heterotypical mitoses appeared to throw a new light on its occurrence and meaning. We have pointed out that the characteristic changes accompanying the heterotypical mitosis in the reproductive tissues are absent from cancer cells undergoing what we regarded as this form of division, and that the want of correspondence extends to the stages which precede and follow it *. We have also illustrated some of the appearances simulating bivalent chromosomes, but in reality conforming to the type met with in ordinary (somatic) karyokinetic cell-division †. In what follows we shall illustrate other sources of error on the basis of a renewed analysis of the preparations from which the figures of heterotypical mitosis in our previous papers were made and by other figures not yet published. In the sexual cells of animals the heterotypical mitosis is preceded by a stage known as the "synapsis." In it the chromatic filament is split longitudinally and gathered into a rosette at one part of the nucleus, the nucleolus lying to one side and usually flattened against the nuclear membrane. In this stage the chromosomes are believed to unite in pairs, thus giving rise in the equatorial plate of the heterotypical mitosis to bivalent chromosomes, half as numerous as those characteristic of ordinary somatic cells. The examination of many sporadic and transplanted malignant new growths failed to reveal a corresponding sequence in their nuclei. We therefore undertook a renewed analysis of the preparations of the stages in cell-division already figured from transplanted mouse tumours, and of other preparations resembling them, to determine whether or not their identification as heterotypical were justified. We shall confine our statements mainly to five transplantable mouse tumours because they permit of control observations with varying methods of preservation and staining in a manner not possible with material from sporadic new growths; but our remarks apply also to the figures we have published from sporadic tumours of the trout and cat. Von Hansemann has combated the statements on the presence of heterotypical mitoses in malignant growths, and ascribes the appearances figured to clumping of the chromosomes, and to pathological abnormalities in their form. He adheres to his conclusion that the numerical diminution is not an exact halving of the normal number, but is Loc. cit., and Lancet,' April 1, 1905. † Loc. cit. Biolog. Centralb.' vol. 24, 1905; vol. 25, 1905; Verhandl. physiol. Ges., Berlin, 1904. irregular, and due to (1) asymmetrical mitosis, and (2) casting out and degeneration of chromosomes. Häcker reproduced three of our figures in a paper in which he admitted the striking similarity to heterotypical mitosis, but suggested that adhesion of the chromosomes in pairs*, together with longitudinal splitting of the couples thus produced, might account for the phenomena encountered in cancer. The appearances in some preparations are explicable in the manner suggested by von Hansemann †, but all the forms of cell-division resembling the heterotypical mitoses of reproductive tissue cannot be accounted for in this manner. Some of the nuclear divisions previously figured have been found on reexamination to be due to an artificial grouping together of distinct chromosomes. Fig. 1, reproducing at higher magnification fig. 3 of our Royal Society Paper, and fig. 27 of the First Scientific Report of the Imperial Cancer Research Fund, affords an example of this source of confusion. The chromosomes seem to be bivalent, i. e., to have the form of rings and loops. This mitosis is not completely depicted in fig. 1, the remainder of the chromosomes being in the next section. The preparation has been carefully restained. The result is shown in figs. 2 & 3. The "rings" and "loops" resolve themselves into a larger number of ordinary short chromosomes, split longitudinally. The dense equatorial plate of the next section shows clearly the presence * Häcker uses the term "heterotypical" in a purely morphological sense, and embraces in it the mitoses of the cells destined to give rise to the sexual products of Ascaris and the Copepods. In the latter he has shown that exposure to ether ('Anat. Anzeiger,' 1900) may cause all the cells of the developing egg to exhibit this modification (viz., cohesion of the chromosomes in pairs). In the strict sense of the word, the term "heterotype" is applied to the form of mitosis characteristic of the first ripening division of the spermatocytes of amphibia, as first described by Flemming. In the Salamander, ring-shaped chromosomes, half as numerous as the longitudinally split chromosomes of somatic mitoses, are stretched out into elongated ellipses upon the spindle, giving rise to a barrel-shaped figure. Chromosomes of similar form are associated with their numerical reduction in many animals and plants, but it must be borne in mind that different forms of chromosomes occur in the corresponding mitoses of some animals. In Ascaris, c. g., both ripening divisions appear to be effected by a longitudinal splitting of chromosomes arranged transversely on the spindle, and in others ring or loop chromosomes are never formed. It was, of course, conceivable that reduction-divisions might occur in cancer by means of chromosomes unlike those in the reproductive tissues of the same animal. The frequency of cells with diminished numbers of chromosomes led us to examine many tumours for evidence of their occurrence, but without result. + 'Biolog. Centralb.,' vol. 24, 1904. of many short chromosomes split longitudinally, and arranged in the manner described below. The mitosis is therefore somatic, and not heterotypical. The ordinary scheme of karyokinetic cell-division presents, in its phase of equilibrium (amphiaster or equatorial plate), a series of V-shaped loops with limbs of equal length, arranged around a central spindle and all lying in a plane at right angles to its axis. This arrangement is by no means universal. Some of the deviations are of great importance to a proper understanding of what occurs in cancer. Frequently the limbs of the V-shaped loops are of unequal length. When this is the case the attraction fibres are relatively few in number and attached only to the apex of the V and its immediate vicinity, and therefore nearer one end of each chromosome than the other. consequence the longer limb does not come to a position of equilibrium in the equatorial plane, but may take up one inclined to the axis of the