J. Hered. 39: 311-325 (1948)
I. Initial observations on Allium cepa
Department of Botany, University of Wisconsin

IN the course of studies of chromosomes in differentiated regions of the roots of Rhoeo discolor28 two haploid cells and two nuclei having chromosome catenations, similar to those characteristic of its pollen mother-cells, and apparently undergoing meiosis were found. Two haploid cells were also found in Allium roots. These Rhoeo and Allium plants had been treated with indole-3-acetic acid, but there is no reason to suspect that it induced the chromosome reduction (contra Mather36). A haploid cell was also found in untreated Paphiopedalum (Duncan, unpublished). Later,23 reduction-divisions and haploid nuclei were found in considerable numbers in roots of onion bulbs growing in water containing 1-4% sodium nucleate. In subsequent extensive studies of the effects of sodium nucleate, somatic reduction has been found to occur occasionally in the controls in several different species of plants, and with fairly high frequency in roots from some wilted onion bulbs that had been kept on a laboratory shelf all winter. These findings have demanded an examination of the literature for other cases of somatic reduction, or data that could be so interpreted, and seem to warrant a preliminary assay of its possible wider occurrence and significance.

Berger5 showed that somatic reduction is a regular feature in the ileum of mosquito larvae. His report was received with considerable skepticism (see 8) but has since been confirmed and extended by Grell19,20 who has also clarified certain details in the process. The sporadic occurrence of reduction in somatic nuclei has been noted in various plants and animals by a number of workers. Gates18 observed reduction-divisions in the nucellus of Oenothera lata and from conversations it appears that many cytologists have observed them in this tissue in various plants, as I did in Matthiola incana in 1930. Few such observations have been published because they are rarely "well-fixed" and are therefore usually dismissed as of doubtful validity. The occurrence of reduction is record­ed in many cytological studies of mammalian tumors, but again with varying degrees of validity on account of the frequent clumping of chromosomes and of irregularities of many kinds in such material. Unpublished reports of haploid nuclei in Paphiopedalum roots and of reduction-divisions in roots of brome grass have recently been received from R. A. MacLeod and I. W. Knohloch respectively.

Battaglia2 has recently shown that in Sambucus Ebulus L. "the somatic cells constituting the basal portion of the style tissue . . . undergo an usual [sic] heterotypic division with formation of bivalents and binucleate cells [two haploid nuclei]". Christoff and Christoff10 have found that in Hieraceum hoppeanum "an unusual reduction-division [occurs] in the integumental cells, which is influenced by some external factors." It results in plants with 45 chromosomes arising from one with 2n=90.

*It was a "graft-hybrid"; the parents (or rather, parental components) had 2n=24 and 72. The sub-epidermal layer had 48 chromosomes; Winkler's assumption was that two parental nuclei fused to give one with 96 chromosomes which had then undergone reduction to give the cell-layer with 48. Endotetraploidy in the diploid component now seems more plausible.

The most completely analysed case of somatic reduction in plants is that recorded recently by Brown8 in cotton. In this instance it seems that the cytogenetically unbalanced nature of the parent polyploid plant favored the development of a reduced tissue. There seems little doubt that reduced cells occur not too infrequently in many plants but that they rarely form flowering branches or large segments. Winkler's58 Solanum darwinianum* which is usually cited as the classical case of somatic reduction is not so listed here nor are early inconclusive accounts of reduction in Amphibian tissues. On the other hand, some of Nemec's42 early reports of reduction following chemical treatments might perhaps be worth reinvestigation.

Premeiotic reduction as a regular feature in a portion of the hermaphrodite gonad was established by Hughes-Schrader21 in the coccid Icerya purchasi. She found no chromosome pairing or heterotype divisions preceding the appearance of haploid nuclei and was unable to suggest a mechanism for reduction. But she pointed out that in Pseudococcus and Lecanium there is a perfectly regulated segregation at anaphase without any prior synapsis and that Metz38 had shown regular segregation in Sciara "without any previous pairing, synapsis, or even any alignment in an equatorial plate stage." She noted that the appearance of haploid nuclei is concomitant with degeneration of many diploid nuclei. These observations may bear on the experimentally induced reduction-divisions reported below.

In an interspecific Sciara hybrid Metz39 found a mosaic salivary gland in which some of the nuclei had chromosomes of only one of the parental species. The segregation had apparently taken place a few cell generations before the final mitosis in the gland.

It seems strange that with all these cases of reduction noted in the literature (and doubtless many more would be found were a more thorough search made) more extensive consideration has not been given to their possible bearing on such questions as the reversal of polyploidy, the origin of sexuality, unusual cases of segregation, somatic "mutation," etc. Some of the possibilities have been mentioned elsewhere.23 Consideration of them in fuller detail may profitably be deferred until more data are available from the experimental methods now at hand for increasing the frequency of somatic reduction and segregation, but they should certainly be borne in mind. Re-reading of Bateson's papers, particularly his last, may be salutary in this regard. It shows that a major factor in his hesitant and incomplete acceptance of the chromosome theory was his conviction that: "Though segregation is commonly effected at the reduction-division, evidence steadily accumulates showing that at least in plants of many kinds comparable segregations occur in somatic divisions also." And again: "Somehow a somatic cell is evidently able to divide in such a way as to produce cells dissimilar either from the parent cell or from each other or both." He warned that "to speak of all such sports as 'mutation' . . . is likely to introduce confusion." Though other explanations than somatic segregation have been found for some of the cases that troubled Bateson, most of them have been ignored or forgotten during the past quarter century when it has been fashionable in some schools of cytogenetics to ignore rather than to "treasure your exceptions" as he advised (cf Sehrader48). That Bateson could not envisage somatic segregation in terms of chromosomes, as we can, is not surprising in view of the opinion then fairly widely held (and cited by him from Strangeways) that "the chromosomes . . . disappear altogether after telophase and pass into solution." Further, he accepted the cytologist's dictum that in the developing zygote "whatever the resulting tissue, its chromosome content is the same," How could he reconcile his evidence of somatic segregation with these cytological concepts? We now know: (a) that the chromosomes do not dissolve at telophase and (b) that quantitatively the chromosome constitution is not the same in all cells.27,28 Further, we now have serious reason to suspect that chromosome segregation in somatic tissues may be of widespread occurrence.

