the reproductive elements, and is brought about by half of the entire number of chromosomes formed during a mitosis passing unsplit into one daughter-cell, and half into the other.

The value of these hypothetical speculations touching the nature of the phenomena immediately antecedent to fertilisation, appeared to be enormously enhanced by O. Hertwig’s description of a process answering to the Reductionstheilung in the final stage of the spermatogenesis of Ascaris, because if this process should turn out to be universal, as at one time seemed probable, it would give to the “id “theory an actual demonstration in fact. Unfortunately, however, for the Reductionstheilung, as well as for the enormous superstructure which Weismann has lately piled upon it, O. Hertwig’s observations have been shown by Brauer to be quite erroneous, there being in the spermatogenesis of this animal no such thing as a division in which alternate chromosomes pass unsplit to daughter-cells. So also, during the Elasmobranch spermatogenesis with which we have been dealing (the course of which will be found summarised in these pages at the end of each spermatogenetic period), there is nothing comparable with the “Reductionstheilung “of Hertwig, which is made such an integral part of Weismann’s last theory of heredity. It is true that there is a numerical halving of the chromosomes, between the first and second spermatogenetic periods, but this is brought about in the synapsis which separates the one period from the other, and has nothing to do with division at all.

It is so necessary to be quite clear about this, that I have subpended a few lines of Weismann’s treatise in which his conceptions of the “Reductionstheilung” are given in full. On page 11 of the English translation of the ‘Germplasm,’ its author, after) speaking of the necessity of a “Reductionstheilung,” and as though the universality of its occurrence was an established fact, goes on to say :—” The hypothesis of the Reductionstheilung has been thoroughly substantiated by subsequent observations—in fact, it has even been proved that in many cases this reduction occurs exactly as I had foretold and represented in a diagrammatic figure ; that is to say, by the non-occurrence of the longitudinal division of the chromosomes, which occurs in ordinary nuclear division, and by the distribution of these in the daughter-nuclei. This holds good for the ovum as well as for the sperm-cell in animals, and as far as is known, in plants also. The germ-cell must in all cases by division get rid of half its nuclear rods.” Again on page 236:—”We now know that this reduction in the number of the ids, by one half, is of general occurrence, and is effected by means of the nuclear divisions which accompany cell division. The divisions which result in the formation of the polar bodies perform the function of the Reductionstheilung as regards the ovum, and the final divisions of the sperm mothercells have this function in the case of the spermatozoa. In both cases the Reductionstheilung does not consist in the idants (chromosomes) becoming split longitudinally, and in their resulting halves being distributed equally amongst the two daughter-nuclei, as in ordinary nuclear division, but in one half of the entire number of rods passing into one daughternucleus, and the other half into the other.”

46. The absence in the spermatogenesis of Elasmobranchs of any Reductionstheilung is thus of peculiar interest, because the fundamental way in which Weismann has used this conception of a Reductionstheilung as a basis on which to build up his supposed explanation of heredity, renders it evident that any widespread collapse in the alleged universal existence of this process, either among animals or plants, will in the long run bring down the whole speculative superstructure with it.

47. Now with respect to plants, the two highest living authorities, Strasburger1 and Guignard,2 have already dissented from Weismann’s views; and the former, in his address to the Biological Section of the British Association meeting at Oxford last year, expressed his opinion of the Reductionstheilung as follows :—” There is no such thing among plants as nuclear division resulting in the reduction of one-half of the chromosomes. Such a conception involves the assumption that the entire, not longitudinally-split, chromosomes of the mothernucleus become separated into two groups, each of which goes to form a daughter-nucleus.” So we may take it that the Weismannistic conception of the “Reductionstheilung,” “so far as is known in plants,” fails.

