On Multipolar Mitosis as a Means of Analysis of the Cell Nucleus

Über mehrpolige Mitosen als Mittel zur Analzyse des Zellkerns. Verhandlungen der physicalisch-medizinischen Gesselschaft zu Würzburg. Neu Folge 35: 67-90.

Theodor Boveri, 1902

Translated by Salome Gluecksohn-Waelsch for Foundations of Experimental Embryology (B. H. Willer and J. Oppenheimer, eds.), Prentiss-Hall, Inc., Englewood Cliffs, NJ, 1964. This paper is adapted here with permission of Drs. S. Gluecksohn-Waelsch and J. Oppenheimer. Boveri's footnotes are brought into the text and appear in parentheses.

It has been known since the investigations of Fol and 0. Hertwig that the penetration of two spermatozoa into the sea urchin egg results in the formation of a tetrapolar spindle and consequently in the simultaneous division of the egg into four blastomeres.{Cf. Boveri (8) concerning exceptions to this rule}. Driesch (13) isolated 82 such quadripolar eggs and found that they were unable to develop beyond the stage of an abnormal blastula (the so-called stereoblastula); at the very most, the first beginning of invagination could be observed. "Even an approximately typical gastrula was never formed." I had this same experience three years ago in unpublished investigations in which approximately ten dispermic eggs of Echinus were cultured separately. Later I was able to achieve an analogous effect in normally fertilized eggs as a result of suppression of the first cleavage division, a procedure which, as in dispermy, results in the formation of four centrosomes instead of two in the cell. Eggs of this sort cultured in isolation did not develop beyond stereoblastulae.

Several possibilities could conceivably account for this result, and it was possible to subject at least a few of these to experimental tests. I therefore decided to try to ascertain if it were possible to solve the question by means of a most careful analysis of doubly fertilized eggs.

1. Since O. and R. Hertwig (21) had found that damage to the eggs facilitated the penetration of several spermatozoa, it was necessary first of all to consider the possibility that the pathological development of dispermic eggs was caused not by the penetration of two spermatozoa, but by a pre-existing pathological condition of the eggs. The following experiment designed to test this possibility is based on an observation of mine, reported previously (7, p. 439), which indicated that the percentage of doubly fertilized eggs depends largely on the amount of sperm present. Undamaged eggs of a female were divided into two groups, one of which was exposed to very little, the other to very much sperm; examination of these eggs after the appearance of the first cleavage furrow showed that the first group had very few and the second group very many dispermic eggs. The percentage of abnormal larvae in both groups corresponded to that in dispermy.{Numerical evidence for this and other statements will be given elsewhere}. It was demonstrated therefore that the pathological development of the eggs is a result of the dispermy.

2. Since, as one of the certain consequences of dispermy with resulting quadripolar eggs, each of the four blastomeres contains as a rule a different number as well as a different combination of chromosomes, the next question to be asked was whether this differential distribution of the chromatin might have an effect on the properties of the four cells. The discovery of Herbst (20), who showed that calcium-free sea water separated the individual blastomeres of the sea urchin egg from each other, makes it possible to isolate each single blastomere and to follow its fate. In this way Driesch (17) discovered that quarter blastomeres of normally fertilized eggs are able to develop into normally formed though dwarfed plutei. Whether each one of the four blastomeres is capable of such development Was not determined by Driesch. I therefore repeated the experiment in such a way that I separated isolated eggs after the appearance of the second cleavage furrow into their four blastomeres, culturing separately each of these four cells of common origin. As was to be expected, each cell gave rise to a pluteus. The result is completely different if the four blastomeres of a dispermic quadripolar egg are separated from each other. First of all, such blastomeres will not-except for rare cases-develop into plutei. Secondly, however, many of them develop into at least more or less normal gastrulae in such a high percentage that on the average almost one quarter gastrula is found for every two dispermic eggs; therefore, if we take into account the fate, previously mentioned, of whole dispermic eggs, certain quarters achieve more separately than do all four quarters together. Thirdly and finally-and this is the most significant result of the experiment-as a rule, each of the four blastomeres develops differently. Since development in most of them does not proceed very far, these differences do not ordinarily amount to very much; however, there are also striking cases in which beside one quarter that broke up into separate cells at the blastula stage, or that became a permanent stereoblastula, a more or less normal gastrula is found, or even a young pluteus with segmentation of the gut and skeletal primordia. Whereas, therefore, the four blastomeres of a normally divided egg are equivalent, the properties of the blastomeres of a dispermic egg differ from each other in many respects and to varying degrees.

3. After this result it was to be expected that different potencies of the four quarters should frequently be demonstrable in the development of whole dispermic eggs also. This indeed turned out to be the case. When the eggs have developed into swollen blastulae with polar differentiation, a stage at which, as a rule, they still appear perfectly normal, one or two of the quadrants of the blastula located between two meridians now begin in many cases to slough cells into the interior. Consequently, this entire marginal portion appears opaque, or the whole quarter dissolves, shedding its cells to the outside. Finally the remaining part, which is at first still open, closes up again into a vesicle. But there are further events, and here my results differ from the experiences of Driesch mentioned above and from my own earlier ones. Some of the quadrants of the embryo develop into gastrulae and form a skeleton, but now usually in a way that expresses the different potencies of the individual regions, e.g., with an archenteron asymmetrical in degree of differentiation or in location, or with the skeleton present on one side only and even there more or less abnormal.

4. However, even plutei varying from grossly abnormal to completely normal in structure may develop from dispermic eggs, and here an experimentally obtained variety of dispermic development is particularly illuminating. In the experiments described above in which the development of isolated blastomeres of dispermic eggs was studied, it was necessary to remove the fertilization membrane, which can be accomplished, according to Driesch, by shaking the eggs a few minutes after fertilization. With this procedure, I observed almost regularly the phenomenon which Morgan (24) already described but whose origin has not been explained, namely, that some eggs divided simultaneously into three blastomeres. There are two ways to demonstrate that the three blastomeres are derived from dispermic eggs in which, as a result of shaking, one of the sperm centrosomes did not divide; this one gave rise to one pole, whereas the other two poles were the results of the normal division of the second sperm centrosome. {In respect to this and other eveidence I refer for the time being to the more detailed description elsewhere}. Morgan studied the development of ten such tripolar eggs and three of these reached the stage of fully formed gastrulae. I myself have cultured more than 900 isolated specimens some of which were whole, some of which were separated into their three blastomeres; in principle I made the same observations as on quadripolar eggs, noting, in particular, the same lack of equivalence of the regions originating from multipolar cleavage. However, the tendency for normal development is much stronger in the tripolar than in the quadripolar eggs and consequently a quite considerable percentage of plutei, some of them completely normal, are obtained from them. Even the quadripolar dispermic eggs give rise to plutei in a very small percentage of cases; however, I never found a completely normal one among these.

