Acta Horti Bergiani, Band 17, N:o 9 (1958)
Hip and Seed Formation In Newly Formed Rosa Polyploids
Folke Fagerlind


In the genus Rosa there exist both diploid and polyploid wild types. The basic number is in all cases 7. The diploids must of course be considered to be the more primitive. In a few cases (cf. Darlington & Wylie 1955) different degrees of polyploidy have been observed in individuals belong to one and the same Rosa species. In the majority of cases, however, it is quite impossible for us to decide whether the polyploidy in the genus is of interspecific or intraspecific origin.

With the aim of investigating the manner in which the origin of species and forms has taken place in the genus I have for several years carried on crossing experiments. The diploid Rosa hybrids produced all show the same syndetic properties as the diploid species. In spite of this there is a varying, not infrequently very pronounced degree of pollen abortion. These conditions are possibly to be explained in the light of the assumption that several of the young spore cells have lethality-producing gene combinations, or at least combinations producing lethality when the cell in question has been developed in the milieu presented by the mother bush and the surrounding medium. Certain phenomena, to which I intend to revert on another occasion, indicate that properties in diploid tissue of the mother bush itself exercise a certain influence.

The diploid hybrids do not show any disturbed chromosome distribution during anaphase I. The special preconditions for the formation of restitution nuclei are thus absent. Nor have any such nuclei been found. Nor, again, have any individuals with doubled chromosome number been observed in any of the subsequent generations derived from the diploid hybrids.

In some cases the supplying of the diploid species with pollen from a tetraploid one has resulted in, besides triploids, a single tetraploid. Comparison between the latter, the parents and the triploid siblings favours the assumption that the mother-plant has supplied two of the four genomes. These tetraploids do not, or at least not markedly, differ from the crossings between different tetraploid species in respect of their hip and seed forming capacity.

In seed progeny from the triploid hybrids—this progeny is in relation to the wealth of flowers extremely inconsiderable—tetraploid individuals are predominant, or possibly individuals with a chromosome number oscillating closely about the tetraploid number. These secondarily arising tetraploids or "approximate tetraploids" show varying fertility qualities. They do, however, yield hips and seed, generally a rather high frequency of these, in fact.

In order to be able through comparative studies to trace the history of the tetraploids' origin I have on several occasions tried to bring about a doubling of the chromosome number with the help of colchicine. A first report on these experiments was published more than 10 years ago (Fagerlind 1945). The resulting doublings have in the course of the years shown a marked increase in growth. Many of them have flowered. As to the hip and seed formation, they deviate markedly from other Rosa polyploids, including those mentioned above, and from what is regarded as the normal. This is shown in greater detail in Chapters 2 and 4. The background to these reactions is discussed in Chap. 5.

Chapter 1. Material

The method used in the treatment with colchicine was to begin with the same as that described in the aforementioned work. With a little steel pen one drop of colchicine solution was applied between the still closed cotyledons of the young sprout. Later, however, a modification of this method was employed. For twenty-four hours the cotyledons and the tip of the shoot were allowed to grow in a rather narrow test-tube filled with 0.5 per cent colchicine solution and mounted upside down over the seedling.

The "doublings" formed show, as compared with their origin, distinct gigas properties. These are revealed in an increased ratio between the breadth and length of the different organs, in a stouter general structure, in the size of the cells and of the organs, in a deeper green colour, etc. In contradistinction to what was stated in my first work, the doublings show, from several other viewpoints, changed physiological properties. They are, for example, considerably more sensitive to winter cold and to drought in summer than the types from which they originated. Many show increased susceptibility to rust. The ability to survive in competition with ordinary garden weeds is often much reduced. Thus many of the doublings formed have frequently—at a rather early stage of development—withered in the field. This notwithstanding, I have for a longer or shorter period of years been able to study the following doublings.

  1. Rosa rugosa—tetraploids from diploids—several individuals of varying ages; in the cultures there remain 9 different specimens from the year 1942, 4 from the year 1947, 2 from 1948 and 2 from 1950.
  2. Rosa rugosa x Willmottiae—tetraploid from diploid—one older, now defunct individual.
  3. Rosa rugosa x Prattii—tetraploids from diploids—one older and two younger, annually flowering individuals remain in the cultivation series.
  4. Rosa rugosa x Beggeriana—tetraploids from diploids—three older individuals were included in A/B KABIS' cultivation series, now abandoned.
  5. Rosa rugosa x "Bakeri"—tetraploids from diploids—four older richly flowering individuals still remain.
  6. Rosa rugosa x multiflora—tetraploid from diploid—one older individual, which languished on for a long time, died without having bloomed.
  7. Rosa rugosa x villosa—tetraploids from diploids—several individuals, of which two older, richly flower-producing specimens still survive.
  8. Rosa rugosa x pomifera—tetraploids from diploids—several individuals, of which one older, richly flowering specimen survives.
  9. Rosa rugosa x rubiginosa—tetraploids from diploids—three individuals, of which one older, richly flowering specimen survives.
  10. Rosa rugosa x Afzeliana—tetraploid from diploid—one still surviving, flower-producing individual.
  11. Rosa multiflora—tetraploids from diploids—three individuals, badly frozen each year. Earlier, one of them flowered rather richly, but not at all in recent years. All these specimens are now languishing considerably.
  12. Rosa Sweginzowii—16-ploid from 8-ploid—one older individual which flowered feebly on one occasion, very frost-bitten every year.
  13. Rosa canina coll.—10-ploid from 5-ploid—five-year old individual which has not yet flowered. Rather considerable annual frost-bite.
When the results of the colchicinization experiments have been positive, chimaeras formed of unchanged and "doubled" tissue have not infrequently arisen. The positive result must of course be conceived always primarily to consist of such a mosaic-individual. Here, however, there generally occurs a rapidly succeeding "splitting process", which results in the putting forth of "pure" shoots. Different shoots of this kind, both undoubled and doubled, may appear in one and the same individual. The background of the "splitting" must be that the one or the other kind of cell is eliminated in the competition in the growing meristem.

The undoubled and doubled shoots from one and the same individual can be cultivated as separate specimens by layering. If this separation is not effected in time, the weaker or the less dominating partner is generally rather soon eliminated. By suitable pruning one can, at least in the case of R. rugosa and R. rugosa x "Bakeri", retain the occurrence of pure diploid and tetraploid shoots on one and the same root year after year.

In a number of periclinal chimaeras, however, the chimaera character is obstinately retained year after year. I have a diploid-tetraploid periclinal chimera of R. rugosa from the year 1942 which has proved to be conservative, inasmuch as no single pure diploid or tetraploid shoot has ever been produced from it.

One may possibly a priori be inclined to suspect that the phenomena described in the following account to be ascribed to elements of diploid layers and sectors of different strength in the shoots and bushes characterized as tetraploid doublings. I wish, however, at the very outset to emphasize it as my conviction that these shoots (and bushes) are really pure. The chimera shoots have completely different properties.

Chapter 2.Hip-forming capacity and hip quality in the Rosa rugosa doublings.

Before the summer of 1954 hips were found on only one of the doublings produced. One year one of the multiflora individuals bore, in relation to the quantity of flowers, a very modest number of hips. These were feebly developed. They contained only a few more or less developed achenes. The seeds were without germinating capacity.

The absence of hips before the summer of 1954 in all the other doublings is not due to poor flowering on the part of the latter. Many bushes had a prolific production of flowers long before the year mentioned.

