Proceedings of the Royal Society of London. Series B 109(761): 126-148 (2 Oct 1931)

Embryo-sac Formation in Diploid and Polyploid Species of Roseae.
By C. C. Hurst, Ph.D. (Cantab.), F.L.S.
(Communicated by Sir Daniel Hall, F.R.S.—Received January 13, 1931.)
[Plates 10-15.]

Introductory Note.

During the past ten years, the author has been carrying out cytogenetical experiments with the numerous species of the Tribe Roseae of the Family Rosaceae. In the course of this research, 691 varieties have been examined eytologically and taxonomically, including representatives of 198 Linnean species and many natural and genetical hybrids of the six genera of the Tribe. Several workers have investigated the processes of pollen formation and male gametogenesis, but, owing to various technical difficulties, no one has yet published a consecutive and comprehensive account of the processes of embryo-sac formation in the species of this Tribe. Hoffmeister (1858) and Strasburger (1878) examined a few species of Rosa and found that several embryo-sacs were formed in one ovule. Péchoutre (1902) discovered two fully-developed embryo-sacs in one ovule in one species of, and both he and Strasburger found that the micropylar cell of the quartet develops to form the embryo-sac. Täckholm (1920, 1922) examined the formation of embryo-sac mother cells in the irregular polyploid Species of the Caninae Section of the genus Rosa, and discovered the remarkable unequal reduction division that takes place in the embryo-sac mother cells of these species, but technical difficulties in sectioning the older ovules and achenes prevented an investigation of the later stages. It seems desirable, therefore, to publish a general account of the processes of embryo-sac formation in Roseae, since the research covers new ground.

The various stages of female gametogenesis have been investigated in 193 varieties of 104 Linnean species of the Tribe, including the diploid species of Rosa L., Platyrhodon Decn., and Hesperhodos Ckll., and the regular and irregular polyploid species of Rosa L. Many crosses and hybrids have also been studied in this respect, but owing to their peculiar behaviour it is necessary, for a clear account, to give comparative pictures of both male and female gametogeneses, so that they cannot be included in this report. The outstanding eytological feature of the species of the Tribe Roseae is the constancy of their basic number of chromosomes. With the exception of a few hybrids, all the 1130 individuals so far examined by seven workers in Europe and America are Euploid, with 7 as the basic number. Of the six genera composing the Tribe, all the species are diploid, except those of Rosa L., in which the prevalence of polyploidy is a remarkable feature. In this genus there are diploids with 14 somatic chromosomes, triploids with 21, tetraploids with 28, pentaploids with 35, hexaploids with 42, and octoploids with 56 somatic chromosomes. Of these, most of the tetraploid and hexaploid species, and all the examined octoploid species are quite regular in the formation of their male and female gametes. The triploids are all of hybrid or garden origin, and no established triploid species has so far been found in Nature. The remaining tetraploid and hexaploid species and all the pentaploid species, which together comprise the old Section Caninae (sensu latissimo), i.e., Canineae Christ, 1873, irregular in their cytological behaviour and are distinguished by a peculiar mechanism of gametogenesis which is at present unique in plants and animals. In these irregular polyploid species the functional male gametes have only one septet of 7 chromosomes, the remaining septets—3 in tetraploids (21), 4 in pentaploids (28), and 5 in hexaploids (35)—being lost in the processes of gametogenesis. The male heritage of all these irregular polyploid species is therefore that of a simple diploid species.

On the other hand, the female reduction division in these species is regular but unequal, producing female gametes with 3 septets of chromosomes (21) in the tetraploid, 4 septets (28) in the pentaploid, and 5 septets (35) in the hexaploid species; 1 septet of chromosomes (7) being lost in each case. The female heritage of these irregular polyploid species is consequently that of a regular polyploid species higher than itself, i.e., an irregular tetraploid produces the female gametes of a regular hexaploid, a pentaploid those of an octoploid, and an irregular hexaploid those of a regular decaploid species.

By means of this remarkable mechanism, the specific somatic number of chromosomes is maintained whenever sexual reproduction occurs, notwithstanding the inequality of the reduction divisions.