achromatic figure. When the attraction fibres are attached to the chromosomes in this manner in a cell of elongated form, the longer limbs may even come to lie parallel to the spindle axis, and nuclear divisions closely resembling heterotypical mitoses may result. In such mitoses pairs of distinct chromosomes whose longer limbs lie on opposite sides of the equator, while their apices are closely opposed, simulate bivalent chromosomes, cf. figs. 4 to 10. In polar view the apparent halving of the chromosome number due to superposition, and the unusual vertical extension of the free longer limbs, are even more deceptive; unless the longitudinal splitting of the chromosomes is very clear, the resemblance to heterotypical mitoses may be almost perfect. During the separation of the daughter chromosomes such nuclear divisions are especially deceptive, because the longer limbs adhere for some time after separation of the apices and short limbs. Barrelshaped forms result, in which the crowding together of the chromosomes renders their enumeration impossible, but at the same time conveys the impression of a diminution in their number. A much more serious source of error results from individual differences in the size of the chromosomes in one and the same nucleus. Montgomery and Sutton + have drawn attention to this phenomenon * * T. H. Montgomery, Proc. Acad. Nat. Sc., Philadelphia,' 1901: 'Biol. Bull.,' vol. 4, 1903. + W. S. Sutton, 'Biol. Bull.,' vol. 4, 1902. in the sexual cells of invertebrates. We have found it to be present also in many vertebrate mitoses. When the chromosomes attain the position of equilibrium in the equatorial plate the smaller take up a position nearer the axis of the central spindle than the larger or more massive ones. Seen in profile the larger may then completely screen. the smaller from view, and lead to an under-estimate of the chromosome number. Normally, nuclear division takes place by means of bipolar mitosis distributing the halves of the chromosomes to two daughter nuclei. An interesting abnormality results, when, from any cause, this segregation of the daughter elements does not take place. This may result from the centrosome remaining single, or, when after division of the centrosome, one only becomes attached to the chromosomes by the attraction fibres. The chromosomes then remain in one group, but the daughter elements separate slightly from each other before combining to form one large nucleus, containing twice as many chromosomes as the mother nucleus. This form of mitosis, known as a "monaster" (from the presence of only one active attraction sphere), has been studied and described in detail by Th.* and M.† Boveri in invertebrates: figs. 13, 14, and 15, from a squamous cell carcinoma of the tongue (human), represent a cell in this condition. The nuclear membrane has disappeared, but there is no trace of radiations or centrosomes to be seen. The chromosomes are arranged as a hollow sphere around a central clear area, and for the most part consist of two parallel daughter elements in different stages of separation. Some chromosomes in which the separation of the daughter elements is least advanced have a horseshoe shape. In others, the daughter elements are parallel and widely separated. In a few, the separation is incomplete, the ends only remain apposed, giving the appearance of rings. In fig. 14 this cell has a striking resemblance to a heterotypical mitosis such as occurs in testis. Its cytoplasm is clear and voluminous, and there are no intercellular bridges between it and the surrounding epithelial cells. The number of chromosomes in the cell, however, is not diminished, but amounts to 40 to 50 when both sections are examined. Such a mitosis, instead of reducing the number to half, really results in the original number of * Th. Boveri, 'Sitz.-ber. Phys.-med. Ges.,' Würzburg, 1897; 'Zellenstudien,' vol. 4, Jena, 1901. + M. Boveri, 'Jen. Zeitschr. f. Naturwiss., vol. 37, 1903. F chromosomes being doubled, because the separated halves of the chromosomes co ubine again to form a single nucleus. After elimination of these sources of confusion, there remain other apparent heterotypical mitoses which cannot be explained in any of these ways. The chromosomes present an irregular contour and become drawn out like a viscid fluid in the later stages of separation. Individual chromosomes cannot be made out. The achromatic spindle develops at such a rate that the evolution of the chromosomes cannot keep pace with it, and they are drawn towards the spindle and stretched upon it before they have completely contracted and condensed. All these abnormalities may occur in nuclei possessing the usual number of chromosomes. They also occur in cells with a greater or lesser number associated with the presence of multipolar and asymmetrical mitoses in other cells. In such cases there is no evidence warranting the assumption that the diminution in the number of the chromosomes is due solely to a nuclear division effecting a reduction comparable to that of the sexual cells. Galleotti and von Hansemann† have shown that nuclei with diminished numbers of chromosomes (hypo-chromatic) may arise from larger ones by asymmetrical mitosis, in which entire chromosomes pass to one daughter cell, because they are only attached to one or other attraction sphere, and also by "casting out of chromatin." "Casting out of chromatin" is merely an exaggeration of what occurs in asymmetrical mitosis; in it some chromosomes remain unattached to either attraction-sphere, and therefore fail to be included in either daughter nucleus. Krompecher ‡ and we ourselves § have shown that multipolar mitoses may also lead to a diminution in the number of chromosomes. We stated that nuclei with diminished and half the somatic number of chromosomes occur without it being possible to determine whether the diminution has been effected by asymmetrical mitosis, casting out of chromatin, multipolar or heterotypical mitosis. We have given our reasons for now believing that the mitoses we formerly assumed confirmed the occurrence of a heterotypical reducing division in cancer, are, in reality, somatic mitoses. Although we do not presume to explain in the above manner all the figures which may be brought forward resembling that form of nuclear division, we submit that the occurrence of heterotypical mitoses in cancer requires further * Ziegler's Beitr.,' vol. 14, 1893. Centralb. f. Path u. Anat., vol. 13, 1902. † Loc. cit. $ Loc. cit. |