In the year 1954 hips were produced by some of the rugosa doublings. These all belonged to the bushes of the year 1942. Other plants from that year, all the younger rugosa doublings and all the rest of the doubling material showed, also in 1954, a total absence of hips despite the fact that many of the specimens had flowered abundantly or very abundantly. The hip-producing rugosa doublings yielded only small quantities of hips (pl. I, fig. 4). Most of the hips were poorly developed. They contained a small number of apparently normal achenes, rather a lot of only partially developed achenes and very many not so well-developed pistils. When sown, the seeds showed extremely poor germinating qualities. A collection of young plants was, however, nevertheless obtained. These all appear to be tetraploid. (In R. rugosa it is, to judge from the available evidence, easy to distinguish diploid and tetraploid young plants from each other.)

In the summer of 1955 none of the doublings—not even those which had borne fruit in 1954—formed any hips.

In the summer of 1956 the doublings which had been hip-producing in 1954 once more showed the capacity to form hips. The same capacity, however, was also found in all the other rugosa doublings from the year 1942. All rugosa doublings representing subsequent years and all the doubled rugosa hybrids were still, despite a rich flowering, without any tendency to form hips.

In the non-hip-producing doublings many of the undeveloped rudimentary hips persist for a strikingly long time (pl. I, fig. 2) Thus for periods the plants acquire an appearance which sharply distinguishes them from the diploids. Especially striking does the picture become when it is a matter of a bush composed of diploid and doubled branches. The diploid shoots then bear well-developed, deep red hips in more or less rich positions; the "doubled" shoots bear collections with only brown, dried-up remnants of flowers (cf. pl. I, fig. 3).

In 1956 the hip-producing rugosa doublings showed an extremely variable hip quality. Only a few of the mature hips had an external appearance that would justify their designation as normal. A few more were represented by small green dwarfs or dwarfs faintly touched with red. These never acquired the soft, juicy consistency of the normal hips. This notwithstanding, they are not identical with the dried-up, withered flower-remnants.

Besides hips belonging to the normal and dwarf classes, there were also those representing a richly differentiated chain of intermediate types. In the dwarf group and the intermediate series there was an abundance of hips that showed a more or less strongly asymmetrical development. In the intermediate class the hip as a whole showed a larger or smaller part, or several such parts, of frequently much delayed maturity or total incapacity to acquire the red colour and consistency characteristic for the rugosa hips. In sectors and parts there was a frequent tendency to blackening and hardening. It was often possible to observe parts that had been strongly attacked by infections. Various degrees of wrinkling of the hip also occurred. The varying "external" hip quality is shown in Plate II.

If sections are made of hips of different quality (pl. II), it proves that a greater or smaller number of ovaries have always started their development to achenes, but that this development has arrived at different stages in different cases. The most extreme dwarf hips contain only rather soft, more or less dwarf-like achenes, which are easily cut through during the operation. These "soft achenes" were in all cases without vital contents.

To judge from the external appearance at least, there are, however, normal achenes far down the intermediate chain. The quality of the seeds and their relative number are discussed in greater detail in Chap. 3.

At the time when the well-developed hips of the doublings have attained maturity one finds "fruit clusters" of very varying appearance among the latter. Some contain only withered flower-remnants, others a small number of normal or defective hips. Withered remnants and a varying number of normally or abnormally formed hips make up other collections: etc. A number of withered lowers which had never shown any tendency to develop into hips, as well as of whole partial inflorescences retaining a varying number of flower-remnants, have, however, already fallen. Remanent scars often bear witness to their earlier existence. But it would seem that at a late stage it is difficult or impossible to observe remanent scars from all the flowers present earlier.

The frequency with which hips of different types and flowers incapable of further development appeared in the summer of 1956 in the various rugosa doublings from the year 1942 varied very considerably. This was evident even to superficial observation. To throw further light on the variation, four biotypes were selected for closer analysis. The types selected were the strongest hip producing biotype—4R6 (this is represented by three clonally formed individuals) —, two of the weakest hip-forming biotypes—4R5 and 4R10 —, and one—4R7—with a more intermediate hip-forming capacity. The hip production in these biotypes is shown in Table I. The table gives the different "types of fruit cluster" and their frequency. The types of cluster are indicated with the help of a series of figures. The last figure of the series indicates the sum of the number of remanent withered flowers not developed into hips, and distinguishable scars from fallen or partial inflorescences. The first figure represents the number of well-developed and fairly well-developed hips. The second figure gives the number of hips of poorer quality. The sole criterion of quality applied here has been the external appearance. There is only in a few cases any difficulty in dividing the hips into the two classes. The last figure does of course not constitute any real measure of the number of non-hip-producing flowers. What it does constitute is a kind of minimum value; the figure is perhaps mostly too low. As pointed out above, it is impossible to decide whether an observed scar is from a solitary flower or from a partial inflorescence that has fallen. There is also the risk that many scars are overlooked during the inspection.

Table I. Hip formation in the year 1956 in 4 different biotypes of doubled Rosa rugosa from 1942.
(Two relatively richly and two relatively poorly hip-forming individuals have been chosen for analysis.)

Quality of fruit
cluster (cf. the text
on pp 233, 235)
Number of different fruit clusters in the different biotypes and individuals
Biotype 4R6
individual 1

Biotype 4R6
individual 2

Biotype 4R6
individual 3
0-0-1 29 3 2 18 1  
0-0-2 13 1 2 10 3 1
0-0-3 3     16 12 1
0-0-4 6     15 14 8
0-0-5 4     4 3 3
0-0-6 4     7 1  
0-0-7       2 2  
0-0-8 2       2  
0-1-0 4     1   1
0-1-1 3     1   1
0-1-2 5     8   1
0-1-3       5 2 2
0-1-4 2       1  
0-1-5       1    
0-1-6       1    
0-2-1 3     1    
0-2-2 1     2    
0-3-0 1          
0-3-1       1    
0-3-4 2          
0-4-1 1          
1-0-0 3 2 3 1    
1-0-1 7 2       1
1-0-2 1 3 7 2    
1-0-3   1 5      
1-0-4 1   1      
1-0-5     1      
1-1-0 3 1 1      
1-1-1 3 1        
1-1-2 1 1 2      
1-1-9 1          
1-2-0 1 1 1      
1-2-2 2 1        
1-3-1   1        
2-0-0 3          
2-0-1 2          
2-0-2 1     2    
2-0-3       1    
2-1-0 3   1      
2-2-0 1     1    
2-2-1   1        
2-3-4     1      
3-1-0     1      
4-0-2 1          

Table II. Hip-forming capacity in the "individual complex" Å 29:12
composed of diploid and tetraploid (4R10) parts.

Quality of
fruit cluster
Number of fruit clusters
Biotype Å 29:12
— 2R
Biotype Å 29:12
— 4R (= 4R10)
0-0-2 0 1
0-0-3 0 1
0-0-4 0 8
0-0-5 0 3
0-1-0 0 1
0-1-1 0 1
0-1-2 0 1
0-1-3 0 2
1-0-0 1 0
1-0-1 0 1
2-0-0 2 0
3-0-0 2 0
4-0-0 3 0
5-0-0 1 0

Table III. Hip-forming capacity in different rugosa biotypes

Biotypes Primary figures
(cf. Tables I-II)
percentual values
4R6— individual 1 . . . 51-54-220 16-17-67
individual 2 . . . 16-12-23 32-24-45
individual 3 . . . 28-10-52 31-11-58
Summary . . . 95-76-295 20-17-63
4R7 11-28-282 3-9-88
4R5 0-3-160 0-2-98
Å 29:12; 4n-sector ( = 4R10) 1-5-69 1-7-92
Å 29:12; 2n-sector 28-0-0 100-0-0
4n-individuals from later year-classes   0-0-100
2n-individual from later year classes and 2n-sectors
in some of the last-mentioned doubled individuals
  very high-0-very low

It has already been mentioned that a number of my experimental plants bear both doubled and undoubled shoot-systems. Among these are also rugosa individuals from 1942. The comparison between the hip-producing capacity in 1956 in the tetraploid and diploid branches of these is very eloquent. The figures are given in Table II.