5. To explain the fact just reported, we had best start with the questions of possible origin of the differences in developmental potencies of the cells resulting from multipolar cleavage. The differential cannot lie in the cytoplasm. The reason for this is that the tetraster of a dispermic egg-only these were used for the blastomere separation experiments-and similarly the triaster are located in a plane (karyokinetic plane) perpendicular to the axis of the egg {Cf. my evidence in 9 and 10}. This can be directly observed in Strongylocentrotus by the relation to the pigment ring and can be indirectly concluded from a consideration of the two abnormal cleavage types studied by Driesch (13) and Morgan (24) in connection with the axial relations demonstrated by myself. The four blastomeres of the quadripolar egg are just as equipotential in their protoplasm as those of the normal four blastomere stage. Similarly, a differential in the centrosomes cannot be assumed. The reason for this is that each two of the four centrosomes of dispermic eggs correspond to one of the two of the normally fertilized egg which have identical properties which can be concluded from the study of normal development and all pertinent experiments. At most in the tripolar eggs the possibility might be considered that the one undivided centrosome differed qualitatively from the two others However, quite aside from the completely identical behavior of the three cells in the subsequent stages of division, it is specifically the tripolar eggs that may give rise to completely normal larvae. But even different potencies of the centrosomes could not cause what is demonstrated to us by the development of the dispermic eggs, namely, an almost unlimited variability from complete normality to abnormalities of the highest degree, and specifically the so extremely variable and in each particular case so differently combined potencies of the blastomeres derived from simultaneous multiple division. These phenomena could only be explained on the basis of a process which itself is subject to corresponding variability and such a process is presented only in the manner of distribution of the chromosomes.

After I realized in 1887 (1) that the karyokinetic figure results from a secondary connecting together of two cell organelles previously independent of each other, on the one hand the centrosomes with their spheres, on the other hand the chromosomes, I was able to demonstrate in 1888 (3, pp. 180ff), as a result of the first detailed analysis of multipolar division figures, that the distribution of the chromosomes between more than two poles is determined by chance. "Karyokinesis, which in the presence of two poles is a mechanism of almost ideal perfection for the purpose of dividing a nucleus into two daughter nuclei identical in quantity and quality, turns these advantages practically into the reverse as soon as a larger number of centrosomes begins to take effect.... Number, size, and-if we have to assign different qualities to the individual chromatic elements-also the quality of the resulting daughter nuclei are determined by chance" (3, p. 185). If we consider our particular case, the number of chromosomes of the mature Strongylocentrotus egg is approximately 18, and the identical number is found in the spermatozoon. The first cleavage spindle therefore contains 36 elements each of which divides into half, so that each daughter cell similarly contains 36. The number of chromosomes in the doubly fertilized eggs amounts to 3 X 18 = 54. As a result of division of each chromosome into two halves, 108 daughter chromosomes are produced, which are distributed (in the typical case) into four cells. In the case of equal distribution, each of these four cells would contain 27 elements, that is nine less than normal. Actually, such an equal distribution occurs probably only in extremely rare exceptions; the four cells therefore, obtain on the average not only fewer, but also different numbers of chromosomes, and particularly quite different combinations of them. If we designate the individual chromosomes of the dispermic first cleavage nucleus A, B, C, D, etc., then we see that in the case of multipolar cleavage of the egg into four blastomeres only two blastomeres can have a representative of A, or of B, etc., whereas no representative of this particular chromosome gets into the other two blastomeres. {A figure illustrating this is included in this file}.

6. Now the question arises: Is the different potency of each individual blastomere of the quadripolar egg based on unequal quantitative distribution or do we have to ascribe different qualities to the individual chromosomes in order to explain this heterogeneity? That a particular number of chromosomes in itself is not required for normal development was demonstrated by myself in experiments, confirmed by Delage (11) and Winkler (32), in which the development of enucleated egg fragments fertilized by one sperm was studied and in which normal plutei developed although they had only half of the normal amount of chromatin and number of elements, that is, only the chromosomes of one sperm nucleus. What was demonstrated here for the sperm nucleus has since been shown to be true also for the egg nucleus as a result of the investigations on artificial parthenogenesis by J. Loeb (23) and E. B. Wilson (30).

{It is irrelevant for this argument whether the normal number of chromosomes is restored in later embryonic stages, a fact which, incidentally, I still doubt on the basis of my investigations. Delage, however, recently extended (12) to artificial parthenogenesis his contention, proposed originally for merogony, that the normal number of chromosomes could be found in later embryonic stages. He was able to count with certainty and in numerous cases, 16 to 19, on the average 18, chromosomes in the cells of parthenogenetic embryos of Strongylocentrotus, and thus he considers his earlier statements completely proven. However, it escaped him here that the normal chromosomes number of Strongylocentrotus is not as he assumes 18 on the average, but 36, as I have found without exception in three different years (1888, 1896, and 1902); the chromosome number of the individual pronucleus is therefore the average of 18, a figure which R. Hertwig (22) actually determined (16 to 18) for the egg nucleus in preparation for independent division. According to the hypothesis of the individuality of the chromosomes, we therefore expect the average number of 18 as found by Delage in the parthenogenetic as well as in the merogonic egg of Strongylocentrotus; thus Delage's new accounts prove exactly what he believes to disprove, namely the failure of the regulation of chromosome number}.