These irregular polyploid species of Rosa are, however, only facultatively sexual in their reproduction, since their seeds are for the most part apomictical, reproducing the mother parent in every detail. Since the frequent crosses and hybrids produced by sexual reproduction are truly perpetuated from the first by apomictical reproduction, these species are extremely polymorphic in their characters.

Judging by the fossil remains of these species, this double process of hybridisation and apomixis has been going on from the time that they first appeared as hybrids or chromosomal transmutations of the regular polyploid and diploid species on the retreat of the ice of the Great Mindel glaciation in the Pleistocene Period. The present geographical distribution of these species is confined to Europe, Western Asia and North Africa, over an area which broadly corresponds with the area influenced by the Great Mindel glaciation in its advance and retreat.

Material and Methods.

Material.—Flower buds were collected from wild bushes growing in the Midland and Eastern Counties of England and in North Wales; in Switzerland, in the Upper and Lower Engadine, in the Cantons Uri, Ticino and Schwyz, and in the northern foothills of the Jura Mountains by the author and his wife; in Siberia, around Lake Baikal and in Turkestan, by Professor T. D. A. Cockerell, of Colorado, U.S.A. Collections were also made in the Royal Botanic Gardens, Kew, in the Botanic Gardens of Oxford, Cambridge, and Basel; in the collections of Colonel Gravereaux at La Belle Boseraie de l’Hay, Paris, of the late Canon Carew-Hunt at Albury, Oxford, and of the author at Burbage, Leicestershire (now at Cambridge). Many interesting results were obtained from plants grown from seeds collected for me by Professor Heslop Harrison, E.R.S., in Co. Durham, Professor Stanley Gardiner, F.R.S., in France, Professor Cockerell in America and Asia, Dr. Eileen Erlanson in America, the late Dr. Gadow, F.R.S., in Mexico, Dr. Hugh Scott in Abyssinia, and several correspondents in Canada and Japan. I am much indebted to Professor A. C. Seward, F.R.S., and Professor Stanley Gardiner, F.R.S., for laboratory accommodation at Cambridge, and to Mr. F. T. Brooks, F.R.S., for his personal instruction and early supervision in cytological technique. My wife has given me much assistance in the microscopical preparations and has made all the camera lucida drawings. I am especially indebted to the generosity of the Government Grant Committee of the Royal Society for a grant in aid of these experiments and researches, which, without their assistance, would have come to an untimely end.

Methods.—For the purpose of investigating embryo-sac formation, Carnoy’s Fluid was found to be the most satisfactory fixative, since its rapid fixing and greater powers of penetration ensured good preparations, which were not easily achieved by other methods when dealing with the hard bony achenes of Rosa. In the course of the work it was found advantageous to leave the material in Carnoy’s Fluid for a considerable period, a t least 24 hours and preferably longer, even up to 3 or 4 weeks, since material treated in this way showed better fixation and good penetration of the paraffin wax, which consequently led to easy sectioning.

Various stains were used, according to the stage to be investigated. Iron-Alum—Haematoxylin gave good pictures of embryo-sacs, but for counts of chromosomes Gentian-Violet—Orange G was found to be a more satisfactory stain. In other cases Flemming’s Triple Stain, Cyanin-Erythrosin and other transparent stains were used.

In examining the microscopical preparations, a Watson’s Oil-Immersion Condenser was used (aperture 1.34) with either a 1.5-mm. Zeiss apochromatic objective or an apochromatic 1/12 with x10 eye-piece for searching and x20 for detailed examination.

All the drawings were made with an Abbe camera lucida with the board at table level.

Embryo-sac Formation in Diploid and Regular Polyploid Species of Roseae.