If the figures in Table I are added up, one obtains the series of figures given in Table III as an expression for the capacity of the different biotypes and individuals to form hips of different quality. These data are supplemented with information concerning the conditions obtained for doubled rugosa belonging to the younger year-classes and the approximate normal figures for R. rugosa. (There are, however, certainly diploid rugosa individuals which diverge markedly from the normal. I can mention, for example, a couple of individuals in which the hip-forming capacity is restricted to the flowers developing relatively late in the vegetation season. Other flowers show defective development of their archespores, anthers and pistils. They form, consequently, no hips.)

The figures in Tables I and III show that the rugosa doublings from 1942 are extremely labile as regards their hip-forming capacity. The values do not contradict, though neither do they prove, the assumption that different biotypes show this lability to different extents. A comparison between the three members of the 4R6-clone shows, however, that even individuals representing one and the same biotype appear to give different reactions. The study of one and the same bush—thus of one individual which has not been divided up clonally—does, moreover, give the general impression that different parts and branches often react differently. It would seem that the hip-primordias produced by the particularly strong-growing shoots do as a rule give rise to hips with considerably higher frequency than those on weaker growing shoots. This is perhaps the background to the fact that the hip-primordias from "1-flower clusters", Table I (though these "1-flower clusters" probably in reality often held more than one flower cf. pp. 233, 235), produce hips with lower frequency than those from "2- and 3-flower clusters". One also gets the impression that flowers developed during different parts of the vegetation season have different hip-forming capacities. It is not possible, however—at least not for the present—to confirm this general impression by figures.

Chapter 3. The formation of achenes in the hips of the doublings

In diploid Rosa rugosa the mature hips contain a large number of fully developed achenes and a small number of pistils not developed into fruits. The number of the latter generally varies between 3 and 7. In exceptional cases, however, one finds hips with a considerably larger number of rudiments in the form of "down". The contents of three arbitrarily chosen hips and a hip chosen for its extremely large quantities of down are shown in Plate III. In these cases the down undoubtedly represents pistils that have remained unpollinated or that have remained unfertilized possibly because they themselves have not developed viable embryo sacs—cf. p. 242. In an earlier work (Fagerlind 1948) it has been demonstrated that in Rosa fertilization is, at least as a rule, followed by development of (seed-containing or empty) achenes.

The proportion between the number of fully developed achenes and the quantity of down is markedly altered to the advantage of the latter in the hip-producing rugosa doublings. The variation is also much greater. Further, poorly developed achenes not infrequently occur; evidently pistils that have started, but sooner or later stopped in their development. Such intermediate types are extremely rare in diploid R. rugosa and, altogether, in pure species of the Rosa genus. Plates III and IV show, among other things, the achenes and down from a number of normally or fairly normally developed hips in 4R6 and 4R7.

As already mentioned, one does not find, even judging solely from the outer appearance, any normal achenes, in the extremely dwarf hips, but only so-called "soft achenes". They consist of more or less strongly developed ovaries that have never acquired the normal consistency. Unlike the normal achenes, they may easily be squashed with the handle of a scalpel or may be sectioned without difficulty with a razor blade. Real "soft achenes" are as a rule absent already in the slightly better developed hips.

(The apparently well-developed achenes of the doublings are extraordinarily large as compared with the achenes of diploid R. rugosa. The order of magnitude appears to be such as to exceed what one would expect as a result of the doubling of the chromosome number. Presumably another factor also enters here. Even in one and the same diploid rugosa individual the size of the achenes is variable—cf. the 2n-material in Plate III. The giant hips of the strongest shoots have as a rule larger achenes than the smaller hips of the more feebly growing shoots. Hips of approximately equal size with different numbers of pistils or with a different proportion of down show differences in the size of the achenes. I am therefore inclined to connect the difference between the size of the seeds in the tetraploid hips in Plates III and IV also with the different relative quantity of down. The size of the seeds appears to a certain extent to be a function of the "conditions of competition" in the hip. In the doublings there are relatively large amounts of down. The considerable size of the achenes in the doublings is therefore perhaps in part due to the existing polyploidy and in part to the fact that relatively few developing achenes are competing for the available supplies of nutrition and space.)

To give an idea of the connection between the hip quality, the number of pistils, the extent of achene formation, etc., the accompanying correlation diagrams (textfig. 1) have been drawn for 4R6. A small number of normal hips (a) and a number of more poorly equipped ones (b) have been registered. The hips with only "soft achenes" are completely excluded. The diagrams have been drawn on the basis of an all too limited material. They show, however, that the frequency with which the pistils have matured to achenes oscillates about the 25 per cent line. The 50 per cent line is not reached. The lowest value is somewhere between 5 and 10 per cent. The hips which have been designated as the more normal evidently contain more (developed + undeveloped) pistils than do the more abnormal hips. There does not, however, seem to be any distinct difference in the capacity to produce seeds between the pistils belonging to the two hip categories studied. Nor does the diagram show any connection between the capacity referred to and the number of pistils in the hips.

Hip primordias produced on more strongly growing rugosa shoots contain as a rule more pistils than do those on weaker shoots. This circumstance, together with the above-mentioned between the quality of the hip and its number of pistils, once more leads us to the conclusion that the stronger shoots of the doublings have better chances of development than the hips on their weaker shoots.

Chapter 4. Seed quality of the doublings

With the intention of determining the quality of the content in the achenes harvested from the doublings, all the latter from the biotypes 4R5 and 4R10 have been photographed by X-rays. The achenes from a number of 4R6 and 4R2 hips of different quality were treated in the same way. The poorest hip quality, viz., the dwarf hip, which had only soft achenes, was here completely excluded. To get a basis for comparison a total of 20 achenes taken from 20 different diploid rugosa individuals were also photographed. Some of these derived from diploid branches of the previously mentioned complex individuals composed of diploid and tetraploid sectors. Inter alia, the diploid part of Å 29:13 (its tetraploid part is 4R5) was represented.

From earlier experience I know that the achenes from diploid R. rugosa almost always contain—to judge from the morphology—well-developed seeds. Nor, of the 20 X-rayed achenes, did any show more marked defects (pl. V, figs. 1-2). Slight variations in the degree of blackening do, however, perhaps some minor differences in quality. I rather suspect that the four seeds with the lowest degree of blackening are not germinative.