We could now also put up the following argument: The number of chromosomes, above a certain minimal limit, does not matter as long as the same number is present in each cell. If the individual regions of the same embryo contain nuclei with different numbers of chromosomes, then abnormalities occur. However, this assumption can also be refuted in two ways. First of all, on the basis of the experiments of blastomere dissociation. Each isolated blastomere of the quadripolar egg has on the average more than the necessary minimal number of chromosomes, and even in the most unfavorable distribution, at least two of the four blastomeres have to obtain more than the minimal number. According to our assumption therefore, each quadripolar egg dissociated into its four blastomeres should yield at least two plutei, which is however not the case.

However, the untenability of this hypothesis may be demonstrated in the whole dispermic egg. Also, I had determined previously in my experiments on merogony (5, 7) that larvae from enucleated egg fragments had considerably smaller nuclei than those from nucleated fragments or from whole eggs. This observation I have found to be confirmed in the clearest possible way in a repetition of these experiments just completed, as I shall report in detail in a separate paper. I only would like to mention here the following observation: If one selects from the fragmented eggs of a female on the one hand nucleated, and on the other hand enucleated fragments, and fertilizes these with identical sperm, the larvae developing from the latter fragments contain considerably smaller, and, as I now must add, considerably more nuclei than larvae of the same size and age developing from the former fragments.

{This sentence holds true also, even if not quite as strictly, for larvae developing from fragments of different sizes. It is true, furthermore, not only for cases with decreased but also for those an with abnormally increased number of chromosomes. I have succeeded in obtaining cleavage of the egg with twice the normal number, that is in Strongylocentrotus, with approximately 72 instead of 36 chromosomes. The larvae contain accordingly much bigger nuclei than those developing from normal control eggs, and, in connection with this, much bigger and many fewer cells. They show only about half the normal number of mesenchymal cells, and never produce, obviously because of this small number of cells, completely normal plutei}.

Thus, the size and number of the nuclei and accordingly also the size and number of cells of a sea urchin larva are respectively-other things being equal-directly and inversely proportional to the number of chromosomes in the mother cell. I was able to determine without doubt, by raising larvae in which I knew with certainty the chromosome number of the individual blastomeres, that this rule holds not only for different larvae but also for different regions of one and the same larva, provided that these regions are derived from blastomeres with different numbers of chromosomes. I cite here merely the not infrequent case of dispermy where only one sperm nucleus unites with the egg nucleus and a normal first cleavage spindle is formed by the chromosomes of these two, whereas the other sperm nucleus comes to lie in a separate spindle which therefore contains only half as many chromosomes. Such eggs almost always divide, as I have described previously (8), into two cells, each with one large and one small nucleus. However, it occurs, sometimes in the beginning, more frequently in one of the later cleavage divisions, that regions with small nuclei separate cleanly from those with large nuclei and consequently, larvae develop whose properties and significance I shall discuss below. {Dr. E. Teichmann will report more details from his own investigations about the variations which occur here}. Here it may suffice to state that these larvae consist of one part with large nuclei and another part with small nuclei and correspondingly more numerous cells, quite in the same relation that we have established between larvae developed from nucleated and those from enucleated egg fragments.

Once this is established, the conclusion is unavoidable: since larvae from normally cleaved eggs possess nuclei of equal size in corresponding regions of the organism, those larvae in which regions with different nuclear sizes are found must be derived from eggs in which each of these regions starts out from a blastomere with a different chromosome number. Now, I have found, among the almost normal plutei of the tripolar eggs described above, four in which one-third of the larvae contain very small and correspondingly more numerous nuclei, and the remaining two-thirds contain large nuclei. {It should be noted that this third with small nuclei may belong to different regions of the pluteus. It always, however, represents a region which comprises all zones from the animal to the vegetative pole so that therefore a third of the gut also has correspondingly small nuclei}. In the living pluteus, the border at which two such regions meet usually cannot be recognized. Thus it is proven that the different number of chromosomes as it happens to be distributed to the various regions of the egg as a result of multipolar division is not to blame for the pathological development of dispermic eggs.

It should be noted as an important supplement to this statement that I have obtained numerous highly pathological products from tripolar eggs in which size differences between the nuclei could not be demonstrated; in these, therefore, an equal or practically equal distribution of the chromosomes into the three primary blastomeres must have occurred. In this way, the irrelevance of a definitive amount of chromatin has been proven from the opposite direction.

One other idea might perhaps come to mind, namely, that different numbers of chromosomes could occur in the individual cells, but only certain quite definite numbers, namely, those typical for the individual pronucleus and their multiples, but no intermediate numbers. If we consider that it is possible to obtain hybrids between species with different numbers of chromosomes, then such a possibility already becomes highly improbable. However, it also can be disproved more strictly. Namely, if we assume for the plutei described above with a third containing small nuclei, that these small nut lei contain the chromosome number of the individual pronucleus, then it is a necessary consequence that the two other thirds with their nuclei of almost identical size present intermediate numbers, thus refuting the hypothesis.

7. Thus, only one possibility remains, namely that not a definite number, but a definite combination of chromosomes is essential for normal development, and this means nothing else than that the individual chromosomes must possess different qualities. At the moment, we are unable to give a more definite setting to this irrefutable conclusion; only one more exact statement can be made, namely, that all qualities, at least those essential to reach the pluteus stage that can be raised in our aquaria, have to be represented in the dispermic egg at least three times. For according to the experiments on merogony and parthenogenesis the egg nucleus as well as each sperm nucleus contain all qualities.

If we now consider all chromosomes of a dispermic egg to be unselectively mixed in a uniform first cleavage nucleus, and if we suppose that each quality exists in three chromosomes only-in one element from each pronucleus-then the probability that a certain quality is transmitted into each blastomere amounts to 70% in tripolar and 40% in quadripolar eggs. If the individual qualities of each pronucleus are distributed between nine chromosomes (the actual number in Echinus microtuberculatus), then the probability that each blastomere of the tripolar eggs obtains all qualities, that is, at least nine chromosomes, representing the whole set, is calculated at 4%, and in the quadripolar egg at 0.0026%. {This calculation is based on the assumption that all three or four poles are connected by spindles into a triangle or quadrangle}.