Jones29,30 also has emphasized that many variations in maize endosperm which earlier were described as somatic mutations, may actually be due to segregation. He interprets much of his later evidence31 as indicating translocations, deficiencies, etc., rather than regular somatic crossing-over. It is possible, however, that, like Bateson earlier, his interpretations have been limited by the cytological picture available to him. Negatively, we have, of course, to discount the concept which he cites from Weatherwax, that the nuclei of the endosperm commonly undergo "direct," i.e., amitotic, division. Then, to current cytological evidence of the widespread occurrence of polysomaty, somatic segregation and reduction, there has to be added the fact that in the later stages of development of the maize endosperm its nuclei are polytene (Duncan16). This latter provides an obvious mechanism for the delayed appearance of the effects of segregation, which apart from polyteny must be a complex process in a triploid tissue. Further discussion of Jones' invaluable genetic data or of McClintock's various beautiful and precise cytogenetic analyses of the behavior of maize chromosomes, would, however, be beyond the scope of the present paper.

Genetic evidence of somatic reduction is given in the well-known work of Bridges7 relative to sex-determination in Drosophila, but this was considered due to the maternal chromosomes being unable to "maintain the normal division-pace." Evidence of crossing-over and segregation in somatic cells was given by Muller41 in an early paper on "The Mechanism of Crossing-over." He showed that 100% of the offspring of a certain male fly were crossovers and that the crossing-over must have occurred in an early embryonic cell that was ancestral to both somatic and gonial tissue. There was no evidence of haploidy accompanying the segregation; on the four-stranded crossing-over concept, two crossover chromatids must have segregated into a nucleus destined to become ancestral to spermatogonia and two noncrossover chromatids into somatic cells, with all nuclei remaining diploid. Much other evidence of somatic crossing-over in Drosophila is given by Stern,51 Whittinghill54 and others. "Somatic pairing" or association of homologues occurs regularly in the Diptera and this has tended to create an impression that somatic crossing-over in Drosophila is a special case and perhaps the only one for which a definitely known mechanism exists; the evidence does not warrant this conclusion. The "somatic pairing" described in many plants by numerous authors (see Watkins53) will not be considered in detail here since the value of the evidence for it varies greatly in different cases. It is, however, of interest to note Strasburger's52 insistence that in plants attraction of chromosomes occurs only in pairs and not between higher multiples. This is not true of the Diptera.37 The definite association of non-homologous somatic chromosomes through interchanges of chromatids as found by Peto45 is a distinct issue, though he termed them chiasmata. They are apparently due to random breakage and fusion under the influence of the chloral hydrate treatment. Doubtless this was involved in some of Nemec's results also.

Gonomery should also be disregarded here, though it gives superficially similar cytological appearances. Where it occurs, the two parental genomes may remain distinct within the one nucleus and mitoses similar in appearance to some of those shown herein therefore occur regularly. But as Wilson55 pointed out, it is found "only in the earlier stages of development" and it "represents no more than a tendency on the part of chromosomes to remain in separate maternal and paternal groups." This tendency is soon lost and "chromosomes of maternal and paternal ancestry become intermingled as development proceeds." Gonomery, therefore, can assuredly have no significance in causing the onion reduction figures since these cells are removed by many mitotic cycles from the original zygote. Even more does this hold for similar figures obtained in Tradescantia nuclei which have a very long history of vegetative reproduction. To settle this matter experimentally, and the related one of possible genome segregation, a report will be published shortly of a study which has been made on an orchid hybrid in which six of the maternal and paternal chromosomes can be distinguished.

In a 42-chromosome wheat plant that was an F7 from a pentaploid hybrid, Love35, found two pollen mother-cells that had undergone "premeiotic" reduction and whose 21 chromosomes were undergoing a second, ordinary meiotic, reduction. One of these cells had 7II + 7I and could therefore have produced either one or two 7-chromosome, ancestral-type gametes.

Either somatic or gonial reduction followed by gonocyte meiosis or a gametophytic reduction following the latter could account for cases such as (1) East's17 recovery of a 7-chromosome ancestral-type gamete from a 42-chromosome strawberry, (2) Nishiyama's43 diploid oat resembling Avena strigosa from a pentaploid hybrid of A. fatua X A. barbata, (3) Kiellander's32 18-chromosome Poa resembling P. trivialis from a 72-chromosome P. pratensis and other such, It is probable that many such cases are known to workers with polyploids. They have not usually been emphasized because in most cases the possibilities of contamination with foreign pollen or accidental admixture of seeds cannot be ruled out as they were by East and Nishiyama. There are doubtless also many more unpublished cases for, as with cytological data above, several have been communicated to me personally since the first announcement22,23 of this work was made. Dr. A, H. Sparrow has kindly drawn my attention to three cases recorded in his M.S. thesis.50 These were 28-chromosome plants arising in a 42-chromosome hybrid wheat line. They regularly formed 14 bivalents.