The material that was X-rayed was divided according to the hip quality into three groups: (1) normal and fairly normal, (2) intermediate, and (3) defective hips. As has already been mentioned, this does not cover the whole range of variation. The most defective hips are not represented. The achene mass and the content of the achenes from four 4R6 hips of this category 1, two of those of category 2 and one of category 3 are reproduced in Plate V, figs. 3-6, Plate VI and Plate VII, fig. 1. The material shows that the content of the achenes varies greatly. I restrict myself, however, to a division of the seeds into two categories. The designation "+/- normal seed" is used when the embryo shows a rather marked degree of blackening, a relatively normal degree of filling out in relation to the cavity inside the achene and a fairly even surface contour. And as belonging to the same category, where it is a matter of determining frequencies, I count all cases in which the embryo is absent but in which the achene contains a well-developed Megastigmus larva. From earlier experience I am almost certain that the presence of well-developed larvae imply that the development of embryo and endosperm has been directed into normal channels. More poorly developed larvae indicate that embryo, endosperm and maternal seed tissue have at an early stage shown lethal qualities. The achene-content referred to the category "undeveloped seed" is for the most part represented by wrinkled, veil-like tissue remains. After boiling up these in ammonia it is possible to observe the presence of embryos and endosperms that have been at an early stage arrested in their development and are now dead and more or less abnormally formed. Thus fertilization has taken place also in these cases. The products of fertilization have, however, soon succumbed. Thus we have here a parallel to the conditions earlier observed to be a common effect of the pollination of a Rosa individual by an alien species (Fagerlind 1948).

It is quite certain that the category "undeveloped seed" is entirely without germinative capacity. And it is probably almost equally certain that the category "+/- normal seed" contains not only germinative units. A division of the last fraction would of course be interesting. As Rosa seeds of one and the same kind germinate after resting periods of very different duration, this division would entail much time-consuming work. Division on the above-adduced lines is at least for the present not feasible.

The X-ray pictures show a certain connection between the seed-quality and the structure of the achene-wall. The occurrence of a more marked blackening of the inner-most layer of the wall runs parallel with relatively good seed qualities. The size of the full achenes is on an average greater than that of the empty ones. A division of the categories based on these conditions is, however, not possible.

Without doubt the normal achenes with undeveloped seeds (= the "empty achenes) constitute a kind of intermediate category. They form a connecting link between the normal ones with developed seeds and the so-called soft achenes. The content of the latter has presumably been arrested in its development at an earlier stage than has been the case with the empty normal achenes.

The X-ray analyses show that the examined biotypes with a particularly poor hip-forming capacity and poor hip quality—the biotypes 4R5 and 4R10—have also the poorest achene-content (pl. VII, figs. 2-6). In the summer of 1956 there was in these not a single achene containing a developed seed. The achene-wall, moreover, is in all these cases poorly developed.

The X-ray photographs illustrate that in 4R2 there were both filled and empty achenes. The frequency figures in their connection with the hip quality are shown in the correlation diagrams in Text-figs. 2 and 3. In the biotypes studied the number of developed achenes per hip is greater in the class +/- normal hips than in the class abnormal hips. The third type occupies an intermediate position. The abnormal hips show a rather distinct correlation between the number of developed achenes per hip and the vitality of the seed. This relation is diffuse or absent in the other diagram sections.

Chapter 5. The background to the sterility or the poor fertility of the doublings discussed.

A. The sterility or the poor fertility of the above-discussed doublings is not conditioned by any inherent weakness already existing in the same measure in the corresponding diploid original types and inherited from these. This assumption is supported by the fact that the attributes referred to are found in all of the doublings produced, and conclusive proof is afforded by a comparison between diploids and doublings derived from one and the same colchicinized young plant. The former show the same good hip and seed forming capacity as do ordinary rugosa diploids; the doublings, on the other hand, are completely sterile or, as regards rugosa of the year 1942, show poor fertility (cf. Tables I-III).

(In an earlier paper—Fagerlind, 1945—I have pointed out that the observed partial asyndesis in doubled R. multiflora might be due to asyndesis in the original diploid individual. There still seems to be some justification for this suggestion. However, in the light of the circumstances and conditions illustrated in the following it is subject to certain reservation.)

B. The pollen mass in doubled R. rugosa is in all cases more poorly developed than in the original diploid plant. This is not very surprising as regards an intraspecific polyploid. What is more remarkable is the fact that the relation is the same in all rugosa hybrids where the material permits of comparison between doublings and original hybrids. The diploid hybrid has both normal and abortive pollen—different hybrid types and different individuals of the same type show the two kinds of pollen in different frequencies; corresponding doublings have in all cases defective pollen in considerably higher frequency. In some cases the comparison has been made between diploids and tetraploids deriving from one and the same colchicinized young plant.

The course of the meiosis in the diploids and their doublings will be described when the preliminary studies have been completed.

On different occasions pollen from the doublings has been used in crossing experiments. These experiments gave no positive results in the first years, i.e. the first years in which the doublings flowered at all. Later—though also before 1954, i.e. the year in which the rugosa doublings first produced hips—positive results were, however, obtained. (I now have in cultivation a small number of triploid R. rugosa.) The positive results show that the lack of hip-forming capacity or the poor hip production and the poor hip quality are not due to lack of vital pollen in the types in question.

C. Diploid R. rugosa is self-incompatible (Fagerlind 1944, 1948). Pollen from diploids transferred to tetraploids and polyploids—here I am disregarding entirely the section Caninae—does not as a rule give hips. (Fagerlind 1944, 1948). One might therefore a priori be inclined to assume that a complete absence of hips in an isolated rugosa doubling or a doubled rugosa hybrid growing among its diploid kin is due to the interaction of the two above-mentioned phenomena. In my cultures, however, several doublings are quite close to each other; the distance from them to bushes representing different tetraploid wild types is moreover quite inconsiderable. This eliminates the possibility of explaining the sterility or the poor fertility of the doublings as a consequence of the interaction of the two factors referred to.

D. In the course of my earlier studies of compatibilities and incompatibilities in Rosa I have got the impression that fertilization of an embryo-sac leads to the surrounding ovary growing into a more or less fully developed achene and to hip formation independently of the vitality of the fertilization products. This, however, is what determines whether the achene becomes filled or empty. In the light of these conditions I consider it probable that the more poorly developed pistils (=the downs), which alone fill the quite rudimentary hips and are found in large numbers among the achenes in all the hips harvested from the doublings, were never fertilized.

If hip-primordia belonging to young flowers of the doublings are cut open, one often finds that some, many, or all pistils have poorly developed ovaries. Several of them—but not all—are without embryo sacs; they are frequently even without dead remains of sacs. The large number of pistils that remain unfertilized in the doublings must be due entirely or in part to these circumstances.

E. Preliminary studies on younger pistils have shown that the existing lack of embryo sacs is not solely, and perhaps not even chiefly, due to the fact that the meiosis of the doubling has led to the formation of spore nuclei with lethal chromosome or gene combinations. In many of the macroarchespores the meiotic process does not come about. The archespore cells then show no sign whatever of even having begun the meiotic development. Instead, the cells are abnormally stretched and remain undivided. Thus we have here in the sporophyte itself a "weakness" contributing to the reduction of fertility, a weakness which did not exist or was not manifested as strongly and as often in the undoubled primary type.

Archespores without meiotic tendencies and those which develop more or less normally occur in one and the same individual and even in one and the same hip primordium. It is therefore scarcely possible to relate the differences in the development of the archespores either to different external milieus or to different gene-equipments. In these circumstances the only remaining explanation seems to me to reside in differences in the "internal milieus". These are probably determined by the interplay between the external milieu and "reciprocal competition" between the different organs and systems of organs. In the concept "reciprocal competition" I include the competition of the different sporophyte sections—branches, shoots, flower-clusters, pistils, archespore cells, etc.—for the available "supplies", as well as the effect of their reciprocal exchange of substances. Thus this term includes also the effect exercised upon the surrounding organs and systems of organs by development-furthering and (or) development-inhibiting substances with or without hormonal character formed in a certain organ or system of organs. If the conclusion is correct, the plants must—before doubling, in contradistinction to after the same—have possessed such general physiological properties that the above-mentioned competition factors did not in normal cases have injurious results.