A strictly absolute value cannot be assigned to these figures. There exist in these distributions conditions which evade all calculations. In addition, it is doubtful what still can be considered a normal pluteus, and particularly it is conceivable that an occasional larva is excluded whose abnormality may be explained in another fashion than by a defect of chromatin qualities. Finally, the number of cases available to me (695 tripolar and 1170 quadripolar eggs) is not very considerable; also, these are derived from three different species. If we take all this into consideration, then we shall find the agreement of the expected with the actual result nevertheless significant. I have found, among the 695 tripolar eggs cultured as wholes, 58 almost normal plutei, that is 8.3% compared with the expected 4%. Among the ten larvae considered to be plutei obtained from the 1170 quadripolar eggs, there was not a single one as normal as the 58 obtained from the tripolar eggs and just mentioned. According to our calculation of course, their occurrence here can hardly be expected.

8. However, not only are the chances for a favorable distribution of the chromosomes greater in the tripolar eggs in general, but there is the additional possibility of a particularly regular distribution which is lacking in the quadripolar eggs. In the case of three poles, it is possible for a complete pronucleus to get into each of the three spindles and from this constellation there could be derived a chromatin complement of the three blastomeres which corresponds almost completely to that of the normal; each cell would possess the normal number of chromosomes and a double set of all qualities, one, that of the egg nucleus and of one of the sperm nuclei, the second, that of the egg nucleus and of the second sperm nucleus, the third that of both sperm nuclei. From this condition, I would like to derive, in addition to the very interesting asymmetrical forms, the few absolutely normal plutei that I obtained from tripolar eggs and whose nuclei of consistently equal size demand the assumption of an equal distribution of chromatin.

9. If we now examine the abnormal and pathological larvae from the point of view we have reached, they may be usefully divided into three groups:

a) Highly pathological. Here I include all those larvae that had not attained the morphological properties of a normal pluteus in any respect. This group contains, by the way, great variations that will not be considered in detail here.

b) Larvae in which certain pluteus characteristics are missing or appear to have developed abnormally whereas the others are normal. A few examples of this group may be cited: A case where a vesicle-shaped, completely anenteric larva has a perfectly developed skeleton on the left side, whereas the right one is missing completely; furthermore, cases where an otherwise normal pluteus from a tripolar egg lacks completely one or two thirds of the skeleton as though they had been cut away; finally, cases also from tripolar eggs where one third of the pluteus, otherwise practically normal, has no pigment cells. In some of these cases, it was possible to conclude from the nuclear size that the borderline of the defect coincides with the borderline between two regions derived from two different primary blastomeres.

c) Larvae derived from tripolar eggs which though normal within each separate part show the same trait in a different "individual" type on each side, so that they appear more or less asymmetrical. These cases relate closely to my earlier experiments of hybridization of enucleated egg fragments, and if on the one hand they do not operate with differences as great as species differences, on the other hand they are also not open to the objections that could be raised against the hybridization experiments. It is undoubtedly true in the cases just mentioned that different nuclear substance in identical cytoplasm produces a different larval type. These larvae could be explained by the kind of particularly regular chromatin distribution mentioned under 8 where it may happen, for example, that one larval third which contains the right part of mouth and anus contains in addition to the derivatives of all the maternal chromosomes those of one sperm nucleus, whereas the corresponding parts of the other side have obtained in addition to the maternal elements those of the other sperm nucleus. In this case, the individual differences inherent in the two sperm nuclei, differences which otherwise would appear in two different larvae, develop next to each other in the symmetrical organs of one and the same larva. As a matter of fact, I was able to make drawings from the different types of normal control larvae where I combined the right half of one type of larvae with the left halfof another one; such pictures corresponded almost exactly to the plutei of tripolar eggs in question.

The experiments just reported appear to me to explain the symptoms of pathological dispermy. We are able to say: double fertilization and pathological development are not related to each other only indirectly in such a way that a pathological condition of the egg leads to pathological development on the one hand, and on the other hand makes dispermy possible, but pathological development is a consequence of dispermy, since the penetration of two spermatozoa ruins the previously perfectly normal egg. However, dispermy does not under all circumstances lead to pathological development, but only under certain conditions which however are almost always present. These conditions are not necessarily inherent in the increase of the number of centrosomes to three or four; for we have found that plutei may develop also from tripolar and quadripolar eggs. And since the cytoplasm behaves alike in all cases of multipolar division, it follows from the development of these plutei also, in agreement with the results of E. B. Wilson (31), that simultaneous division into more than two cells is not injurious to the cytoplasm. Rather, the harmful effect of multiple poles is due to the fact that as a rule they cause an abnormal chromatin complement in the daughter cells. {Already, in 1888 (3, p. 185) the analysis of multipolar mitoses led me to the conclusion, "that indeed the cell substance is prepared for a simultaneous multiple division, not, however, the nucleus." However, just now I have reason to assume that this statement does not apply to all egg cells}.

The number of cells is irrelevant in this abnormal distribution. To be sure, the chromosomes as carriers of different qualities have to be present in each cell in a certain minimal number comprising all qualities; but beyond this, their number is irrelevant up to an upper limit harmful for other reasons; and in the reverse sense, a normal chromosome number in all cells, which is possible in tripolar eggs, does not guarantee normal development.

That in spite of the distribution of the dividing chromosomes into more than two cells, all these cells still may obtain all qualities is due to the fact that each quality is represented at least in three different chromosomes in the dispermic eggs. In the case of simultaneous multiple division of a normal first cleavage nucleus each cell could also obtain all qualities; however, the occurrence of such a case would be almost infinitely improbable.

These results on dispermy are in complete agreement with those obtained in studies of cells which for other reasons contain several centrosomes. As mentioned in the beginning, the suppression of the first cleavage furrow has the effect that the two spindles that should belong to the two half-blastomeres come to lie next to each other in the undivided egg. E. B. Wilson (31) recently described cases of this kind, in which this condition led to a direct quadripolar division of the egg, and he obtained normal plutei from such eggs. In my own earlier experiments (8), such eggs divided first of all into two binucleate cells, in which again two spindles appeared and this condition was perpetuated until, sooner or later, the two spindles united into a four-polar figure; then, due to quadripolar division, uninucleate cells resulted which in turn continued to divide regularly. As mentioned above, these eggs did not develop beyond the stage of pathological blastulae. The experiences with dispermy provided the simple explanation for this different development. In Wilson's cases, each of the four simultaneously arising blastomeres obtains the very same chromosomes as in normal cleavage; in the cases observed by myself, the nuclei are also normal as long as they remain separate; however, as soon as a four-polar figure replaces the two separate spindles, the daughter nuclei must, as a rule, just as in dispermy, obtain faulty combinations of chromosomes and thus become pathological.