General Remarks and Acknowledgements

It has long been obvious that nucleic acids must play some very significant role in chromosome mechanics; the speculations on this role are too numerous to cite. The most clear-cut data and forthright conclusions were given by Caspersson.9 He showed that the spermatogonia in a grasshopper are "nucleic acid-poor" and the primary spermatocytes "nucleic acid-rich." The synthesis of nucleic acids is completed before chromosome contraction in meiosis and it seems to have been synthesized in the spermatocyte nucleus as there is no appreciable amount of it in the cytoplasm or testicular fluid during the time so many cells are developing so quickly. He concludes: "These facts make it probable that nucleic acid plays a role in the synapsis or in the division of chromonemata or in both."

*Consider Miescher40 speaking of a more "exact" science: "Alles, was man finden will, muss man zuerst voraussetzen und vermuthen"—an overstatement, one hopes, but a useful corrective to the idea that it is necessary only to "describe what you see."

As part of a research program22,23 which aims, in general terms, to combine the methods of "experimental cytology" with those of cytogenetics, various experiments with ribose nucleic acid, sodium nucleate and various components of nucleic acids were planned and the materials for them obtained in 1946-7. Dr. M. Kodani, who was with us temporarily during October and November 1947 as a Project Associate, was encouraged in his desire to determine their effects on onion roots. His prime interest, arising from some of his earlier studies,33,34 was to compare their effects on root growth and on chromosome structure with those of a salivary-gland extract; mine, following Caspersson's "lead," was on their possible effects on chromosome reproduction and synapsis. As not uncommonly happens in cytology, we both found what we were looking for.* Dr. Kodani is publishing separately his observations on root growth, the production of "lampbrush" and condensed "meiotic-like" chromosomes and on irregularities of chromosome behavior and distribution. Herein are the observations, the first of which were made on Dr. Kodani's slides, which convinced me that one of the effects of sodium nucleate is the production of an effective, albeit in part unusual, type of reduction-division that may provisionally be termed "somatic meiosis."

It must be emphasized that it was im­mediately recognized that Allium cepa is unfavorable material for establishing definitely the detailed process of "somatic meiosis" and its results. This species has eight pairs of chromosomes of which at most, five can be individually identified, and those almost always very doubtfully; and it is not favorable for genetic tests. Several experiments on different plants were started immediately. and Dr. Kodani planned one on Drosophila but was obliged, owing to a prior commitment, to leave for Japan. At the meetings of the Genetics Society in Chicago, December 28, 1947, the preliminary observations on Allium, Tradescantia, and Rhoeo, the latter with catenations in treated root-tip nuclei similar to those of its normal pollen mother-cells, were demonstrated. Cytological, cytochemical and cytogenetic tests of the effects of nucleates and their components and of amino acids on the nuclei of various plants and animals are now being made by our research group. These probably color to some extent the interpretation of the Allium data here presented, but they will be offered separately and as far as possible independently.

This will necessitate giving an apparently hesitant and tentative interpretation to some of the Allium figures that in the light of fuller knowledge from other materials could be interpreted more boldly. If this results in a presentation that does not carry complete conviction it may, as a corollary, show why naturally occurring somatic reduction-divisions have so generally been overlooked. Many cytologists must have seen, in untreated roots, figures like many of those herein, and, quite properly, decided not to attribute significance to them. Only when they are found repeatedly and their details correlated do they mean anything. Certainly such a presentation will show the necessity for diverse experiments on various organisms if all details of the process are to be elucidated. Rhoeo, for instance, being a translocation heterozygote, will give evidence on the specificity of pairing. At the same time it may help clarify the role of heterochromatin in somatic association. Crocus gives evidence on "distance conjugation" without which some of the Allium observations seem inexplicable. Trillium and Crocus both give evidence on the specificity of the repulsion between homologues. The tomato is favorable for cytogenetic analyses. The irregularities of mitosis that are found in many of the treated nuclei will not be considered here since they are being described independently by Dr. Kodani. It seems at present that in their effects on chromosome reduction, nucleic acids are different in kind from many mitotic poisons, such as colchicine, acenaphthene, ethylene glycol, chloral hydrate, etc., while in the production of various irregularities, many of which are simply necromorphisms, in part due to over-dosage, they may be similar.

I am particularly indebted to Dr. Kodani for the use of the initial preparations and to Dr. Lotti Steinitz, Dr. Rhona Leonard-Bennett, Marianne Weisz, Norah Stewart, Dr. Irving Galinsky, R. P. Patil, and Walter Drapala for making subsequent treatments and from them sectioned and squash preparations. Replicated tests of six samples of sodium nucleate were also made by 12 students as a class exercise. To Dr. G. B. Wilson I am indebted for recent help in examining preparations and selecting nuclei for illustration, to Dorothea Voss-Helmen for help with drawings and to Alfred Owczarzak with the photomicrography.


The method, used initially by Dr. Kodani, is simply to grow onion bulbs in tap water and then to place them with their roots in a solution of ribose nucleic acid or its sodium salt, in distilled water. The sodium nucleate has simply the ad­vantage of greater solubility. Concentrations ranged from 0.5% to 8.0% and period of treatment from 3 to 36 hours. For the onion the optimum is probably between 1 and 2% for 6 to 12 hours, but it seems that this varies greatly with the condition of the bulbs and growth-rate at the time of treatment. In some treatments the pH was adjusted to neutrality, usually with NaOH. It may be added that different plants will tolerate, and require for an effect, very different concentrations.