The dead remains of more or less well-developed embryo-sacs found in many of the more or less markedly withered pistils belonging to the young flowers of the doublings may derive from the chromosome and gene combination of the gametophytes themselves. One cannot, however, exclude the possibility that in these cases combinations are not, or at least not always, lethal in their tendency. The lethality in the case of the embryo-sacs may have a background similar to that of the incapacity for normal development of the archespores.

F. Among the hip-producing rugosa doublings there were, besides undeveloped pistils and normal achenes, also those with dead fertilization products. Here we had so-called soft achenes as well as achenes with normally or almost normally constructed walls. Among the more or less abundantly hip-producing biotypes the frequencies of the different achene types in the hip fluctuated extremely (Tables I and III). In some extreme cases there were only so called soft achenes. In others there were, besides the undeveloped pistils, only the better class of empty achenes. A third extreme class consisted of hips in which nearly all the achenes had on the whole a satisfactory content.

Different individuals from a small clone showed different frequencies. All these circumstances—together with the general impression that the hips with the best content correspond to the hip primordia that were strongest from the outset and are bound to the more strongly growing shoots—lead to the following conclusion:

The variation in the vitality and lethality of the products of ferilization cannot solely be the consequence of their possessing different gene and chromosome combinations. The decisive variable must be looked for elsewhere. If not by chromosomes and genes, it must be conditioned by external influences. Here these should be taken to include not only the external milieu in the strict sense of the term, but also influences exercised by the surrounding masses of tissue, whether these belong to the mother sporophyte itself or to other products of fertilization. Thus, as in connection with the circumstances discussed above concerning the non-hip-producing flowers, I consider it extremely probable that the varying vitality of the products of fertilization in different hips and in different pistils belonging to one and the same hip is in a high degree conditioned by the existing "internal milieu".

This conclusion does not imply a denial of the possibility that the lethality of the spores, gametophytes and fertilization products of the doublings may be in part due to the fact that some of the gene and chromosome combinations inherent in them have an in et per se more "injurious" effect. Rather do I consider it certain that such an influence is also a contributory factor. The circumstance that no correlation has been shown between, on the one hand, the hip quality and the hips' absolute number of pistils, and, on the other hand, the extent to which the pistils have been transformed to (full + empty) achenes, may perhaps be due to the fact that a rather large number of the macrospores and embryo-sacs are in themselves doomed.

G. If the conclusion I have just declared myself inclined to draw is correct, this implies that the doubling of the chromosome number has transformed a more advantageous physiological system to one that is from certain points of view less advantageous. Thus in contradistinction to the corresponding doublings, the original diploid types would according to this theory have such general physiological properties that in the normal cases the variations in the "internal milieu" would not suffice to bring about the injurious effects in question to any more noticeable extent. It should, however, be observed that at least some of these effects have been noted also in the diploids where the external milieu has been more extreme. Diploid R. rugosa that has suffered from rather extreme lack of water shows, even in relation to the number of flowers existing in this case, impaired hip-forming capacity and presence of hips of different quality and different degree of fullness. Soft achenes are also present. Thus in this case the conditions of the milieu seem to have imposed upon the individual such general physiological properties that the internal variations in milieu are able in several ways to influence the final fertility.

In the abnormal diploid rugosa biotype or group of biotypes already mentioned on p 236 (Chap. 2) hips are formed exclusively in the flower-clusters formed relatively late in the growing season. Other flowers which in contradistinction to the hip-producing ones possess dwarf-like, pollen-free, compact anthers, never yield hips. In the later developed clusters there are fertile as well as sterile flowers. If different clusters are compared, one finds fertile and sterile flowers that must have set and developed at approximately coincidental periods. Thus we find here in the diploid a gene-conditioned disturbance of the general physiology on account of which the internal milieu-variation gives a visible variation in fertility and sterility during the later, but not the earlier, part of the season.

In the course of experiments with species crossing in which the pollen-receiving partner has chiefly been R. rugosa it has been observed that hip primordia in one and the same individual give different reactions after one and the same treatment (Fagerlind 1948). The one flower gives in normal way filled hips, in the other the hip germ shows no tendencies to develop further, although the pollen used has been the same and in all cases applied with a large surplus and in about the same quantity. The failure to form hips here seems to be due to the fact that there has been no fertilization. The frequencies of the two classes are often different on different branches. Flowers from the weaker shoots and the youngest ones in the multiflorous inflorescences of the stronger growing rugosa shoots show a lower hip-forming frequency than the flowers placed otherwise. One thus gets the impression that on account of the internal milieu-variation the pistils in some hips offer pollen and pollen-tubes optimum conditions, while others do not offer suitable conditions for germination. I have earlier in this connection spoken of the different types of shoot showing different degrees of "sensitiveness" (Fagerlind 1948).

That extreme conditions in the external milieu allow the internal milieu-variations to become visible in the form of reduced fertility is to all appearances really a general phenomenon. If, for instance, one cultivates diploid tomato plants under bad conditions, many flower-buds, though not always all, stop developing more or less early; the same applies to the fertilized pistils and their products of fertilization. That the influence of the internal milieu-variation upon fertility appears when the "basic disturbance" is exercised by special genes is also shown by, inter alia, the genus Galium. In several species (Fagerlind 1937)—diploids and polyploids — there are individuals with extraordinarily variable fertility conditions. The variation implies firstly a variable development of the archespore—this develops in a meiotic direction or refuses stubbornly to develop further—and secondly a normal meiosis or one disturbed in different degrees. Different shoots on the same plant, different parts of an inflorescence, different flowers in one and the same part of a flower cluster, different stamina in the same flower, etc., may react differently.

In certain respects the importance of the internal milieu for the formation of flowers and seed appears as something extremely common, even where the external milieu may be regarded as optimal. The factor is here so universal that perhaps precisely for this reason it may be overlooked. I am here thinking of the common phenomenon that buds in the apical parts of the inflorescences often stop their development prematurely, and of the numerous cases (e.g. fruit-trees) in which autonomous thinning of the bulk of fertilized pistils takes place sooner or later. In both these cases there is an elimination of units which have started their growth at the same time and in part developed parallel with other units which are not being eliminated. The differences in behaviour are due neither to gene-differences nor to different external milieus. They must be the result of a kind of internal competition, the effect of the changing internal milieu.

(The fact that "homologous organs" having different internal milieus show different reaction-norms undoubtedly constitutes a "weaker parallel" to the ordinary determination conditions. The difference between the two categories is very indistinct in the cases in which the inflorescences regularly contain a varying or a more constant number of apical buds incapable of development, or in which certain flowers are developed differently from the others. In the last-mentioned case—as in many others—the boundary line between external milieu and internal milieu becomes extremely diffuse.)