48. With respect to animals, Boveri,3 in his “Befructung,” as far back as 1890, after postulating the chromatin as the hereditary substance, argues, like Weismann, for the necessity of some sort of chromatic reduction, before the maturation of sexual cells ; but he comes also to the conclusion that the reduction processes hitherto described are numerical reductions of the chromosomes, and not quantitative with respect to their substance. He draws a sharp distinction between “id “reduction and chromosome reduction, the latter of which he apparently disregards, but he seems to entertain the idea that the former, by the numerical reduction of the chromosomes, may in reality be carried out. He shows further that the numerical reduction of the chromosomes in ovogeneses, like that of Echinus and Pterotrachea for example, represented in the diagram which I have borrowed (Diagram II), and which will become intelligible on reference to my § 44, is not brought about by any of the divisions in the first ovogenetic period, a, and up to the formation of the first ovocyte after the rest β, i. e. the ovum before the extrusion of the polar bodies, but that there are only half as many chromosomes in the first ovocyte when it emerges from the rest, β, in the polar bodyspindle as there were before, and this number is retained after the polar bodies are extruded in the ovum. Therefore the numerical reduction is not brought about by any division of the ovogenesis, but occurs during the synaptic rest, β, and before the prophasis of the first “Richtungspindel.”

Diagram II.

representing course of typical ovogenesis (after Boveri). Reference letters same as in I.

Diagram II.

representing course of typical ovogenesis (after Boveri). Reference letters same as in I.

Finally, he shows that the so-called reduction processes hitherto described are irreconcilable among themselves, and concludes with the following characteristic phrase:—”Durch die vorstehenden Erdrterungen, glaube ich gezeigt zu haben dass zwar gewisse Vorgange bescrieben worden sind, die vielleicht mit der Chromosomenreduction in Zusammenhang stehen, dass uns aber eine wirkliche Einsicht in diesen Vorgang bisjetzt fehlt. Es bleibt weitere Forschung vorbehalten, dieses Dunkel aufzuhellen.”

49. In 1893 I published1 a preliminary account of some investigations concerning the course of the spermatogenesis of mammals, which I summarised in these words: “There is in the Rat (i)2 a period of indifferent cell formation, terminated by a mitosis with sixteen chromosomes, both in the primary and daughter-nuclei ; (ii) a period of growth (or rest) during which the sixteen chromosomes are converted into eight, and terminated by a division in which the daughter-nuclei spermatids still retain the number eight ; (iii) a period in which the spermatids are converted into spermatozoa.” If we now construct a diagram of the first and the second spermatogenetic periods in Mammalia (as in Diagram III), and place it side by side with that of the Elasmobranchs given in § 44, it will be seen that the former differs from the latter only at the beginning and the end. These differences are produced by the shortening up of the generations of the first spermatogenetic period in mammals, (a) into what is practically a kinetic budding, so that there is only one distinct homotype division with sixteen chromosomes (I, 16) before the synapsis ; (β) (equal growing cells) in which the chromatic individuals are reduced or fused together into eight. The process of transformation and the succeeding heterotype correspond exactly with that of the Elasmobranchs. But the daughter-elements produced do not, as in Elasmobranchs, divide again. They are converted directly into spermatozoa, and it thus appears that one of the two generations of ciliated cells, present after the heterotype in Elasmobranchs, in some mammals is unrepresented.

Diagram III.

showing course of Mammalian spermatogenesis. Reference letters same as in I.

Diagram III.

showing course of Mammalian spermatogenesis. Reference letters same as in I.

50. Brauer,1 as I have said, in his admirable account of the spermatogenesis of Ascaris, published in 1891, also denies the existence of the “Reductionstheilung “described by O. Hertwig, both in the uni- and bivalent form of this curious worm. There is a period of cell multiplication, equivalent to the first spermatogenetic period, with two or four chromosomes, as the case may be, and in the divisions of which, as Professor Brauer has recently informed me, the chromosomes split longitudinally, like those in ordinary divisions. Then a period of rest, equivalent to the rest of transformation, in which the number of the chromosomes is halved, followed by divisions of a totally different character, in which there appears to be precocious splitting of the chromatic elements and rapid separation of daughter-cells,2 without the nuclei returning into rest. The closeness of the similarity of the spermatic reduction described by Brauer with those detailed above is perhaps best seen when presented in the same schematic form.