Some good results have been obtained with excised roots placed in water or culture solution to which sodium nucleate has been added. With some plants injection and spraying the leaves have both been effective. Standardized methods have yet to be developed and the role of the components of nucleic acid and their interactions with certain other substances in the cell have yet to be determined.

The sodium nucleate used on onions was obtained from the Schwarz Laboratories, Inc., New York. It is made from yeast and is not critically purified. Since it can scarcely be imagined that the complicated processes of meiosis are specific effects of any one chemical substance, the lack of purity was no disadvantage in the initial studies reported herein. Six small and one large sample from Schwarz have been tested since the results obtained with the original lot (S.N. 4509) were first reported.23 All eight samples cause chromosome segregation, but there seems to be less pairing with the latest large sample (S.N. 4704) than with the first. On the other hand, this may be due to other variables, not ignoring subjective ones. The six small samples did not permit any quantitative evaluation of their effects.

Sectioned material has been used to only a very limited extent, since essential details of chromosome structure are much more favorably shown in squash preparations of root tips fixed in 3:1 alcohol-acetic, hydrolyzed with HCl and stained with Feulgen or aceto-carmine.


Many of the nuclear divisions in the meristem of treated onion roots seem to be completely ordinary, normal mitoses, A considerable proportion show unusual features, such as shortened and thickened chromosomes, but appear to be dividing regularly. A few show grossly abnormal chromosome distributions and others have "lampbrush" chromosomes. A reductional type of division occurs in not less than 2% of all mitoses in some experiments. A maximum cannot be stated since many division figures, naturally, cannot be interpreted with certainty see below. In most of the "somatic meioses" the cells appear to be in a normal, healthy condition.

The frequency with which miscellaneous irregularities occur is directly related to dosage and length of treatment and can be predicted from the degree of flaccidity of the roots before fixation. In roots that appear turgid and healthy when fixed, the chromosomes almost all have a normal appearance ; they are not clumped, fused or ragged in outline as after many chemical treatments. In general the cytological impression in healthy looking roots that have a high proportion of reducing divisions is me of disturbance within the range of viability or "normality." It is, therefore, particularly interesting to note that several of the features which are unusual (and therefore "abnormal") for the onion, resemble normal features of mitosis or meiosis in other organisms.

Separation of 16 condensed chromosomes, that look like meiotic univalents, into two groups of 8, as in Figure IA, can be found in a number of cells; to decide whether in any one case it is a significant grouping or a chance result of pressure in making the squash preparations may be more difficult. Groups of eight in two successive sections have twice been found; this may seem more convincing than evidence from squash preparations, but the latter method has advantages for the determination of exact numbers and of chromosome structure.

Figure 2

Root-tip cells of Allium cepa treated with sodium nucleate showing the equivalent of a second meiotic division and the overlapping of two divisions. A and B: Cells with two haploid nuclei; C-F: the overlapping of two divisions in "somatic meiosis"; E, Upper group at mid-anaphase II, lower group at early anaphase If. See text for details. D', E' and F' are drawings interpreting Li, F and F. All ca 1300X. Squash preparations, C aceto-carmine, remainder Feulgen.

*The topological and interpretative complexities involved in the correlation of cytological chiasmata and genetic crossovers are not always recognized, (cf Huskins and Newcomb and Schrader48). The great advance in cytogenetic theory made possible by the brilliant analyses of Belling4 and Darlington13 has been to a considerable extent vitiated by subsequent uncritical work on chiasma frequencies, terminalization coefficients, etc. (including some by myself prior to 1935). Pending the completion of both genetic and cytological studies now in progress on other materials the only comment that will be made on the problem of chiasmata in somatic nuclei is the following: When pairing and segregation without chiasmata, such as that occurring in Figure 2, was first observed it was emphasized, because in no plant has gonocyte meiotic pairing and segregation without chiasmata so far been established (though some evidence for it even in Trillium26 was presented in 1935). Now the evidence indicates that pairing with chiasmata is the exception in "somatic meiosis" induced by sodium nucleate and a change of emphasis, necessitating different experimental materials, has been found necessary. K. W. Cooper's critical analyses11,12 of the chiasma problem in the Diptera give an indication of the issues that have to be faced.

Actually, in the present study, the first evidence convincing to me came from incomplete separations such as those shown in Figures 1B,C, and D. In each of these there are seven chromosomes at each side of the cell and two of similar size and shape seem to be in process of separating from one another. In Figure 1D this eighth pair appears to be associated by a sub-terminal chiasma. Certainly the configuration would be so described by many cytogeneticists if seen in a meiocyte and it would be interpreted as in the accompanying line diagram.* There is no indication of any prior chiasma formation between the members of the eighth pair in Figures 1B or C. Neither has any sign of a spindle been seen, though Fast Green has been used with some of the Feulgen preparations. However, this aspect of the problem is now receiving special attention.

In Figure 1E two chromosomes of different size are united by an almost terminal chiasma. In another cell a median chiasma united two such chromosomes. The significance cannot be judged without a study of normal meiosis in the same plant. If the two chromosomes are "heteromorphic homologues" and have a region in common, the chiasma could be the result of crossing-over, but such configurations can also result from irregular breakage and reunion of chromatids.

In a number of cells, chromosomes of similar shape and size have been found lying together in pairs without chiasmata joining them, but "somatic pairing" cannot be studied satisfactorily if homologues cannot readily be identified.

In Figure 1F there are apparently eight chromosomes at one pole, seven at the other and one lying between them. The cell is intact but slightly constricted at the middle as if about to be divided by furrowing.