H. If my tentative conclusion is correct, this implies that the doubling of the chromosome number in R. rugosa and its hybrids transforms an advantageous, more insensitive physiological system into what is from certain viewpoints a less advantageous and often more labile system. This appears in et per se to be not unlikely. It is now a well-known fact that doubling of the chromosome number brings about rather great physiological changes. There are considerable changes in the relative amounts of the substances involved. Amongst other things, there is a relative increase in the amount of water, with consequent changes in suction power and osmotic properties. The doubling of the chromosome number should therefore be able to change the sensitiveness of the different systems of organs to internal milieu-variations, just as changes in the external milieu or the introduction in the system of sensitivity-increasing genes.

Does other material than Rosa biotypes studied by me show that doubling of the chromosome number influences the sensitiveness of the system to changes in the internal milieu?

Most descriptions of experimentally produced doublings give no information concerning impairment of fertility as a direct consequence of the general physiological state of the new type. Instead, any disturbances of fertility that may have occurred have been explained as a consequence of the meiotic complications. Referring to the observations of a number of researchers, however, Müntzing (1936) and Fagerlind (1937) have asserted with considerable emphasis that the low fertility often found in the newly formed polyploids is not solely, and perhaps not even chiefly, due to the disturbances of the reduction division, but to the fact that the sporophyte possesses a general physiology such that it exercises an injurious effect upon the young spore cells, or upon their mother cells. Müntzing also believes that the strikingly pronounced apical sterility of the inflorescences in rye-wheat produced through doubling of the chromosome number may be a direct consequence of the high-polyploid state. At least relatively newly formed doublings of tomato show greater lability of fertility and sterility under optimum cultivation conditions than do the diploids (Schlösser 1934). At least relatively newly formed Bryum caespiticium doublings, produced and studied by Wettstein (1937), showed a markedly restricted capacity to form and develop sporophytes as compared with the undoubled original plant.

Still more striking are the parallel cases where the doubled original plant is itself a doubling of relatively young age (Gottschalk 1956). Here we have disturbances and lability not only during the forming and development of the reproductive organs, but also in connection with the building up of the vegetative systems.

The above brief survey gives the following evidences: Doubling of the chromosome number creates a tendency to disturb the earlier developmental equilibrium. However, the strength of this tendency varies in different cases. The irritability of the more insensitive types appears, however, if a further doubling is carried out.

I. The fact that the lability appearing after doubling of the chromosome number finds such different expressions in a number of different cases—disturbance of the meiosis (?), lethality of the products of fertilization, etc.—shows that it is in reality a matter of a quite general disturbing influence. Certain phases in the cycle of development—the development of the vegetative system in the angiosperms—are in this connection so little susceptible to irritation that obvious disturbances (at least as a rule) do not occur after a first doubling of the chromosome number. They are, however, at least in the tomato, distinctly manifested after a repetition. During that part of the cycle of development having to do with the forming and development of the reproductive organs, on the other hand, there seem to be several more sensitive points, though in different organisms these seem to manifest different degrees of irritability. Hence after doubling we possibly will find, in one material an increased inhibition of the bud-development in the apical part of the inflorescences, in another inhibited development of the macroarchespores, or increased lethality in the fertilization products, etc.

J. As regards a number of Rosa hybrids I have an impression—though further studies are necessary—of certain parallels to the lability of the hip-producing rugosa doublings. One cannot avoid the suspicion that the more marked or less pronounced impairment of fertility in the diploid and triploid hybrids is here in part a consequence of the general physiological status of the sporophytes, and not only due to the gene and chromosome constellations of the different spore cells. It would seem that the new consolidated systems of the hybrid possess such a general physiology that the sensitivity to the changes of the internal and external milieus is greater than normal. Although this is a matter of still more extreme cases, I will point out the agreement in principle with such hybrids—e.g. Mahoberberis and Fatshedera—in which the total lack of flowers renders all sexual reproduction impossible.

Chapter 6. Does a secondary change of properties occur in the
chromosome-doubled Rosa rugosa and other polyploids?

A. The Rosa rugosa doublings of the year 1942 formed hips and seed only in the years 1954 and 1956. This notwithstanding, the flowering was of normal extent during a succession of years before 1954 and in the year 1955. If the culture series had only comprised these doublings I should undoubtedly have drawn a conclusion that the differences in the different years were conditioned by climatological factors. The summers of 1954 and 1956 had more rainfall than 1955 and the years immediately preceding 1954. The rugosa doublings of the years 1947, 1948 and 1950 showed, however, a total absence of capacity to form hips in 1954 and 1956 as well as in the intermediate year and the preceding years. The same was also observable in all the doubled rugosa hybrids. Thus climatological differences can not, or at least not alone, explain the facts in question. The doublings representing the different year-classes have throughout been cultivated under similar conditions in the Hortus Bergianus. Only a few steps separate the rugosa doublings of 1942 from several of the younger doublings. It therefore seems quite improbable that the differences in reaction emerging in 1954 and 1956 on comparison between the doublings of different years are conditioned by variations in the external milieu.

The seed used in the production of the different rugosa doublings was all of similar origin. It was collected without selection of any special bushes from a rugosa hedge growing in front of the institute building in the Hortus Bergianus. Chance can scarcely have entered in such a way that the 9 doublings of the year 1942 all derive from rugosa individuals with certain common properties and genes which are, on the other hand, absent in all the 8 individuals from which the later doublings derived. It therefore appears quite unlikely that the differences in reaction emerging in 1954 and 1956 on comparison between the doublings of 1942 and the younger ones should have been caused by gene differences in the primary material.

If the two eliminations are in agreement with the actual conditions, then there scarcely remains more than one explanation. The much debated difference is due to the fact the members of the year-class 1942 are at least five years older than the other rugosa doublings. The rugosa doublings of 1942 must then have undergone some kind of autonomous change, an "auto-adjustment". The relatively low age of the other doublings must have prevented the change in these from being carried so far as to lead to a noticeable effect.

The tetraploid Rosa species existing in nature must have been formed from diploids through chromosome doubling. All the newly formed Rosa doublings studied by me—both the intraspecific and the interspecific—are primarily completely or (in one case) almost completely sterile. This means that they have not been able to give rise to tetraploid species unless they have first changed their character. Thus we have here another argument for the opinion supported by the result of the analysis carried out above. The proof is not, however, conclusive. One cannot exclude the possibility that the doublings studied by me—chiefly R. rugosa and its hybrids—constitute a different case from many other Rosa doublings, and that the tetraploid species in nature derive from the latter. But this possibility seems to be rather slight. Tests in which doublings are produced without rugosa elements are desirable, nonetheless. Although these will be very time-consuming, it is my intention to try to carry them out.

B. The literature contains a few descriptions of phenomena which in a way seem to constitute parallels to the demonstrated rugosa doublings of which an account has been given above. SCHLÖSSER (1934) speaks of a gradual, time-consuming stabilization in a 52-chromosome clone of tomatoes. The clone derived from a newly formed (4n+4)-plant among 4n-siblings. This diverged from its sisters in a striking way. The latter showed normal fruit-forming properties. The diverging plant shed its flower-buds when these had attained an approximate length of 3 mm. It was reproduced by means of top-cutting for 9 generations. Each generation required a period of 2.5-3 months. With each generation the bud-shedding tendency was reduced. In the fourth generation, "kam es fast zur Anthese bei einigen Knospen, doch wurden sie aus inneren Gründen im letzten Augenblick abgestossen". In the sixth generation fully developed flowers appeared. In the fifth generation the buds contained lethal pollen in 80 per cent of the cases, but in the seventh the figure had sunk to 35 per cent, a feature thought to be due to changed chromosome distribution during the meiosis. The majority of PMC were believed to distribute the chromosomes in such a way that 24 went to the one and 24 to the other pole, while the 4 surplus ones remained lying in the plane of division. In the ninth generation fructification occurred. The fruits often showed a normal amount of seed. The F1-plants that were pulled up proved to be tetraploid. Control analysis showed that the members of the clonic chain of generations had the whole time possessed the chromosome number (4n+4).