51. There are thus several well-established cases of spermatogenesis in which the reduction process described by Weismann is departed from. Besides Boveri’s account, it is apparent from Brauer’s1 figures of the ovogenesis of Branchipus, published in 1889, that the twenty-four chromosomes of the cells of the first ovogenetic series are reduced to twelve, during the synapsis, before the commencement of the second, while each of these twelve chromosomes splits twice at the beginning of the first “Richtungspindel.” The quadripartite chromosomes thus formed, divide and redivide in the two subsequent mitoses, without any intervening rest, so that there are twelve single chromosomes left finally in the ovum.

52. There are thus several well-established cases of both spermato- and ovo-genesis in which the reduction process described by Weismann is departed from, not only in the absence of the “Reductions “—as opposed to the “Equationstheilung,”—but also in the fact that the halving of the number of the chromosomes takes place in resting nuclei, one or more generations before the formation of the final sexual cells—from all of which it will have become sufficiently apparent that the Reductionstheilung of Weismann is universal neither among animals nor plants, and although an attempt may possibly be made to foist the theoretical burden which it carries on to the “synapsis “instead, there are cogent reasons for believing that the advocates of such a process will simply travel further, and in the end fare worse. It is obvious that the objections which have been raised, by botanists, against the numerical halving of the chromosomes in the resting reproductive cells of plants having anything in common with the Reductionstheilung, can be used with equal weight in the case of the synaptic phenomena in the animals I have just described. And there is yet another and much more formidable obstacle to such a view, namely the possibility, if not probability, of both the synapsis among animals and the analogous processes in plants being interpretable on common and quite different grounds.

53. It will have been seen that throughout the whole course of the evolution by which the halving of the number of the chromosomes in the above animals is produced, there exists at least a superficial similarity to that accompanying the formation of the spore mother-cells and embryo-sacs in plants; and Strasburger,1 in the address to which I have referred above, has already put forward, in a more or less provisional way, the ingenious suggestion that the halving of the number of the chromosomes in the reproductive cycles of living organisms may be interpretable on phylogenetic and not on physiological grounds. This attempt, however, to bring the whole of the phenomena into line is sadly hampered, thanks to the influence of the “Reductionstheilung” on investigation, by the insufficiency of recorded observation on the zoological side. I am enabled now, however, with these new facts relating to the reproductive cycles of Elasmobranchs, to draw Strasburger’s comparison between animals and plants much closer, and to show that the phylogenetic interpretation of the numerical halving of the chromosomes of both is probably true.

54. It will be seen on reference to § 19 of the descriptive part of this paper that the prophasis of the heterotype division following the synapsis in Elasmobranchs is preceded by a peculiar readiness of the chromatin to contract into forms like those represented in fig. 39, which is characteristic of this particular phase in the spermato- and ovogenesis in a great variety of animal forms. Now, exactly similar figures are obtained before the division of the pollen mother-cells, during the formation of the so-called “paranucleus “in plants ; but considerable diversity of opinion exists on the botanical side as to the real or artificial nature of the paranucleus and its associated contraction figure, i. e. whether the whole appearance is not in reality more a “Gerinnung’s Erscheinung “than a reality. However this may be, I do not believe that the one-sided nuclear figures seen at a corresponding period in the spermato- and ovo-genesis of animals, and with which most histologists must be quite familiar, are artifacts at all ; and whether the contraction really exists in plants or not, it has been generally conceded by the botanists I have interrogated on the subject, that it is especially related to the time in question, while Professor Farmer tells me that such shrunken nuclear figures are practically diagnostic of the synapsis in certain liverworts of Ceylon, and so there can be little doubt that there exists, at any rate at this period, a peculiarly sensitive condition of the chromatin, common to both animal and plants.

55. In Elasmobranchs, Mammals, Amphibia, and probably many other animals, the division which immediately follows the synapsis is, as will be seen from §§ 23—37, different from all those preceding it. The chromosomes as they emerge from the reticulum of rest, being no longer longitudinally-split rods, but closed loops or rings, the divisions thus fall under the category of Fleming’s heterotype. In animals the exact form and placing of these closed loops differs a good deal in different forms, but they all agree in this, that the loops split finally in the equatorial plane. In Elasmobranchs, Amphibia, and many other forms, the loops at first become bent up in the equatorial region of the spindle, so that when seen in profile they present the appearance of two Greek omegas placed side by side, the ends of which unite towards the poles (Diagram IV, 1, b, c). The outer curves and the closed ends of these figures are much thickened, and consequently the space between the two enclosed omegas is reduced to a mere slit. Viewed from above, such chromosomes present the appearance represented in Diagram IV, 1, c, and when the chromosome divides, the polar extremities lengthen out, while a transverse split appears across the equatorial thickening, and the daughter V’s, gradually separating, present a curious fourfold appearance represented in d, e.