Figure 1G shows a cell in which two groups of eight chromosomes lie slightly separated from each other. Each chromosome has its chromatids widely separated except at the kinomere and resembles a typical metaphase II dyad in pollen mother-cell meiosis, or a "c-pair" resulting from treatment with colchicine. In their grouping they look like meiosis II plates; such groupings are, apparently, not commonly found after colchicine treatments. Figure 1H shows a group of eight dyads and far removed from them, with other nuclei in between, a condensed group that probably contains eight. Figure 11 shows a smashed cell in which there are also two groups of eight chromosomes well separated from each other. The number of smashed cells containing groups of eight chromosomes seems too high for random frequency and suggests that a cell with two such groups in it may be constricted and hence easily broken into two. Obviously no conclusions could be reached from such cells alone, but taken in conjunction with the next two figures they seem significant (see also Figure 3F). Figure 2A is an example of the binucleate cells which are interpreted as having two haploid metaphase II plates, i.e.. as being a stage succeeding that in Figures 1G and I. Figures such as 1G, H, and I or as 2A could not alone carry any conviction; but the former permit recognition of chromosome shape as like that of typical metaphase IT, while the latter shows clearly that there are two metaphase plates in one cell. In such plates there appear to be 8 bipartite chromosomes, similar to those of Figure 1 G, but their crowding prevents clear identification as such. Figure 2B shows a binucleate cell in which the chromosome numbers cannot be determined definitely but there are clearly only about half as many in each nucleus as in that of the adjacent uninucleate cell. It cannot be determined whether this has resulted from separations such as those in Figure 1G-I or from such as 3A-F.

Figures 2C-F must be considered together and also in relation to Figure 1G. In Figure 2E and E' there are clearly two groups of about 16 chromosomes each. The lower group can be interpreted as at early anaphase and the upper at mid-anaphase, but figures somewhat resembling this can be found after treatment with various mitotic poisons. Figure 2C was the first to indicate in itself that a second division, similar in function to that of ordinary meiosis but differing in its time of onset may occur in these sodium nucleate treatments. In the lower half of this cell there are two groups of eight single chromosomes that appear to be separating from each other while in the upper half there are five bipartite and six single chromosomes. Unfortunately this preparation was lost or mislaid before it could be drawn. However, Figure 2D and D' shows an essentially similar cell excepting that the 16 single chromosomes in the lower half are not clearly separating into two groups of eight. In the upper there are twelve single chromosomes and two that are bipartite. In both groups single chromosomes of similar shape lie in pairs in many places. Figure 2F and F' shows what is interpreted as a complete overlapping of the two divisions of "somatic meiosis" and as such clarifies Figures 2C and D as partial overlaps. In it there are clearly two groups of 16 units each comprising eight pairs (the members of the second pair from the top in each group may actually still be united at the kinomere). Attention may be drawn to the constriction at the end of each member of the pair at 3 o'clock. This plant may have been heteromorphic for this chromosome, since none appears in the left-hand group, but we know from study of differential regions (cf Wilson and Boothroyd56,57 contra Darlington and LaCour15), that too much significance must not be placed on variations in the appearance of such constricted regions. Its present significance lies in the evidence it provides that adjacent single chromosomes have been sister chromatids. The only apparent alternative to the interpretation of Figure 2F as a second division separation of chromatids during a first anaphase that is giving numerical reduction (we cannot from onion data conclude that segregation of homologues is occurring) would be that it is a mitotic metaphase plate that has somehow separated into two groups of eight chromosomes. Against this is, first, its general appearance and the unlikelihood of such a separation. In the fact that separation of chromatids is occurring last, instead of first, at the kinomere the pairs of associated single chromosomes (or chromatids) seem obviously to have been derived from bipartite chromosomes like those of Figures 1G-I. They thus resemble either typical meiotic metaphase I! dyads or colchicine "c-pairs." But colchicine has not, so far as I know, been shown to give anaphase-like segregation coincident with dissociation of the "c-pairs."

Figures 3A-F show a condition not noticed during the first months' study of Allium The chromosomes are separating into two groups during what look like various stages of prophase. Figure 3F and F' shows segregation of two groups of eight chromosomes that are contracted to a degree characteristic of very late prophase. Perhaps Figures 1B, C and D should also be interpreted as very late prophase segregations instead of as anaphases. Obviously, however, the standard terminology has little significance here—at least not until we have definite data on the presence or absence of a spindle.

This type of prophase segregation was first noted in the water-poppy, Hydrocleis nymphoides, which Mr. Peter Nelson had given sodium nucleate treatments in concentrations of from 1-8%, He found no pairing or segregation such as shown in Figures 1A-2F. On one of his slides, however, I found five nuclei resembling that of Figure 3A, On checking back they were found in the union slides also. They are now found far more frequently than any other type following treatment with sodium nucleate. This applies to 11 different plant species currently being studied. There is an obvious subjective factor involved, but there seems also to be more than that. Present indications are that sodium nucleate chiefly affects repulsion between chromosomes and, further, that it is the phosphorus constituent which is the effective agent (Galinsky). Chromosome pairing seems to depend upon some other nuclear constituent, preliminary indications from experiments with two amino acids being that they play a significant part in it. If so, the difference it, frequency with which pairing and segregation of condensed chromosomes took place in the early experiments and prophase segregation in the more recent ones may be due to differences in the quantity of impurities in the samples of sodium nucleate and/or the water in which they were grown until treated. The tap water used in the early experiments had a high content of organic matter. On the other hand, growing conditions, growth rate and, with onions at least, the condition and age of the bulb all affect the results. These issues are currently all under investigation.