The processes of megaspore formation are so similar throughout all the regular species, whether diploid or polyploid, that one description will cover them all. The first differentiation of the megasporangial tissue takes place immediately before or during the first division of the pollen mother cells. The integuments appear at this time and grow rapidly, the nucellar tissue quickly adding layer after layer. During this growth the embryo-sac mother cells, of which many form in each ovule, become elongated and pass into synizesis about the time of the reduction division in the pollen mother cells or rather later. This condition persists for a considerable time, and while the pollen mother cells complete their second division and while the resulting microspores grow into pollen grains, the size of the ovule rapidly increases. The integuments grow up and join at the top of the nucellus and many more layers are added to the latter. The embryo-sac mother cells become much elongated and form a nest of long and rather narrow cells in the lower part of the ovule. The number found in each ovule is variable even in the same flower-bud, but there are commonly about four layers at the chalazal end each with four cells and frequently one or two apical layers above. A few species, notably the CC diploid septet species R. rugosa Thunb. em., have comparatively few embryo-sac mother cells, occasionally only two or three, but usually eight or nine.

The nuclei pass through the same stages as the p.m.c. and the reduction  division (fig. 2, Plate 10) usually coincides with the first division of the nuclei in the young pollen grains. Cytomixis has not yet been observed in the e.m.c., and as it apparently depends much upon external conditions, it is possible that the embryo-sac mother cells being in a more protected position than the anthers are less likely to be affected. During the division stages, the presence of many e.m.c. is a great advantage since one may find divisions going on for a considerable period and all stages may be found, from synizesis to the completed quartet, in one and the same ovule. The inner cells are the first to divide and the early stages and first division are similar in all respects to the corresponding division in the p.m.c., the chromosomes forming typical gemini, 7 in the case of the diploid species (2n = 14), 14 in the tetraploid the hexaploid (2n = 42) (fig. 1, Plate 10), and 28 in the octoploid (2n = 56). The anaphase and telophase stages are carried through with great regularity and a dyad is formed with a cell plate dividing it into two cells (fig. 3, Plate 10). The second division spindles are formed in the same plane as the first division spindle, one above the other with the poles pointing to the micropylar and chalazal ends of the ovule respectively (figs. 2 and 4, Plate 10). Thus a regular quartet is formed, the four cells lying in a more or less straight row (fig. 5, Plate 11).

One remarkable case of irregularity was observed in a tetraploid (labelled R. Beggeriana, but more nearly allied to R. Fedschenkoana Regel), found at Kew. The p.m.c. divisions showed an orderly and regular tetraploid behaviour with 14 gemini and regular divisions. Since the plant was observed to be sterile, the e.m.c. were examined and in most cases there were 28 single chromosomes at diakinesis instead of the 14 gemini, or more rarely a variable number of gemini and singles. The singles appeared to reduce irregularly at the first division, and did not carry through an equatorial division, causing great irregularity and rapid degeneration. The presence of occasional embryo-sacs may have been the result of more complete pairing and consequently regular reduction or by the equatorial division of a 28 single chromosomed nucleus giving a "diploid" embryo-sac. The presence of apomixis consequent on this irregular behaviour is, however, unlikely since the plant was almost infertile and the few seeds sown failed to germinate. It is possible that the plant is a hybrid though, if so, the regularity of the pollen divisions is remarkable. This was the only case of irregularity found in the e.m.c. of species examined, all others behaved normally.

In all the regular species, while the innermost cells, which have divided first, are preparing for further development, the outer cells carry out their division. In some cases, degeneration sets in quite early in these outer cells giving rise to irregularities of division such as lagging chromosomes, so that it is not safe to take the behaviour of the chromosomes of these cells to be typical for the plant. In most cases, however, they perform a regular reduction and many reach a 2-nucleate embryo-sac stage or even attain to the 4-nucleate stage. Their continuance or discontinuance, however, depends largely on the number of e.m.c. present, those ovules with many showing much earlier degeneration of the outer cells than those with few. There is also a great difference in the stage of development between the ovules themselves. In the AA and EE septet species and some sub-species of the BB septet species, in which the achenes are basi-parietal, the ovules in those at the base of the receptacle are much more advanced than those up the sides, while in the CC and DD septet species and other sub-species of the BB septet species, in which the insertion is mostly basal, the ovules are all at approximately the same stage. In some cases, the outer cells remain in synizesis for a long time and one may often see some in this condition at quite a late stage of embryo-sac development. As a rule, however, they degenerate fairly early and are absorbed by the rapidly growing central sacs.