Closely related to these cases is that type of dispermy where one sperm nucleus remains separate, leading to the formation of two separate, usually parallel, spindles. We reported briefly above (6) about the early development of such eggs. Since the separate sperm nucleus possesses all the qualities necessary for pluteus development, the potentialities of this type of doubly fertilized eggs must be essentially identical to those of the cases just described where the first cleavage furrow is suppressed. If, as in Wilson's experiments, simultaneous division into four blastomeres were to occur, normal, even though not completely normal, development would be expected. In the dispermic eggs isolated by myself with two parallel spindles, such an immediate division into four blastomeres never occurred. However, several underwent a tripolar division into one binucleate and two uninucleate cells. I do not want to elaborate here the rather variable details; we may formulate the result on the basis of the various cases by stating that such eggs maintain the ability of normal development as long as, and to the extent that, the four centers present in the egg or the succeeding cell generations are not combined into one multipolar figure. If, before this happens, a division into uninucleate cells takes place, then the region of the embryo originating from this part has been definitely salvaged for further normal development. Thus, it can be explained that the percentage of larvae reaching a stage beyond that of the blastula is considerably greater in this constellation than where both sperm nuclei have united with the egg nucleus; and it is a beautiful confirmation of all our conclusions that the best developed larva that I obtained from a tetrapolar dispermic egg was derived from such a case with a separate sperm spindle.

In general, we may say this for the Echinus egg: multiple centrosomes in a cell are harmless for the cell complex that will eventually develop as long as only two poles each unite into one karyokinetic figure and as long as the original nucleus or nuclei were normal. If eventually a cell is formed around each of the centrosomes and around the nuclei resulting from the successive mitotic processes, then all of these will be normal as Wilson actually showed recently in his experiments on the suppression of cell division, and as we have known for some time from a similar process in some cleavage types. Multiple centrosomes have a pathological effect only if groups or more than two divide the available nuclear substance among themselves. In this case, there is no guarantee, or even the possibility, that all cells will be supplied with a portion of all the different qualities represented by the individual chromosomes.

With this we proceed from a consideration of our special case to the general significance of the results just described. A differential value of the chromosomes, as concluded frequently before from studies of the morphology of mitosis, has now been proven and thus a first step has been made towards the analysis of the physiological constitution of the cell nucleus. {I myself have maintained until now (4, p. 56), primarily on the basis of my experiments with Ascaris megalocephala, that the chromosomes are essentially equal but individually different formations, and the same opinion I find maintained by Weismann in the recently published "Lectures on the Theory of Descent." This assumption has been refuted in the sea urchin egg by my experiments; and it is clear therefore that the simple considerations that Weismann developed for the reduction division also require at least considerable modification since random distribution of the chromosomes into two groups should in general be equally harmful as a multipolar mitosis. Those and related problems, as well as the relevance of this to the results of botanists in studies of hybrids and their descendants, will be discussed separately}.The difference of our experiments on the nucleus from the previous ones lies in the fact that until now, nothing else could be done but remove the entire nucleus and examine the results of its absence {My own experiments on fertilization and particularly on hybridization of enucleated egg fragments constitute a certain exception here}. We supply the cell with a nucleus which lacks certain portions and we follow the consequence of this defect. We have found that such a nucleus is sufficient for certain processes of ontogenetic events but not for others, so that it transmits, for example, the ability of invagination to the derivatives destined to form the gut, but it does not transfer the necessary qualities to the cells destined to form skeleton, or vice versa. We have to conclude from this that only a certain combination of chromosomes, probably no less than the total of all those present in each pronucleus, represent the entire essence of the organism's structure insofar as this is determined by the nucleus.

This recognition leads to the conclusion that the most important aspects of the physiological constitution of the nucleus are completely inaccessible to an analysis with the present methods of physiological chemistry. In this respect, Biology has at its disposal means of analysis of much superior resolving power. Even if the biologist is not capable of removing individual chromosomes as would be the ideal case, he possesses, nevertheless, in multipolar mitosis a tool for the production of the most diverse combinations, and embryogeny during which the qualities of the original nucleus unfold themselves provides the analysis of those qualities which are made possible by the various combinations ("Embryonalanalyse" of the cell nucleus).

What could be demonstrated here for the nucleus of the sea urchin egg is valid, with certain modifications, for all nuclei that divide mitotically. For mitotic division itself, however, we may consider as proved what has been assumed for a long time; namely that its goal lies in the transfer of the qualities present in one nucleus into many nuclei, and that it is specifically the function of the bipolar mitotic figure to multiply successively the nucleus in its totality. These statements, I believe, will from now on be counted among the firm basic principles of general physiology.

If we now consider some details more closely, the experiments offer us the first exact indications about the role of the nucleus in ontogenesis by the certainty with which they permit us to ascribe the disturbances of development exclusively to the chromosomes. It appears that the initial steps up to the blastula stage are independent of the quality of nuclear substance, even though it is essential that the nuclear substance be of a kind capable of existing in the egg. {Cf. in this connection, my discussion in 6 (p. 469) and 8 (pp. 14ff)}. Incidentally, the statements hold of course for the time being only for Echinids}. The necessity for particular chromosomes becomes apparent first with the formation of the primary mesenchyme and from then on shows up in all processes as far as development can be observed. But not only do certain chromosomes prove to be essential in this connection; in addition, it appears that with respect to those characters in which we are able to recognize individual variations, the nuclear substance and not the cytoplasmic cell substance imposes its specific character on the developing trait.