In Figure 3A the separation is 9+7 and in Figure 3B and B' it is 8+7 with the 16th chromosome at one side. In Figures 3C, D and F and most probably in 3E the segregation is 8+8, which is the commonest distribution. In a sample of 131 segregating nuclei Dr. G. B. Wilson has found 61 separations of 8+8, 52 of 7+9, 12 of 6+10 and 6 of 5+11 or more unequal separations.

There is no plausible working hypothesis from which to calculate expectation but if 16 independent bodies went at random to either side of a cell the frequencies of the groupings expected would be 26, 46, 32, and 27. The data from Allium do not merit further consideration because many better data from other plants are now becoming available.


Comparison with the cases cited in the first section of this paper shows that many of the unusual features found in these reduction-divisions induced in Allium roots resemble normal stages of mitosis or meiosis in other organisms. For instance, movement of chromosomes ends first, instead of with the kinomere leading, is found in Sciara. In the coccids, there is segregation without synapsis. In Aggregata eberthii Belar3 showed that the mitotic chromosomes "split" and separate without condensing to the metaphase state, though not without a spindle. Only in having the combination of segregation without either synapsis, condensation, i.e., tight coiling, or a spindle do the Allium "prophase segregations" transcend the range of normality commonly known. From the present material one gathers the general impression that what is being revealed under the influence of the crude sodium nucleate we are using is the dissociation of various adaptive mechanisms which when superimposed on a basic repulsion inherent in the chromosome itself produce the complicated but normally closely coordinated meiotic process we recognize as "typical" for a flowering plant. Such a subjective impression has, of course, in itself little or no scientific value excepting insofar as it suggests test objects and stimulates researches which may give precise answers to some of the questions the present observations raise. We have undertaken many distinct tests and hope that others will initiate experiments with materials more familiar to them than to us.

Though the segregations might, so far as the evidence herein goes, be the result of random scattering they could still, of course, give segregation of complete genomes in a proportion of cases (e.g., ca 2% of the numerically equal distributions in onion with n=8 and 12.5% in Trillium with n=5). Doubling of these through the formation of second division restitution nuclei could give homozygous diploid cells in an otherwise heterozygous organism. In later papers on other organisms it will be shown that association and segregation is not at random. A way is therefore indicated by which, when we have further knowledge of the process and techniques developed, the plant breeders' work may be greatly accelerated at least with some types of plants (tomato callus seems particularly favorable for the extraction of recessives, by somatic segregation, from a heterozygote).

If somatic reduction occurs in polyploids and a reduced tissue forms in which sporogenous cells thereafter undergo ordinary meiosis, some gametes will be formed which are of, or approach, the ancestral-types. Alternately, gonocyte reduction may be followed by a second meiosis in the gametophyte. Either of these types of double reduction may be the method of origin of progeny resembling ancestral forms, such as occurred in the cases cited above. Various minor applications to plant breeding become possible to envisage, such as the transfer of genes from high to lower polyploids or diploids in cases where doubling of the lower form is difficult, or reduces viability or, if accomplished, gives too complex segregation. Reduction without crossing-over, as evidently occurs in most of these induced divisions, gives other possibilities.

Throughout all the studies so far made of the effects of sodium nucleate, the impression has held that it only enhances the frequency of variations of nuclear and chromosome behavior that occur naturally but with such low frequency that they are rarely detected. This is, of course, not surprising since nothing has been added to the cell that is not normally present. However, in vivo these nucleic acids are highly polymerized. In degenerating tissues they are probably depolymerized as in the Schwarz sodium nucleate. That nucleic acids vary greatly in concentration in different tissues and, as shown particularly clearly by Caspersson, in different parts of the cell during its division cycle is well known and was, of course, the starting point of the chain of reasoning that led to the decision to initiate such experiments. However, until more highly purified materials are available, it must be recognized that the effects of commercial nucleic acid might be due in part to impurities.

The pairing which, following Caspersson's "lead," we hoped to effect has been found so far in very few nuclei. Instead we find evidence to be added to that of Metz and many others that the chromosomes have in themselves a property of repulsion (as well as of attraction) which independently of any extraneous mechanism, such as a spindle, is effective in separating either whole chromosomes or their component chromatids. This has of course been recognized by all who have ever worked on the mechanism of mitosis, but the emphasis is new and the data should lead to new points of view. The repulsion is not attributable simply to a doubled state in the microscopically visible chromosome; there is no evidence of attraction in pairs and repulsion between pairs of paired strands (Darlington14) at the microscopic level (though that is of course possible at the molecular). This is particularly clear in Figures 2C-E, where bipartite chromosomes and unitary chromatids are separating from each other at the same time in single nuclei. Again, while the kinomere obviously has a special role in chromosome mechanics it does not necessarily take the lead in moving, the chromosomes apart from one another — see also Rhoades and Vilkomerson,46 Mets35 and others.

It is yet too early to evaluate the implications of the present data for the elucidation of the mechanisms of mitosis and meiosis in general, but it does seem obvious that they should lead the biophysicist to concentrate his attention on the chromosome itself rather than on other features of the mitotic process such as the spindle, astral rays, etc., which seem clearly to be ancillary. The lack of success which has for so long characterized attempts to elucidate the mechanisms of mitosis (see Schrader47) through the use of models that simulate spindle and astral figures, metaphase plate-polar fields, etc., is not surprising if these are merely subsidiary factors. They are doubtless adaptive and help to ensure the regularity of the process but apparently can be dispensed with even in a highly evolved flowering plant, without necessarily destroying the functionally essential features of the meiotic process.