The central megaspores which reduced first are now developing into embryo-sacs. The micropylar cell of the quartet usually enlarges to form the functional embryo-sac (fig. 6, Plate 11), but this is not entirely constant, for one may occasionally see the chalazal (fig. 5) or even one of the two central cells continuing in some cases. This condition does not appear to be confined to any particular species since one may see the two conditions in different ovules in the same flower-bud or even in different quartets of the same ovule in some species. Afzelius (1924) found a similar condition in certain species of Senecio and allied genera. In these genera the chalazal cell usually persists, but in some cases the micropylar or one of the middle cells of the quartet appeared to be enlarging for division. Afzelius doubts, however, whether these would continue and he thinks it is possible that they would degenerate later and their place be taken by the smaller chalazal cell. It is difficult to follow the later stages of the quartet in Rosa in order to determine the ultimate fate of the cells or to judge the previous position of the developing one with regard to the others. The three cells which do not continue, degenerate with great rapidity and their places are quickly taken by the various embryo-sacs which are in course of formation, all trace of their former position being obliterated and the whole mass of growing embryo-sacs and still undivided lateral cells form such a confused assembly that it is almost impossible to sort them out with any degree of accuracy until a later stage, save in the few cases in which only one or two sacs are formed.

The first sign of embryo-sac formation is the enlargement of the cell, which is to continue, often accompanied by vacuolation of the cytoplasm. The nucleus also enlarges considerably and may often be pushed towards the side of the cell by a large vacuole, giving it a very similar appearance to the pollen grain just before the first division, as it appears in many species. The division takes place on the same plane as the previous divisions and the two nuclei resulting lie each at one end of the now rapidly elongating sac, usually separated by a large vacuole (fig. 7, Plate 11). The second division (figs. 8, 9 and 10, Plate 11) occurs quickly, and as a rule after the division there may be seen two nuclei lying closely together at each end of a long and narrow sac separated by a long narrow vacuole (fig. 11, Plate 11). In a few cases, in which the sac is broader and less elongated, the four nuclei lie in a quartet-like row in a short, thick embryo-sac (fig. 12, Plate 12).

The sac now enlarges considerably and penetrates farther and farther up towards the micropyle until finally it reaches as far as the junction of the integuments. Usually it ceases here, but sometimes it may penetrate nearly through the integuments or, curling round half-way, grow up the space between the nucellus and the integuments for the last part of the way. The last division to the 8-nucleate stage (fig. 13, Plate 12) does not take place until very late, in many cases not until the pollen tubes are already half-way down the styles, but more generally when the pollen is just beginning to germinate on the stigma. This last division takes place very rapidly, the cytoplasm surrounding the nuclei furrows and two synergids form at the extreme apex of the sac with the egg cell immediately below, while the first polar nucleus moves quickly down the long narrow sac to meet the second polar nucleus (fig. 16, Plate 12). The meeting takes place about one-third of the way down the sac (fig. 17, Plate 12) except in cases where division of the lower cells has been somewhat belated, when it may take place lower down. The two nuclei lie closely pressed together, but do not fuse before fertilisation. The three antipodal nuclei quickly disappear. They sometimes lie in an axial row (fig. 15, Plate 12), but more usually form a triangle with the apex to the chalazal end of the sac (fig. 14, Plate 12). The cytoplasm furrows and three small cells are formed, which rapidly degenerate.

Meanwhile the other e.m.c. also have been dividing and developing to form embryo-sacs. In some cases a second cell has kept pace with the first, giving two fully developed sacs, but frequently the other sacs formed (usually from two to six) remain at the 4-nucleate stage and degenerate about the time of fertilisation. When two 8-nucleate sacs occur, one is usually rather above the other, often curled over it, and hence is fertilised first (fig. 20, Plate 13), the second sac rapidly degenerating. No cases have been seen of two sacs being fertilised in the same ovule, and only very rarely do more than two 8-nucleate sacs develop in the regular species.