Diagram IV.

representing division of heterotype chromosomes. (1) In Elasmobranchs. (2) In Phanerogams (according to Guignard and Strasburger). (3) In Phanerogams (after Farmer), a, b, c, d, e, corresponding stages in division.

Diagram IV.

representing division of heterotype chromosomes. (1) In Elasmobranchs. (2) In Phanerogams (according to Guignard and Strasburger). (3) In Phanerogams (after Farmer), a, b, c, d, e, corresponding stages in division.

56. In phanerogams the division which succeeds the long rest after the formation of the spore mother-cells, and which in general superficial characters corresponds to the synaptic phase (cf. § 19) among animals, differs, like its zoological counterpart, entirely in the arrangement of the chromatin from all the previous mitoses of the reproductive cycle. But the manner in which the daughter-chromosomes separate and go apart is, according to Strasburger and Guignard, quite different from what obtains in the corresponding animal cells. According to these authors, the chromosomes, after arising as stout, longitudinally-split rods (Diagram IV, 2, b) are attached at one end to the spindle surface, the two halves gradually separating in the manner represented at 2, d, e. Quite recently, however, Farmer showed1 that in the division of the pollen mother-cells of Lilium candidum this description of the origin of the daughter-chromosomes by no means fits the facts. After arising as long closed loops of microsomes, the chromosomes assume the rod form previously known ; but the apparently longitudinal splitting extends only part of the way towards the outer end. When they have become flattened out on the spindle surface—sometimes before, and always as soon as the transverse fission is apparent (Diagram IV, 3, d.)— there is seen another longitudinal split, which converts the chromosomes into a closed loop, exactly comparable to those of the animals I have just described (IV, 1), the four masses into which the equatorial thickenings are divided at the time of separation being very marked (IV, 3, d, c.).

57. Professor Farmer has kindly given me photographs of these details, which I have copied in Diagram IV, 3. It therefore appears extremely probable that the chromosomes described by Guignard and Strasburger are, in reality, like those ofLilium candidum, closed loops bent upon themselves, but that the great shortening up and thickening they undergo leads here, as it often does in animals, to the internal opening being difficult to see.

58. The outcome of all this is that the reproductive cycles of animals and plants correspond, not only in the number of the chromosomes typical of the somatic cells of any species being halved, but also in the successive and complex phases by which their numerical reduction is brought about, as well as in the type of modification which the post-synaptic cellular generations may undergo.

59. Now, with repect to the nature of these post-synaptic generations, it is made obvious by the fact that in Elasmobranchs there are two, while in mammals there appears to be only one, that their number is of no physiological importance- in the formation of the mature sexual cells. They appear rather to constitute a sort of vanishing quantity, the existence of which becomes intelligible only on the supposition that they represent a phylogenetically decreasing succession of postsynaptic generations.

The flagellum which I found in the first cellular generation after synapsis in Elasmobranchs appears to me to indicate more clearly than anything I know, that the cells, before and after the synaptic phase, are morphologically distinct. If the spermatogeneses of Mammalia were the only ones we knew, the tail developed in the generation following the synapsis might legitimately have been regarded as a purely physiological structure acquired simply to suit the exigencies of the case. But the fact that in Elasmobranchs the complex initial phases of tail formation (cf. § 35) are gone through in the first as well as the second post-synaptic generation, is to me quite unintelligible, unless these flagella represent similar structures once possessed by the representatives of the cells’ remote ancestry. This view is greatly strengthened by the complete analogy of structure which exists between the post-synaptic generations of Elasmobranchs and some of the simplest forms of sexually reproductive cells. It is well known that in many Algæ, reproduction can be carried on by means of fusion between two flagellate gametes, and quite recently Strasburger has discovered1 that the flagella of such gametes arise from the kinoplasm, a structure which there is every reason to believe is the vegetable homologue of the archoplasm. Moreover, among these organisms there exist species which exhibit every gradation between those in which both gametes are alike and flagellate, and others in which there is a true tailed spermatazoon, and a tail-less ovum.