Ostergren44 states that "the property of being able to induce spindle disturbances (c-mitosis) is present in most organic substances" and also that many substances as well as X-rays and certain genes can induce chromosome stickiness. He shows that ethylene glycol is a particularly effective agent for inducing stickiness, and therefore chromosome mutations, because of its low toxicity. It must here be emphasized that though there are superficial similarities between some of the effects of sodium nucleate and many "mitotic poisons" an extensive examination of the literature has failed to produce evidence that any of the lat­ter produce separation of the chromosome complement into two groups, with equal or nearly equal numbers in each, with anything remotely approaching the frequency found after sodium nucleate (or phosphates). If they do cause somatic reduction and segregation their significance in this regard has apparently been missed. Further, sodium nucleate in moderate dosage does not produce "stickiness" or any grossly abnormal appearance in the chromosomes, nor have we so far any evidence of mutation accompanying somatic segregation in our cytogenetic experiments. Yet again, it is clear from the occurrence of segregation at very early, prophase stages that it s not disturbance of a normally developed spindle that is involved, as it is with colchicine, etc. In fact it is very doubtful if any ordinary spindle mechanism is involved even in the segregations of metaphase-like chromosomes, but a detailed study is needed before any definite statement is made on this issue.

Finally, it must be emphasized that the study here reported is merely the initial one in what must be an extended series (we hope by others as well as ourselves) before the issues it raises can be adjudicated. These issues currently fall into at least five categories: (1) the cytochemical; (2) the cytological including chromosome mechanics and, even if negatively, the spindle; (3) the cytogenetic effects including somatic segregation and possibly its relation to some cases that have been classed as mutation; (4) the cyto-evolutionary, for which wild forms or species already suspected of having resulted from reversed polyploidy are desirable — Tradescantia is the most promising form currently being studied (5) the possible relation of somatic reduction to the origin of sexuality, for which a "primitive" organism will probably be necessary-this as yet has scarcely been considered. Further issues, if not further categories will, of course, arise as the work proceeds. Different experimental materials are needed for elucidation of the different issues. For instance, for (1), the cytochemical studies, micro-organisms or tissues in sterile culture are likely to be best after preliminary studies of plants in culture solutions are completed and significant "leads" obtained from them. It is obvious that a highly organized structure such as a plant root is not expected to be the most favorable for attempts to produce viable modifications of the mitotic process. It is a very convenient test object for first-level experiments and will suffice for the elucidation of many issues. For more thorough-going attempts to induce and control the meiotic process, less organized tissues, perhaps such as plant callus or animal tumor, or possibly microorganisms, are indicated, For (2) organisms with individually distinguishable chromosomes and also with translocations, etc., are being used. For (3) a plant such as the tomato which reproduces from callus that can easily be treated with the reagents, was obviously indicated for the first tests; the Gramineae, which are most important from the plant-breeders' point of view, were expected to be among the most difficult, but by use of a chromosomally aberrant form of wheat some indication has been obtained that the problem may not be quite as difficult as anticipated.24 Two experiments with tail tips of Triturus, kindly furnished by Dr. G. Fankhauser, have given no results.

Until more of the issues are settled, one hesitates either to call these reducing-divisions meiosis or to coin a new term for them. Etymologically, meiosis means only "to lessen." The term "elaxis," from Greek-diminution, has been adopted by Schreiber49 for what he considers "a proportional and discontinuous reduction of nuclear size during ontogenesis" (perhaps this is through a diminution in the degree of polyteny). If it were not already preempted, elaxis would be a very satisfactory term. For the present these natural and induced somatic reduction-divisions will be termed "somatic meiosis" without, however, any implication that they necessarily involve all the, essential features of the gonocyte meiosis of the organism involved.

Evidence is accumulating that somatic segregation and/or reduction occur not too infrequently under "natural" conditions. Figures such as 2D-F and the fact that few haploid nuclei are formed probably indicate the reason it has so generally been overlooked. If a haploid set of bipartite chromosomes is segregated to each side of the cell and they begin immediately to separate into their constituent chromatids particularly as in Figure 2F, or if two distinct haploid nuclei, as in Figure 2A, are formed but the second division is abortive, the diploid chromosome number is at once restored. Unless somatic reduction occurs fairly frequently it is therefore very likely, to be missed by the cytologist unless the organism has grossly heteromorphic pairs of chromosomes. It will, however, be seen by the geneticist as somatic segregation when it occurs in a heterozygote. If the production of homozy­gous diploid tissues could be induced frequently enough, it would have some advantages and few disadvantages for the plant-breeder over the production of haploids.


By growing bulbs of Allium cepa in an aqueous solution of 1-4% sodium nucleate, chromosome segregation and/or reduction of the chromosome number has been induced in root-tip cells. In treatments of freshly harvested bulbs made November, 1947, with the Schwarz sample S.N. 4509 both pairing and segregation of metaphase-like chromosomes was obtained. A second division which separates sister chromatids and resembles that of a normal gonocyte meiosis occurred in a number of cells. More commonly the two divisions overlap and separation of chromatids occurs during the equivalent of gonocyte meiotic anaphase I.

Segregation of long, prophase-like chromosomes also occurs and this is the more frequent type of "somatic meiosis" found in both the onion and other plants in recent experiments with other samples of sodium nucleate. It has also been found in untreated onion bulbs that were flaccid after several months' storage at room temperature.