In spite of the large amount of material cut, the actual stage of fertilisation has been missed, although pollen tubes have been seen penetrating the micropyle (fig. 18, Plate 12) and the immediate after-stages have also been observed. Apparently fertilisation does not take place until several days after the fading of the flower. After fertilisation the fertilised egg cell remains dormant, while the endosperm nuclei begin to divide. When a single lining of free endosperm nuclei has been formed around the sac, the first division of the fertilised egg cell takes place (fig. 19, Plate 13) and a normal embryo quickly develops (figs. 21, 22, 24, Plate 13). The endosperm now commences cell formation, beginning first in the centre and top of the sac and forming a layer of very thin hexagonal cells (fig. 23, Plate 13). Later it grows inwards and more or less fills up the cavity of the sac, but disappears rapidly as the young embryo grows and fills up the sac, until in the end only a single layer is left. Endosperm divisions observed in R. Sayi, Schwein, a hexaploid species (n = 21, 2n = 42), showed that triple fusion had taken place. The chromosomes were very thin and elongated, which made counting difficult, but it was obvious that the number was about 60, and a few good plates gave clear counts of 63 (fig. 25, Plate 13).

Seminal reproduction in the diploid and regular polyploid species of Rosa is apparently wholly sexual, and there is no genetical or cytological evidence of apomixis in these species.

Table I.—Diploid and Regular Polyploid Species.

List of Diploid and Polyploid Species of the Tribe Roseae in which the Embryo-sacs have been examined by the author, with the number of Varieties examined in each Species and the number of Male and Female Gametic Chromosomes and Somatic Chromosomes found. The Species are arranged in their Cytogenetical Groups with Septet and Linnean Species (Hurst, 1928).