Since the dependence of the developmental processes subsequent to the blastula stage on certain definite chromosomes has been determined, and since, on the other hand, it has been demonstrated that the chromosomes of the sperm nucleus, even in the absence of those of the egg nucleus, possess the qualities necessary for the development of all these characters, it may be supposed that the spermatozoon in the normally fertilized egg has an effect on all the processes beginning with the formation of the primary mesenchyme. If this were not the case, then it would have to be concluded in connection with the results of merogony that the sperm chromosomes serve to make possible the development of the traits under discussion, but that the character of these traits is determined not by them but by the egg protoplasm. Actually, Driesch (15) did conclude from his experiments on hybridization of various sea urchin species that all larval characters with the exception of the skeleton were purely maternal and not affected by the sperm chromosomes. My own experiments however, demonstrated to me that these statements were erroneous. Not only, as I have shown previously (5, 7), the form and skeleton of the pluteus, but also the shape of the larvae before the formation of the skeleton, the amount and the pattern of the pigments and the number of primary mesenchyme cells may be influenced by the spermatozoon. This means, therefore, that precisely from the time when certain definite chromosomes, known to be present both in the egg and in the sperm nucleus, prove essential for further development, precisely from this point on, developmental processes show themselves influenced in their specificity by both parents equally; whereas earlier stages, for which, according to our results, specific chromosomes are not necessary, demonstrate a purely maternal character (Boveri, 6, p. 469; Driesch, 15). From all these facts, it will have to be concluded that the role of the chromosomes in ontogenesis corresponds rather exactly to the views which have found a brief though not very fitting expression in the designation of these structures as "carriers of heredity."

I would like to ascribe to the cytoplasm of the sea urchin egg only the initial and simplest of properties responsible for differentiation. Polarity and bilateral symmetry depend on the cytoplasmic pattern, and all malformations connected with these axial relations, such as duplications of larvae or the perpetual blastulae originating from fragments of the animal half only and incapable of undergoing polar differentiation, are based on disturbances of defects of the cytoplasm.{Cf. here my experiments (9). Since that time I have repeated and extended the experiments on the development of purely animal (completely free of pigment) fragments of the Strongylocentrotus egg; not one of such cultured fragments from three different females developed beyond the blastula stage, whereas all pigmented fragments cultured as controls, and among them considerably smaller ones, developed into plutei}. The structure of the egg cytoplasm takes care, if I may say so, of the purely "promorphological" tasks, that is, it provides the most general basic form, the framework within which all specific details are filled in by the nucleus. Or, the relationship may perhaps also be expressed by stating that simple cytoplasmic differentiation serves to start the machine whose essential and probably most complicated mechanism is located in the nuclei.

I am able to clarify this interpretation still further if I compare it briefly with the opinions that Driesch (18, 19) recently expressed about all attempts to explain ontogenetic events. He says: only a complicatedmachine could achieve what we are facing in ontogenesis; however, we are not dealing with a machine here, since a machine would not remain the same if random parts were removed or if parts were transposed in a random fashion, as can be done without harmful effects both in the cytoplasm and nuclei of the Echinus egg. The contradiction construed here by Driesch which, in addition to other considerations, leads him to postulate an "autonomy of living processes," appears to me not to exist in reality. I want to disregard completely the fact that the statement that any part of the cytoplasm could be removed without harming the potencies of the remaining parts has now had to be restricted very considerably. However, what is more important, also, the assumption that the cytoplasm could be transposed at random in the young egg without harmful effects is based on insufficient experience. I have demonstrated earlier (9), and since have been able to determine even more exactly, that minute translocations of the cytoplasm at the vegetative pole lead to the formation of duplications; in the meantime, I have obtained larvae with a duplicated or even triplicated archenteron and others with severe deformations and malformations of the skeleton from clusters of translocated blastomeres provided that the translocations were not corrected, as is often the case. {My colleague Driesch kindly informed me that he also repeated his earlier experiments on the translocation of blastomeres (14, 16) and that he now obtained essentially the same results as I did}. The Echinus egg therefore, is nothing less than a harmonic equipotential system. Finally, however, and this is the decisive point, any portion of "nuclei" but not any portion of "a nucleus" may be removed from the young Echinus egg. Taking something away from the nucleus has not even been tried in the experiments of Driesch; my own experiments, which did accomplish this, teach us that the nucleus, whose structure may have any degree of complexity, behaves just as Driesch demands from a "machine" in the discussion quoted above.

The conflict that Driesch feels is in my opinion resolved by these facts in a simple manner. It is certainly true that the hypothesis maintained by Roux and Weismann (25, 26, 29) of a differential distribution of that complex structure, postulated now also by Driesch, by way of differential nuclear division, has been disproven, at least for the early development of Echinids, by the experiments of Driesch. {Cf., also, my speculation in 3 (pp. 182 ff) and in 8 (pp. 7 ff) about the difficulties which the development and constitution of multipolar mitoses present for the assumption of a differential nuclear division. The objections raised against these speculations on the basis of the pathological effect of multipolar mitoses are based on a logical mistake}. However, it appears to me that the quite peculiar interaction of the cytoplasm with its simple structure and differential division and the nucleus with its complex structure and manifold total multiplication may still achieve what Weismann and Roux attempted to explain with the help of differential nuclear division. When the primitive differences of the cytoplasm, as expressed in the existence of layers, are transferred to the cleaved egg without any change in the relationships of the layers, they affect the originally equal nuclei unequally by unfolding (activating) or suppressing certain nuclear qualities, as may be visualized directly in the cleavage of Ascaris. The inequalities of the nuclei, in some cases perhaps of temporary nature only, lend different potencies to the cytoplasm, that to begin with was differentiated only by degrees. Thus new cytoplasmic conditions are created which again release in certain nuclei the activation or suppression of certain qualities thus imprinting on these cells in turn a specific character and so on, and so on. In short: a continually increasing specification of the originally totipotent complex nuclear structure, and consequently, indirectly, of the cytoplasm of the individual cells, appears conceivable on the basis of physico-chemical events once the machine has been set in motion by the simple cytoplasmic differentiation of the egg. To explain the origin of normal larvae from isolated blastomeres, as well as from fragments of the egg and the blastula, it is, according to this view, necessary only to propose the assumption-well supported, incidentally-that these fragments obtain from the egg differences such that they release the first nuclear differentiations in the identical manner as does the cytoplasm of the entire egg. The sea urchin egg, apparently one of the eggs with the simplest cytoplasmic structure, teaches us that not every region is able to do this; and we know of other eggs (Ctenophores) in which the releasing egg structure is differentiated so highly that no isolated part of the cytoplasm is able to take the place of the whole.