60. It would appear thus that if the foregoing comparisons are just, the existence of cellular generations with vestigial flagella, after the synapsis and before the spermatids have been formed, indicates that the synaptic phase marks a period in the reproductive cycle at which the cells return to a flagellate condition, with only half as many chromosomes as they had before.

61. It is conceivable that this capacity of periodically altering their chromatic valency, which the cells of both animals and plants possess, and which is accompanied in some by incipient tail formation, may have arisen in either of two ways, viz. :

It is conceivable that sexual reproduction may have begun before the periodic alteration in chromatic valency was evolved at all ; but that, owing to the constant doubling of the number of the chromosomes after every act of fertilisation, a reduction in their number became physiologically necessary ; or it is conceivable that the periodic variation in chromatic valency was evolved first, and that after its introduction, sexual conjugation, with all its attendant advantages, became physiologically possible. I do not know of the existence of any evidence which is decisively in favour of one view or the other. This much, however, is certain : on the one hand, variations in the number of the chromosomes, of a capricious and indeterminate character, are known to exist to-day, as in the case of the cellular structures accessory to the reproduction of many plants ; while, on the other hand, variations which are neither capricious nor indeterminate exist, as we have seen, at certain times and places among the elements of complex animal and vegetable forms;—and I think most people will probably be disposed to agree in regarding this orderly variation as the expression of adaptive selection,which has worked towards some physiological end in the past, and the disorderly variation, if I may use the term, as the remains of something approaching a primitive chaos, on which adaptive selection has not yet acted.

I regard the above speculations, however, of relatively small importance beside the long series of structural homologies which I have established before, during, and after the synaptic phase in the reproductive cycles of both animals and plants, because this close correspondence, among a host of complex structural details, renders it improbable in the extreme that the two series of phenomena can have been independently evolved ; and whatever the synapsis may eventually turn out to be, it is evidently a cellular metamorphosis of a profoundly fundamental character, which would appear to have been acquired before the animal and vegetable ancestry went apart, and to have existed ever since.

In conclusion, I would express my sincere thanks to Prof. Howse for much help, and to the Royal College of Science for granting me the Marshall Scholarship for the completion of this investigation.

Illustrating Mr. J. E. S. Moore’s paper, “On the Structural Changes in the Reproductive Cells during the Spermatogenesis of Elasmobranchs.”

REFERENCE LETTERS

a. Archoplasm. b. c. Chromatic body. c. Centrosome, c.f. Foot-cells. ch. Chromosomes, c. s. Seminiferous cells, b. i. Flemming’s intermediate bodies. Jr. Fragmented portions of foot-cells. l. Linin filaments, tn. Mittelstiick, and portion of flagellum contained in it. n. Nucleolus, n.g. Nuclear groove, n.v. Nuclear sap-spaces, i.s. Inner spindle-sheath, o.s. Outer spindle-sheath, r. Cytoplasmic radiations round the sphere, f. Flagellum. x. Indeterminate body at the base of the flagellum.

The figures were drawn with Zeiss’ horn, immer., 2 mm., 140 ap., and oculars 12, 18, except Figs. 1, 3—10.

Cells of the first spermatogenetic period.

FIG. 1.—Single primitive cells from which the contents of the ampullæ are formed.

FIG. 2.—Single seminiferous cell, of the first spermatogenetic period.

FIG. 3.—Early relative position of seminiferous and foot-cells.

FIG. 4.— Do. do. with foot-cell in division.

FIG. 5.— Do. do. later.

FIG. 6.—Fragmentation of foot-cells.

FIG. 7.—Migration of foot-cells.

FIG. 8.—Arrangement of the contents of an ampulla, before migration of the foot-cells.

FIG. 9.—Arrangement of the contents of an ampulla, with fragmentation products.

FIG. 10.—Division of foot-cells.

FIG. 11.—Early stages of division in the first spermatogenetic period.