Vars Groups Chromosomes
Gametic Somatic
  TRIBE Roseae Genus Rosa. L.    
21              AA Septet Species 7 14
3                R. arvensis, Huds. 7 14
1     R. moschata, Herrm 7 14
1     R. Brunonii, Lindl. 7 14
1     R. Leschenaultiana, Wight et Arn. 7 14
2     R. longicuspis, Bertol. 7 14
1     R. Soulieana, Crép. 7 14
1     R. Helenae, Rhedr. et Wils. 7 14
1     R. Rubus, Lév. et Van 7 14
1     R. Banksiae, Ait. 7 14
2     R. chinensis, Jacq. 7 14
1     R. anemoneflora, Fort. ex Lindl. 7 14
2     R. multiflora, Thunb. 7 14
1     R. Wichuraiana, Crép. 7 14
1     R. setigera Michx. 7 14
2     R. (Linnean species, undescribed) 7 14
22   BB Septet Species 7 14
3     R. sericea, Lindl. (excl. fig.) 7 14
6     R. omeiensis, Rolfe. 7 14
2     R. Webbiana, Wall. 7 14
2     R. sertata, Rolfe 7 14
1     R. Willmottiae, Hemsl. 7 14
1     R. xanthina, Lindl. 7 14
3     R. Hugonis, Hemsl. 7 14
2     R. cabulica, Boiss 7 14
2     R. gymnocarpa, Nutt. 7 14
14   CC Septet Species 7 14
5     R. rugosa, Thunb. 7 14
1     R. nipponensis, Crép. 7 14
4     R. nitida, Willd. 7 14
4     R. (Linnean species, undescribed) 7 14
19   DD Septet Species 7 14
2     R. cinnamomea, L. 1759 (non 1753) 7 14
3     R. Beggeriana, Crép. (p.p. non Schrenk) 7 14
1     R. Marettii, Lêv. 7 14
1     R. Woodsii, Lindl. 7 14
2     R. Fendler, Crép. 7 14
1     R. bidenticulata, Rydb. 7 14
1     R. Johnstonii, Rydb. 7 14
2     R. blanda, Ait. 7 14
3     R. palustris, Marsh (R. carolina, L. 1762 non 1753) 7 14
3     R. (Linnean species, undescribed) 7 14
12   EE Septet Species— 7 14
1     R. macrophylla, Lindl. 7 14
3     R. corymbulosa, Rolfe 7 14
1     R. Giraldii, Crép. 7 14
1     R. elegantula, Rolfe 7 14
1     R. persetosa, Ralfe 7 14
5     R. (Linnean species, undescribed) 7 14
Vars Groups Chromosomes
Gametic Somatic
  TRIBE Roseae Genus Rosa. L.    
11              AACC Septet Species— 14 28
6                R. centifolia, L. 14 28
3     R. rubra, Blackw. 1757 (R. gallica L. 1759 non 1753) 14 28
2     R. damascena, Blackw. 14 28
1   AAEE Septet Species— 14 28
1     R. Davidii, Crép. (R. setipoda, Hemsl. et Wils) 14 28
6   BBCC Septet Species— 14 28
4     R. spinosissima, L. 1762 14 28
1     R. myriacantha, D.C. 14 28
1     R. (Linnean species, undescribed) 14 28
10   BBDD Septet Species— 14 28
1     R. pimpinellifolia, L. 1759 14 28
5     R. altaica, Willd. 14 28
1     R. grandiflora, Lindi. 14 28
1     R. Fedtschenkoana, Regel. 14 28
2     R. (Linnean species, undescribed) 14 28
1   CCDD Septet Species— 14 28
1     R. virginiana, Mill. (R. lucida, Ehr.) 14 28
18   DDEE Septet Species 14 28
14     R. pendulina, L. 1753 (R. alpina, L. 1762) 14 28
1     R. Hawrana, Kmet ex Kern. 14 28
1     R. laxa, Retz. in Hoffm. (non Lindl., nec Hort.) 14 28
2     R. (Linnean species, undescribed) 14 28
4   AABBEE Septet Species— 21 42
2     R. Moyesii, Hemsl. et Wils. 21 42
2     R. Sweginzowii, Koehne in Fedde 21 42
1   AACCEE Septet Species— 21 42
1     R. Hemsleyana, Tackh. 21 42
4   AADDEE Septet Species— 21 42
3     R. nutkana, Presl. 21 42
1     R. Nuttalliana, Hort. Kew 21 42
1   BBDDEE Septet Species— 21 42
1     R. Engelmanii, S. Wats. 21 42
1   CCDDEE Septet Species— 21 42
1     R. Sayi, Schwein. 21 42
2   AACCDDEE Septet Species— 28 56
1     R. acicularis, Hort. Kew (R. acicularis, Lind]. p.p. cum icon.) 28 56
1     R. acicularis, Hort. Gravereaux 28 56
1   BBCCDDEE Septet Species— 28 56
1     R. baicalensis, Turcz. 28 56
          (R. acicularis, Lindl. p.p. excl. fig.)    
          (R. acicularis, Hort. Cantab.)    
      Genus Platyrhodon, Decaisne.    
1     Platyrhodon microphylla, Decn. 7 14
          (Rosa microphylla, Roxb. ex Lindl.) (non Desf.)    
          (Rosa Roxburghii, Tratt.)    
          (Saintpierrea microphylla, Germain)    
      Genus Hesperhodos. Cockerell.    
1     Hesperhodos minutifolia, subsp. mirifica, Hurst 7 14
          (Rosa mirifica, Greene)    
          (Hesperhodos mirifica, Ckll.)    

Embryo-sac Formation in the Irregular Polyploid Species of ROSA L.

The formation of the female gametes has been examined in a representative  set of the Irregular Tetraploid and Pentaploid species of Rosa L., and the discovery of the unequal reduction division in the embryo-sac mother cells by Täckholm (1920 and 1922) has been fully confirmed. A large number of embryo-sac mother cells is formed in most cases, several more than those usually found in the regular species. Up to the reduction division the behaviour is the same as in the pollen mother cells, but after this it is entirely different. As in the regular species, the first meiotic division coincides approximately with the first division of the pollen grain nuclei. In diakinesis seven bivalent chromosomes appear, the chromosomes of the remaining septets being unpaired. In the first meiotic division the seven bivalents form a regular equatorial plate, but the univalents do not form up around them as in the pollen. As the bivalents pass to the plate the univalents all collect a t the micropylar end of the cell (figs. 26 and 28, Plate 14), where they are presently joined by seven of the reduced bivalents which have carried through a normal reduction division (fig. 29, Plate 14). Thus a regular dyad is formed, the lower cell containing a nucleus with only seven chromosomes, while the upper one contains the other seven plus all the univalent chromosomes (figs. 27 and 30, Plate 14), giving the unequal but regular reduction division of 7-21 chromosomes in the tetraploid species, 7-28 in the pentaploids, and 7-35 chromosomes in the hexaploid species.