Whoever has followed the literature on these questions that are under so much discussion knows that various authors have opinions that agree more or less closely in one or the other point with those just expressed. O. Hertwig, Weismann, de Vries, should be cited here. Furthermore, Driesch earlier developed possibilities, which he later rejected again, corresponding in many respects to my own point of view. Also with respect to Roux's doctrine, the common points, such as the interpretation of the nucleus as the real determinant, and that of the differentiation of the cytoplasm as a releasing factor, appear to outweigh by far the difference which lies in his assumption of qualitatively unequal nuclear division. The progress which, as I believe, has been attained by my experiments consists in just this: that now, even though in a field that is still narrow, speculations have been replaced by facts.

Of the manifold relations to other problems inherent in our results, only two points should be considered here briefly; first of all, whether any phenomena are known which appear in a new light as a result of the new insight. In this respect, it appears to me that certain asymmetries, that appear as abnormalities in bilateral animals, particularly in insects, may find a simple explanation on the basis of my results. If a bee has the structure of a drone on its right side and of a worker on its left side, then the right side has developed like a parthenogenetic and the left side like a fertilized egg, i.e., the right side like an egg which has maternal chromosomes only, the left side like one with chromosomes of both parents. On the basis of this consideration, and since it could be demonstrated that in the sea urchin egg asymmetries of a definite kind may be created by an unequal chromosomal composition of different egg regions, the conclusion is almost unavoidable that the reason for asymmetries of insects consisting in a mosaic configuration of male and female areas will also have to be looked for in nuclear differences. In the case just mentioned of purely symmetrical hermaphroditism, it cannot be a question of dispermy. We have to consider a different abnormal chromatin distribution such as I have found earlier in sea urchin eggs (2).{Cf. here alos the more detailed demonstration that E. Teichmann (28) gave in this connection based on material preserved by me}. Here one half-blastomere contains maternal chromosomes only and the other mixed maternal and paternal chromosomes, i.e., precisely what had to be assumed for the hermaphroditic bees if the reason for this abnormality lies in the chromatin. Due to the peculiar conditions of bee development, the occurrence of such an abnormality is apparently much favored since it appears possible that the egg nucleus is already divided before union with the sperm nucleus as a result of its parthenogenetic potencies, and that the sperm nucleus unites only with one of the cleavage nuclei. This union could even be postponed until later cleavage stages and polyspermy, which is known to occur in bees, could have the effect that sperm nuclei unite with certain derivatives of the egg nucleus and not with others. In this way, the most diverse mixtures of male and female characters could result, as has been actually observed.

{Compare C. Th. von Siebold (27). It should perhaps be objected, against the explanation mentioned above, that a cleavage nucleus capable of division by itself and a sperm nucleus together would cause a quadripolar figure and thus pathological development of the corresponding egg region, similarly to two blastomere nuclei with their two centrosomes in the Echinus egg. However, just as the sperm nucleus in the bee egg forms a regular division figure with the egg nucleus capable of independent division, so this will be possible with a later cleavage nucleus. We are dealing here no doubt with conditions of the cytocenters which deviate from those of the sea urchin egg and probably most other eggs}.

Finally, a second question which should be touched upon briefly is that of the consequences of multipolar mitoses in later embryonic stages and in mature tissues. A beginning in this direction may be reported already. In the Echinus egg I succeeded in certain individual blastomeres, e.g., in one of the half or quarter blastomeres, to produce multipolar division figures in the macromeres or mesomeres and thus to render pathological the particular egg region arising from these. The details, interesting in other connections, shall not be discussed here; it is of importance for our considerations that in those experiments which cause pathological conditions exclusively in the derivatives of the macromeres or of the mesomeres, the formation of the pluteus is usually not hindered. {It may however be mentioned that the pathological development of one half blastomere always leads to an exclusive defect in the left or right body half, from which one may conclude that the first cleavage furrow determines the median plane, unless stronger influences, such as I have demonstrated in the deformation of the egg (9), inhibit this}. The pathological cells frequently enter the cleavage cavity (primary coelom) sooner or later in large numbers, but the normal parts group themselves into a smaller whole, just as in the case of complete removal of portions of the egg.

The fate of the pathological cell clusters that enter the interior cannot be determined in view of the limited life span of sea urchin larvae raised artificially. If, however, we want to classify these formations according to the points of view of pathological anatomy, then we have to designate them as "tumors" and thus arrive at the statement that multipolar mitosis might under certain conditions lead to the development of tumor-like formations. Could not this conclusion throw some light on the riddle of tumors? We are confronted here with quite a peculiar phenomenon, namely that a cell complex loses to some extent the normal qualities of its tissue, and by the maintenance or even occasionally an increase of the ability of the cells to multiply, a departure from the parent tissue and an abnormal proliferation contrary to the plan of the whole occur. It is not disease in the sense of a decrease of vitality, but in the sense of an aim in the wrong direction, that is probably the essential property of the tumor cell. Since it could be shown on the one hand that multipolar mitoses lead to the origin of such cells which have lost their balance, and since on the other hand it is known that simultaneous multipolar divisions are found in tumors, the hypothesis of a connection between these two phenomena seems worthy of an examination. However, it would have to be supposed, in addition, that not only in the developing, but even in the originating tumor, multipolar mitoses occur. What may cause these is a second question and I note that my hypothesis is not irreconcilable with the assumption that the first cause of tumors is of parasitic nature. If I survey reports about the etiology of carcinoma and the many suggestions of physical and chemical insults, and if I consider all the other hand that pressure, shaking, narcotics, abnormal temperatures are precisely the agents with whose help we may produce multipolar mitoses in young eggs, then it appears possible to me that we have before us, in the elements just considered, the entire causal sequence of certain tumors.

Dispermic Eggs Figures

(From Theodor Boveri's more detailed 1907 paper: Zellen-studien VI. Die Entwicklung dispermer Seeigeleier. Ein Beitrge zur Beifruchtungslehre und der Theorie des Kernes. Jena Zeit. Naturwiss. 43: 1-292.)