Some superficial similarities to the action of colchicine, ethylene glycol, etc., occur. The occurrence of "somatic meiosis" (either normally or by induction with substances that are used in the normal metabolism of the organism) raises many issues that require the use of different test-objects for their elucidation. Some of these are outlined. The onion data cannot present a complete picture but they indicate some of the lines along which further studies should proceed and in particular show that in an appreciable number of cells a process occurs that is genetically equivalent to gonocyte meiosis. In this the effects of sodium nucleate differ from those of "mitotic poisons."

Literature Cited

  1. BATESON, W. Jour. Gen. 16: 201-235. 1926.
  2. BATTAGLIA, EMILIO. Nuovo Giornale Bot. Italiano. n.s. 54. 1947.
  3. BELAR, K. Arch. Prot. 53: 312-325. 1926.
  4. BELLING, JOHN. Univ. Cal. Pub. Bot. 14: 379-388. 1929.
  5. BERGER, C. A. Carnegie Inst. Wash. Pub. 496: 209-132. 1938,
  6. —————. Cold Spring Harbor Symp. Quant. Biol. 9: 19-21. 1941.
  7. BRIDGES, C. B. Science 72: 405-6. 1930.
  8. BROWN, META S. Am. Jour. Bot. 34: 384-388. 1947.
  9. CASPERSSON, T. Arch. f. exp. Zellf. 22: 653-656. 1939.
  10. CHRISTOFF, M., and M. A. CHRISTOFF. Genetics 33: 36-42. 1948.
  11. COOPER, K. W. Proc. Nat. Acad. Sci. 21: 109-114. 1941.
  12. —————. Genetics 30: 472-484. 1943.
  13. DARLINGTON, C. D. Proc. Roy. Soc. B. 107: 50-59. 1930.
  14. —————. "Recent Advances in Cytology," 2nd Ed. Philadelphia: Blakiston, 1937.
  15. ————— and L. LACOUR, Jour. Genetics 40: 185-213. 1940.
  16. DUNCAN, P. F. Jour. Heredity (in the press). 1949.
  17. EAST, E. M. Genetics 19: 167-174. 1934.
  18. GATES, P. P. Ann. Bot. 26: 993-1010, 1912.
  19. GRELL, SISTER MARY. Genetics 31: 60-76. 1946.
  20. —————. Genetics 31: 77-94. 1946.
  21. HUGHES-SCHRADER, SALLY. Z. Zellforsch. 6: 509-540. 1927.
  22. HUSKINS, C. LEONARD. Am. Nat. 81: 401-434. 1947.
  23. —————. Nature 161: 80-83. 1948.
  24. —————. Proc. 8th Int. Genetics Cong. (in press). 1948.
  25. ————— and H. B. NEWCOMB, Genetics 26: 101-127. 1941.
  26. ————— and S. G. SMITH. Ann. Bot. 49: 119-150. 1936.
  27. ————— and LOTTI M. STEINITZ. Jour. Heredity 39: 34-43. 1948.
  28. —————. Jour. Heredity 39: 66-77, 1948.
  29. JONES, D. F. Proc. Nat. Acad. Sci. 22: 645-648. 1936.
  30. —————. Genetics 22: 484-522. 1937.
  31. —————. Genetics 29: 420-427. 1944.
  32. KIELLANDER, C. I. Svensk Bot. Tidskr. 35: 321-332. 1941.
  33. KODANI, MASUO. Jour. Heredity 33: 115-133. 1942. 
  34. —————. Proc. Not. Acad. Sci. 34: 131. 1948.
  35. LOVE, P. M. Nature 138: 589-590. 1936
  36. MATHER, K. Nature. 161: 872-874. 1948.
  37. METZ, C. W. Biol. Bull. 43: 369-373. 1922.
  38. —————. Science 63: 190-191. 1926.
  39. —————. Amer. Nat. 86: 623-630. 1942.
  40. MIESCHER, F. "Histochemische und physiologische Arbeiten," F. C. W. Vogel. 1897.
  41. MULLER, H. J. I-IV. Amer. Nat. 50: 193-221, 284-305. 350-66, 421-34. 1916.
  42. NEMEC, B. Jahrb. Wiss. Bot. 39: 645-730. 1904.
  43. NISHIYAMA, 1. Jap. Jour. Genet. 8: 107-124. 1933.
  44. OESTERGREN, G. Hereditas 30: 213-216. 1944.
  45. PETO, F. H. Can. Jour. Res. 13: 301-314. 1935.
  46. RHOADES, M. M., and HILDA VILKOMERSON. Proc. Nat. Acad. Sci. 28: 433-436. 1942.
  47. SCHRADER, F. "Mitosis," New York: Columbia Univ. Press. 1944.
  48. —————. Science 107: 155-159. 1948,
  49. SCHREIBER, G. Reviste Agric. 18: 458-474. 1943.
  50. SPARROW, A. H. M.Sc. Thesis, Univ. Sask. 1938.
  51. STERN, C. Amer. Nat. 73: 95-96. 1939.
  52. STRASBURGER, EDUARD. Jahr. Wiss. Bot. 44: 482-555. 1907.
  53. WATKINS, G. M. Bull. Torr. Club. 62: 133-150. 1935.
  54. WHITTINGHILL, M. Genetics 22: 114-129. 1937.
  55. WILSON, F. B. "The Cell," New York: Macmillan Co. 1928.
  56. WILSON, G. B., and P. BOOTHROYD, Can. Jour. Res. C. 19: 400-412. 1941.
  57. —————. Can. Jour. Res. C. 22: 105-119. 1944.
  58. WINKLER, H. Ber. Deutsch. Bot. Gesel. 28: 116-118. 1910.