FIGS. 12—20.—Successively later stages of division in the first spermatogenetic period.

FIG. 21.—Polar view of monastral spindle-figure.

FIG. 22.—Details of the division of the chromosomes.

FIGS. 23—26.—Later stages of division.

FIG. 27.—Final details of the spindle, before separation of the daughtercells.

FIG. 28.—Final details of the spindle, before separation of the daughtercells.

FIG. 29.—Reconstruction of the nucleus in its surrounding vacuole.

FIG. 30.—Reconstruction of the nucleus in its surrounding vacuole.

The synapsis and the divisions of the first spermatogenetic period.

FIG. 34.—Characters of the ampullæ’ at the time of the synapsis.

FIG. 35.—Seminiferous cell in stage of transformation.

FIGS. 36—41.—Successive stages of chromosome formation.

FIG. 42.—Separation of the centrosomes, and initial spindle formation.

FIG. 43.—Superficial distribution of the chromosomes when fully formed, with further divarication of the centrosomes.

FIG. 44.—First stage of spindle formation.

FIGS. 45—52.—Successively later stages of the same.

FIG. 53.—Seminiferous daughter-cells in later stage, with remains of spindle figure between them.

FIGS. 54—56.—Successively later stages of the same.

FIG. 57.—First stage in the reconstruction of the daughter-nuclei, with residual spindle filaments attached to sphere.

FIG. 58.—Later stages of the same, showing translation of, and metamorphosis in the spheres.

FIG. 59.—The same.

Final division and structure of the spermatozoa.

FIG. 60.—Daughter-cells of first heterotype division, with vestigial flagellum.

FIG. 61.—Daughter-cells of first heterotype division, with vestigial flagellum.

FIG. 62.—Daughter-cells of first heterotype division, with vestigial flagellum, showing cystoplasmic condensation about the sphere.

FIG. 63.—Daughter-cells of first heterotype division, with vestigial flagellum, showing cytoplasmic condensation about the sphere.

FIG. 64.—Daughter-cells of first heterotype division, with vestigial flagellum, showing cytoplasmic condensation about the sphere.

FIG. 65.—Initial stage of final division.

FIG. 66.—The same later.

FIG. 67.—The same later, showing divarication of the centrosomes at the cell periphery.

FIG. 68.—Complete formation of chromosomes of the final heterotype.

FIG. 69.—Monastral spindle figure.

FIG. 70.—Monastral spindle figure, polar view.

FIGS. 71—78.—Successively later stages of final heterotype division.

FIG. 79.—Daughter-cells of the final heterotype, showing translation of the spheres, and initial stages of tail formation.

FIGS. 80—81.—Successively later stages of the same.

FIG. 82.—Cells showing the origin of the archoplasmic vesicle, and peculiar form of nucleolus.

FIG. 83.—Cells showing the origin of the archoplasmic vesicle, and peculiar form of nucleolus ; showing elongation of the intra-cellular part of the flagellum, and the attachment of the archoplasmic vesicle to the nuclear wall.

FIG. 84.—Cells showing the origin of the archoplasmic vesicle and peculiar form of nucleolus ; showing elongation of the intra-cellular part of the flagellum, and the attachment of the archoplasmic vesicle to the nuclear wall.

FIG. 85.—Cells showing the origin of the archoplasmic vesicle and peculiar form of nucleolus ; showing elongation of the intra-cellular part of the flagellum, and the attachment of the archoplasmic vesicle to the nuclear wall; showing position of the archoplasm, and re-formation of ring chromosomes.

FIGS. 86—87.—Two views of the tail during its passage away from the archoplasmic vesicle.

FIGS. 88—90.—Successive stages in the elongation of the spermatozoids.

FIG. 91.—Optical section of spindle in first heterotype division, showing outer and inner spindle tubes, o. s., i. s.

With Plates 13—16, and figs, in text.