In the same work, SCHLÖSSER observed that the tetraploids newly formed from the diploids showed, though in a weaker state, the same characteristics as the newly formed (4n+4)-clone. "Aus diploiden Kallusgewebe durch Regeneration gewonnene tetraploide Sprosse von S. Lycopersicum zeigen, wenn sie gesteckt werden, im Vergleich mit gleichgrossen frischgesteckten Stecklingen von alten tetraploiden Pflanzen eine ausserordentlich herabgesetzte Fertilität. Vonden angelegten Knospen werden ohne ersichtlichen Grund eine grosse Zahl auf frühem Entwicklungsstadium bei optimalen Kulturbedingungen abgeworfen. Und von den wenigen, die zur vollen Entwicklung kommen, bringt es wieder nur eine sehr kleine Zahl zum Fruchstansatz und gar zur Fruchtreife. Erst im Laufe einer Anzahl von Stecklingsgenerationen ändern sich diese Verhältnisse so, dass schilesslich sich die tetraploiden Pflanzen von den diploiden in ihrer Fruchtbarkeit nicht mehr wesentlich unterscheiden."

WETTSTEIN (1937) has made an extremely interesting analysis of chromosome-doubled Bryum caespiticium. In contradistinction to the original plant, the doubling was androgynous. Only in exceptional cases did it form more developed sprophytes and sporangia. The latter contained an extremely uneven spore-mass. From spores in one sporangium was reared in 1926 a progeny of 100 individuals. Of these, 22 died early; 63, of which one was called Cae 220, coincided on the whole with the doubled mother-plant, and was, like the latter, diploid. The other individuals were dioecious haploids or diploids. Cae 220 formed sporophytes during the first years, but these died at an early embryonal stage. In the succeeding years there was a gradual increase in the sporophytes' capacity for development. Little by little there appeared more and more capsules with the capacity to develop to maturity. In 1937 such good fertility properties were shown that the type manifested itself as a new species. The spore formation was so even that one could now count on the meiosis running a completely normal course. The restoration of normal fertility ran parallel with the successive elimination of the gigas properties. The new species was therefore very similar to Bryum caespiticium, but differed from this through the doubled chromosome number and the androgyny. It was given the name Bryum Corrensii. Later, this plant—i.e. chromosome-doubled, androgynous Bryum caespiticium—was found in nature.

The transformation, which was concluded after 11 years with the establishment of Bryum Corrensii, took place not only in the original Cae 220-individual, but also within a clonic series reproduced 8 times, as well as in series of plants which on one and three occasions respectively were reproduced with the help of spores occasionally formed in the course of the said period.

WETTSTEIN was inclined to believe that the restoration of the fertility and the disappearance of the external gigas properties were secondary consequences of inner transformations. In his opinion the primary change resided in the successive reduction of the cell size, the external, morphological, observable consequence of which was the elimination of the gigas properties.

C. It has on several occasions been stressed that polyploids which occur spontaneously in nature, and among these also those which may be regarded as intraspecific, often diverge more or less strikingly from the experimentally produced polyploids (cf. for instance Müntzing 1936, Fagerlind 1937, Wettstein 1937). The former frequently have better, the latter poorer fertility qualities; the former are often more feebly, the latter more strongly gigas-accentuated. As the cause of the difference one may of course reckon with the selection that in time takes place in nature. The above-described conditions regarding Solanum, Bryum and Rosa, however, open the way for the interpretation that at least in special cases the differences are due to the presence of a rather slowly running auto-adjustment.

The notion that the above-discussed changes are due to an autonomous capacity in the newly formed polyploids, and not to any other conditions or factors that may have been overlooked, seems perhaps to accord badly with the fact that many experimentally produced polyploids have been studied more or less thoroughly without any signs of auto-adjustment being observed. This, however may depend on presence of variation. I have already pointed out that changes of properties in different organisms after doubling of the chromosome number are in some cases more obvious than in others. It is possible that a less markedly changed system is the object of a less pronounced adjustment. This would imply a greater difficulty in observing the phenomena of change in some cases than in others. It is possible, furthermore, that the adjustment proceeds at a different rate in different organisms. In the tomato it would seem to proceed rather rapidly, in Bryum caespiticium-Corrensii more slowly, and in Rosa rugosa, which after 15 years appears to be at the beginning of its adjustment, the rate would seem to be still slower. The impossibility of observing an adjustment that may take place almost immediately after the chromosome doubling, and also an extremely slow adjustment, is obvious. In the former case one gets the impression that the effect of the chromosome doubling and the adjustment imply the result of a single process. In the latter case the adjustment is not observed if the study of the experimental material is discontinued too early. It is also possible that the capacity for auto-adjustment belongs only to certain species or certain biotypes of one and the same species. According to Wettstein Cae 220 had the capacity for auto-adjustment, while in the sibling plants this capacity was not observed. This would indicate that Bryum Corrensii occurring spontaneously in nature derives exclusively from such doubled B. caespiticium plants, which have had the capacity for a rather rapid auto-adjustment.

In Bryum Corrensii the gradual restoration taking place in the course of the adjustment runs parallel with the gradual elimination of the gigas properties. In Rosa rugosa no such parallelism has been observed. Nor have we any information as to such parallelism in Solanum Lycopersicum. These discrepancies are perhaps due to one of the two changes mentioned—for example the increase in fertility—being more extensive or more rapid than the other within a certain system. Nor, as regards Rosa rugosa, should it be forgotten that in contradistinction to the younger ones the 15-year-old plants are partially fertile, but that it is a matter of a rather inconsiderable degree of fertility. They perhaps correspond to a stage which Bryum Cae 220 showed at the beginning of the 11-year period of adjustment required to transform it into Bryum Corrensii. It is possible that the continuation of the auto-adjustment in the Rosa rugosa doublings to the end of the process will give fully fertile plants without the gigas properties.

On an earlier occasion, when I was not yet acquainted with Schlösser's work, and when Wettstein's study of Bryum Corrensii had not yet been published, I drew attention to the possibility that a process of auto-adjustment was the background to the general difference between the experimentally produced polyploids and those existing in nature. I stated my impression that the regression of properties was connected with a gradually occurring reduction in the size of the chromosomes. This I thought was conditioned by the fact that through the doubling the chromosomes had come to lie in a cell with new physiological proportions. I am not inclined to defend this hypothesis to-day, or at least not in an unmodified form. Nevertheless, I consider it not entirely out of the question that the auto-adjustment may have a "similar" background. When confronted with the new internal milieu the chromosomes and genes perhaps respond with a reduced intensity of activity in one or another respect; for instance, the work of synthesis that is concluded with the monid-reproduction perhaps proceeds more slowly. This implies the reduction of the total mass in the cell of a certain chromosomal fraction, perhaps of the total mass of the genetically active substance. In this way changed physiological proportions would obtain. This signifies a new system with relatively less disturbance of the equilibrium, which is successively adjusted through repetitions of the procedure, though each time on a smaller scale. The auto-adjustment would then constitute a kind of modification, one in which—in a way—the genetic material itself becomes an object for modification. One can not, however, rule out the possibility that the auto-adjustment is the expression for a slow modification of the ordinary kind.