Figure 1

Figure 1   Two figures from plates II and III of Boveri's definitive 1907 publication on the development of dispermic sea urchin eggs. The figure at the left is a pluteus from a Strongylocentrotus egg that divided into three cells at first cleavage, viewed from behind. Magnification is about 650X. The figure at the right is a pluteus larva from a Sphaerechinus egg that divided into three cells at first cleavage, viewed from the front. The border of the third that contained small nuclei is shown by a line (red in the original figure). A few nuclei on either side of this border, along its course on the front surface, are shown to illustrate the differences in their sizes.

Figure 2

Figure 2   Figure indicating the possible random combinations of chromosomes a, b,c, and d on the spindles of tetrapolar eggs. The unstiplled blastomeres contain at least one of each type of chromosome (and Boveri assumed them to develop normally). The stippled blastomeres are deficient in at least one chromosome type (and would therefore be predicted to develop abnormally).

Literature Cited, Including Footnotes

1. BOVERI, TH., Über die Befruchtung des Eies von Ascaris megalocephala. Sitz. Ber. d. Ges. f. Morph. u. Phys. Munchen, Bd. 3. 1887.

2. BOVERI, TH., Über partielle Befruchtung. Sitz.-Ber. d. Ges. f. Morph. u. Phys. Munchen, Bd. 4. 1888.

3. BOVERI, TH., Zellen-Studien, Heft 2, Jena 1888.

4. BOVERI, TH., Zellen-Studien, Heft 3, Jena 1890.

5. BOVERI, TH., Ein geschlechtlich erzeugter Organismus ohne mütterliche Eigenschaften. Sitz.-Ber. d. Ges. f. Morph. u. Phys. Munchen, Bd. 5. 1889.

6. BOVERI, TH., Befruchtung. Ergebn. d. Anat. u. Entw.-Gesch. Bd. 1. 1892.

7. BOVERI, TH., Über die Befruchtungs- und Entwickelungsfähigkeit kernloser Seeigeleier und die Möglichkeit ihrer Bastardierung. Arch. f. Entw.-Mech. Bd. 2. 1885.

8. BOVERI, TH., Zur Physiologie der Kern- und Zellteilung. Sitz.-Ber. d. phys.med. Ges. Würzburg 1897.

9. BOVERI, TH., Über die Polarität des Seeigel-Eies. Verh. d. phys.-med. Ges. Würzburg, N. F., Bd. 34. 1901.

1O. BOVERI, TH., Die Polaritat von Ovocyte, Ei und Larve des Strongylocentrotus lividus. Zoolog. Jahrbücher Bd. 14. 1901.

11. DELAGE, Y., Études sur la Mérogonie. Arch. de Zool, exp. et gén., 3. sér., T. 7. 1899.

12. DELAGE, Y., Études expérimentales sur la Maturation cytoplasmique chez les Echinodermes. Arch. de Zool. exp. 3. sér., T. 9. 1901.

13. DRIESCH, H., Entwicklungsmechanische Studien V. Von der Furchung doppeltbefruchteter Eier. Zeitschr. f. wiss. Zool. Bd. 55. 1892.

14. DRIESCH, H., Betrachtungen über die Organisation des Eies und ihre Genese. Arch. f. Entw.-Mech. Bd. 4. 1896.

15. DRIESCH, H., Über rein-mütterliche Charaktere an Bastardlarven von Echiniden. Arch. f. Entw.-Mech. Bd. 7. 1898.

16. DRIESCH, H., Die Lokalisation morphogenetischer Vorgänge. Ein Beweis vitalistischen Geschehens. Arch. f. Entw.-Mech. Bd. 8. 1899.

17. DRIESCH, H., Die isolierten Blastomeren des Echinidenkeimes. Arch. f. Entw.Mech. Bd. 10. 1900.

18. DRIESCH, H., Die organischen Regulationen. Leipzig 1901.

19. DRIESCH, H., Kritisches und Polemisches. Biolog. Centralblatt Bd. 22. 1902,

20. HERBST, C., Über das Auseinandergehen von Furchungs- und Gewebezellen in kalkfreiem Medium. Arch. f. Entw.-Mech. Bd. 9. 1900.

21. HERTWIG, 0., und R., Über den Befruchtungs- und Teilungsvorgang des tierischen Eies unter dem Einiiuss äusserer Agentien. Jena 1887.

22. HERTWIG, R., Über die Entwicklung des unbefruchteten Seeigeleies. Abh. d. k. b. Ak. d. Wiss., II. Kl., Bd. 29. 1898.

23. LOEB, J., On the Nature of the Process of Fertilization and the Artificial Production of Normal Larvae (Plutei) from the Unfertilized Eggs of the Sea Urchin. Americ. Journ. of Physiol. Vol. 3. 1899.

24. MORGAN, T. H., A Study of Variation in Cleavage. Arch. f. Entw.- Mech. Bd. 2. 1895.

25. ROUX, W., Über die Bedeutung der Kernteilungsfiguren. Leipzig 1883.

26. ROUX, W., Beiträge zur Entwicklungsmechanik des Embryo. III. Breslauer ärztliche Zeitschr. 1885.

27. v. SIEBOLD, C. TH., Über Zwitterbienen. Zeitschr. f. wiss. Zool. Bd. 14. 1864.

28. TEICHMANN, E., Über Furchung befruchteter Seeigeleier ohne Beteiligung des Spermakerns. Jenaische Zeitschr. Bd. 37. 1902.

29. WEISMANN, A., Das Keimplasma. Eine Theorie der Vererbung. Jena 1892.

30. WILSON, E. B., Experimental Studies in Cytology. I. A Cytological Study of Artificial Parthenogenesis in Sea-urchin Eggs. Arch. f. Entw.-Mech. Bd. 12. 1901 .

31. WILSON, E. B., Experimental Studies in Cytology. III. The Effect on Cleavage of Artificial Obliteration of the First Cleavage-Furrow. Arch. f. Entw.-Mech. Bd. 13 1901.

32. WINKLER, H., Über Merogonie und Befruchtung. Jahrb. f. wissenschaftl. Bot. Bd. 36. 1901.