1

Dr. Heidenhain, in his paper published in the ‘Arch, fiir Mikr. Anat.,’ Bd. xliii, p. 496, 1894, is anxious to claim priority over me in the discovery of radii extending from the sphere in leucocytes to the periphery of the cell. The words I used in a former paper were these : “a delicate radiation spreads from the whole sphere to the periphery of the cell” (‘Quart. Journ. Mikr. Anat. Sci.,’ vol. xxxiv, p. 188) ; and my meaning would have been equally well expressed had I used the word “towards “instead of “to “the periphery. I have therefore no claim at all in the matter, and it seems to me to have about as much importance as the conclusions which Dr. Heidenhain has drawn from it.

2

“On the Morphological Value of the Attraction-sphere,” ‘Science Progress,’ vol. ii, No. 10.

1

“Zur Keuntuiss der Spermatogenèse von Ascaris megalocephala,” ‘Arch, fiir mikr. Anat.,’ Bd. xlii, pp. 153—208, figs. 3—5, Taf. xi.

2

‘Arch. für. mikr. Anat.,’ Bd. xxxvii.

1

“Neue Beiträge zur Kenutuiss der Zelle,” ii Theil., ‘Arch, für mikr. Auat.,’ Bd.xxis, p. 389.

1

See Ishikawa, “Studies of Reproductive Elements; Noctiluca,” ‘Journ. Sci. Col. Imp. Univers. Tokio,’ vol. vi, p. 332.

1

Flemming, loo. cit.

1

See Maupas, “Le Rajeunissement Karyogamique chez les Ciliés,” ‘Arch, de Zool.,’ exp. tm. vii, 1889 (pl. ix, figs. 14—20).

1

Brauer, loc. cit. (pl. ix, figs. 12—18).

2

See Toyama, “On the Spermatogenesis of the Silkworm,” ‘Bull. Agrio. Coll. Imp. University Tokio,’ vol. ii, No. 3, 1894, pl. iv, figs. 25, 26.

1

Loc. cit.

2

Gr. , to fuse together.

1

‘Arch, für mikr. Anat.,’Bd. xliii, p. 423; cf. M. Haidenhain, op. cit., Taf. xxv (figs. 14, 21, 22, 23).

1

“On the Germinal Blastema of Cartilaginous Fishes,” ‘Anat. Anz.,’ Bd. ix, p. 547.

1

‘The Germplasm,’ English trans.

1

‘Ans. Bot.,’ vol. viii, 1894.

2

‘Anns. d. Sc. Nat.,’ Bot., 1891.

3

“Ergebnisse der Anat. und Entwicklungsgescliichte,” Bd. i, 1891, pp. 458, 459.

1

“Mammalian Spermatogenesis,” ‘Anat. Anz.,’ Bd. viii, 1893, pp. 683— 688.

2

Dr. Toyama, who apparently writes under the wing of Professor Ischikawa, speaks of my results respecting these phenomena in mammals as being “very improbable,” but as he produces no evidence relevant to the subject, I fail to see the force of such a criticism, and have therefore refrained from applying it to several portions of his own treatise, more especially as very little trouble with any native mammal would have enabled him to see whether the “improbable “was true.

1

Loc. cit.

2

It is probable that all the cases of the so-called “Reductionstheilung “are in reality referable to a process of precocious splitting among the chromosomes, whereby the elements for several daughter-cells are produced at once, and are then distributed, either by successive divisions without rest, or by multipolar spindle formation. An admirable example of the latter method is afforded by Farmer’s description of the spore formation in Pallavicinia decipiens.

In these plants the number of the chromosomes in the sporophyte generation is always eight, but as soon as the spore mother-cells are formed, the eight chromosomes have, during the previous rest stage, been numerically reduced to four. These four chromosomes now split up, first into eight, and then into sixteen, and all these residual chromosomes are distributed by a quadripolar spindle figure in groups of four, amongst four spores, and this final number of the chromosomes persists through all the succeeding divisions of the gametophyte generation. (‘Annals of Bot.,’ vol. viii, pp. 35—51, 1894.)

1

“Über das Ei von Branchipus Grubii,” * Abhandl. d. preuss. Ak. d. Wiss.,’ Berlin, 1892.

1

Loc, cit.

1

“Ueber Kerntheilung in Lilium-Antheren in Bezug auf die Centrosomen-Frage,” ‘Flora. Bot. Zeitung,’ 1895, Heft 1.

1

‘Histolg. Beitr.,’ 1892, Heft iv.