If the auto-adjustment of the polyploids is a fact, it is tempting to indulge in reflections that lead to rather unorthodox conclusions. One might thus, for example, be tempted to develop the following reasoning.

If auto-adjustment actually takes place when the milieu offered to each individual chromosome by the surrounding cell is changed, then the phenomenon should occur independently of whether the new situation has been produced by chromosome doubling, by the introduction of extra chromosomes, or by some other factor. General physiological changes in the cells which closely agree with those occurring after doubling of the chromosome number can be brought about by changing the external milieu. It would thus not be out of the question that the milieu itself has contributed to the creation of new species, ecotypes and biotypes. The classical mneme doctrine (Semon 1908) would thus have a certain justification. We are acquainted with recombination and mutation; we are aware of the significance of selection and of the isolate. This does not, however, by any means prove the absence of other, as yet quite unknown processes of importance to evolution and phylogeny! If for the effect of a "mneme influence" to become perceptible it is necessary that it should have been active for long periods, our genetic experiments, restricted as they must always be in point of time, will never suffice for a demonstration or proof!

Through the following formulation of the ideas sketched above one may eliminate the conflict with the orthodox view.

On a certain occasion and in a milieu referred to as I, genotype A gives the phenotype A-I-1. After a progressively longer stay in this milieu, the successive phenotypes A-I-2, A-I-3, A-I-4, A-I-5, A-I-6 etc. appear. The process will develop eventually towards a point of stabilization. The differences between two successive links in the chain are throughout so small that they cannot be observed. Owing to the additive effect, phenotypes representing completely different parts of the chain will appear as distinctly different also genetically. The observer will thus get the impression that it is a matter of different genotypes. If at a certain juncture, e.g. when the phenotype A-I-5 and the "pseudo-genotype" corresponding thereto have been realized, the system is introduced into a new milieu—milieu II—then A-II-6, A-II-7, etc., are obtained. Owing to the existence of exclusively non-observable differences, A-I-6 and A-II-6 and the links in the chain in their vicinity will appear as if they corresponded to one and the same genotype. This also applies to links corresponding to each other in the more basal sections of the chains; but far up in the series one will on account of the additive effects get the impression that even links corresponding to each other represent quite different genotypes. If the chains develop in a direction running counter to natural purposes the death of the system will of course sooner or later ensue. Chains developing purposefully, on the other hand, heighten their competitive capacity. If a chain has reached, or if it is approaching, the point of stability, a sudden change in the base system, e.g. doubling of the chromosome number, should give rise to a new adjustment. If the above-mentioned conditions have not been arrived at, however, the adjustment must show a tendency to develop in new channels.

I have here pointed out certain theoretical possibilities. This by no means implies that I wish in any to assert that they correspond to the actual conditions, or even that I consider them probable. Our knowledge of the supposed auto-adjustment of the polyploids upon which these speculations are based is as yet very scanty. The drawing of conclusions must therefore be postponed! Continued studies are needed before we can decide whether the auto-adjustment in the polyploids actually does take place, whether in this case it is a matter of a more general phenomenon but one more strongly manifested in some cases than in others, or whether, instead, the impression one gets is due to overlooked sources of error. If should be established through continued studies of the Rosa rugosa doublings existing in cultivation, and the doubled rugosa hybrids of different ages, whether the plants that have been partially fertile in two recent years have the capacity successively to improve their fertility properties, whether they show a tendency to reduce their gigas properties, whether they have been objects for chromosome-diminution etc., and whether the younger doublings tend to follow the patterns of behaviour hitherto shown by the older ones. As the phenomena in question strike me as highly interesting, some of my pupils and I intend to take up similar studies with Solanum and Galium.


The completion of the present work has been delayed. The individuals belonging to the material studied have thus had an opportunity of flowering and setting fruit once more. In the autumn of 1957 the oldest rugosa doublings manifest on the whole the same conditions as in 1956. The variation that these showed in 1956 does, however, seem to be strengthened. One can, for example, in 1957 observe in 4R2 both abundantly flower-producing systems of shoots without any hip-formation and systems with a vary fair number of hips. The hip-forming capacity of 4R6 was much higher in 1957 than in 1956; in 1956 the percentual value was about 20 (16-32)—see Table III —, in 1957 it is at least 40 or 50. Two of the rugosa doublings of 1947 have in 1957 for the first time shown signs of hip-formation. They have produced an extremely modest number of defective hips with down and with soft achenes. Of the doubled hybrids, there were in one individual similar manifestations for the first time. Here, however, there appeared also one, certainly asymmetrical but nonetheless well-developed, fully mature hip. This contained, besides down and soft achenes, also 5 achenes, which, judging from their outer appearance, were quite normally developed. Other doubled rugosa individuals and doubled rugosa hybrids still showed a total lack of hip-forming capacity.

The conditions now adduced do not contradict the impressions given earlier in this work; rather are these in a certain measure strengthened. Against them on the other hand, militates the fact that the not yet "sufficiently aged" 16-ploid hybrid Rosa Sweginzowii x R. Moyesii which flowered for the first time this year showed a completely normal hip-forming capacity.

The Rosa multiflora doublings flowered rather modestly in the summer of 1957. They have yielded, even in relation to the number of flowers, only a modest number of hips, which will certainly not attain maturity. As the climatic conditions in the Stockholm region are such that in my experimental fields ordinary diploid multiflora will scarcely yield mature hips this year, and as the doublings referred to are now rather old, the observations are without any significance in connection with the problems discussed above.


Chromosome-double biotypes of Rosa rugosa and some of its hybrids have been produced during the years 1942-1950. The hip, achene and seed forming capacity of these biotypes has been studied.

In the first year of flowering and in several of the subsequent years none of the doubled types formed hips in spite of good flowering.

There was formation of hips only in 1954 and 1956. In 1956 this occurred in all the members of the oldest year-class (1942), but only in these. The younger plants never formed any hips, either in 1954, 1956 or in any of the other years.

A strong tendency to disturbed hip development appeared in all the hip-producing types. Besides hip primordia without tendency to further development and normally or almost developing hips, there occurred all kinds of intermediate phenomena. The frequency-figures were not only varying in different biotypes, but also in members of one and the same clone and presumably in different parts of one and the same bush.

All the hips contained an abnormally large number of pistils without capacity for development. The number and the quality of the achenes varied greatly in different types and in different hips belonging to one and the same type.

The achenes with morphologically normal exteriors possessed varying seed quality. The products of fertilization showed a varying tendency to die before maturity. The frequency-figures varied greatly.

The disturbance in the course of the development are explained not alone by the gene and chromosome constellation of the gametophytes and the fertilization products formed. The surrounding maternal tissue seems to exercise a strong, lethal influence.

The variation respecting the development of hips, achenes and seeds in one and the same individual appears to be due to differences in the "internal milieu" (see p. 243).

The inequality in hip-forming capacity shown in different years by the doublings of the oldest year-class, and in 1954 and 1956 by these in comparison with the younger doublings, appears to be due to the fact that the newly formed polyploids, which primarily are totally sterile, have a capacity for auto-adjustment. This results gradually in at least partial fertility.

A few cases of auto-adjustment in newly formed genetic systems described in the literature—chiefly those brought about by chromosome doubling—are reviewed. The supposed phenomenon of auto-adjustment is discussed.