The second meiotic division is normal and regular, being equational and non-reductional as in the regular species except for the different number of chromosomes included on the two spindles (fig. 30, Plate 14), the lower one having only 7 while the upper spindle has 21, 28, or 35 respectively in tetraploids, pentaploids and hexaploids. A regular, but unequal quartet of megaspores is formed, the two upper micropylar cells and nuclei being much larger than the two lower chalazal cells and nuclei (figs. 32, 33 and 34, Plate 15). In this way a remarkable mechanism is established whereby, through unequal syngamy, the somatic number of chromosomes is carried on to the next generation. The upper micropylar cell of the quartet, with its complement of all the univalent chromosomes 14, 21 or 28, plus 7 of the bivalents, continues to form the embryo-sac and, being fertilised by a male gamete containing 7 chromosomes, re-establishes the original somatic number of 28 in the tetraploids, 35 in the pentaploids and 42 in the hexaploids.

Some deviations from this regular division occur, some univalents being seen at the chalazal pole and odd chromosomes are extruded from the daughter nuclei, but since these irregularities occur as a rule only in the outer cells which soon degenerate they have no more significance than similar irregularities in the outer cells, of the regular species (pp. 130-131).

So far as observed, the central or inner cells which form the embryo-sacs were entirely regular in their unequal reduction. Tackholm (1922) found that about 75 per cent, of the embryo-sac mother cells carry through the regular unequal division while the author found 72 per cent., which is near enough to Täckholm’s figure to confirm his statement.

These described divisions being reductional are obviously sexual. Although careful search has been made through a large amount of material there is so far little cytological evidence for the apomictical development of embryo-sacs which genetical experiments and data demand. Hundreds of diakineses observed all show the 7 paired chromosomes instead of all single chromosomes as would be expected in apomixis. Three cases have been seen, however, of metaphases in which all the chromosomes were on the equatorial plate (fig. 31, Plate 14) and appeared to be carrying through an equational, non-reductional division, and a few isolated telophases have been observed which appeared to have approximately equal numbers of chromosomes at each pole. Large numbers of quartets examined show two large micropylar cells and two small chalazal cells, but here also there are occasional quartets with approximately equal cells. If the equational divisions observed in the embryo-sac mother cells are the origin of the apomictical embryo-sacs there is a possibility that they may be formed straightway without passing through the sexual dyad and quartet stages, thus reducing the probabilities of observation. In the heterogeneous collection of early embryo-sac development with dyad and quartet formation in both the inner and outer and the upper and lower cells, it is extremely difficult to make accurate observations at this stage in order to determine which are single cells and which parts of dyads, quartets or embryo-sacs, especially as they may be often, not only curled round one another in the struggle for development, but also actually one above the other, giving a row of cells which may contain a single cell, a dyad and a quartet. As in the regular species, many of the outer cells remain in synizesis until a late stage, but these have no significance since later evidence shows that it is the first formed embryo-sacs that produce the embryos so that the apomictical divisions must be carried through first of all and also in the central cells which always produce the surviving embryo-sacs.

In the formation of the sexual embryo-sacs the topmost or upper micropylar cell of the quartet which carries all the univalents and the reduced number of bivalents, is usually the functional one, though in many cases the lower micropylar cell may divide once and occasionally even twice (figs. 34, 35, 36, Plate 15). The two chalazal cells, containing only the 7 chromosomes, usually degenerate rapidly, though even here one may at times find one division taking place (fig 36). There is the possibility that a fusion might take place between one of the micropylar nuclei and one of the chalazal, thus producing an apomictical sac. That the fusion should take place between adjacent quartets is excluded, since segregation would result, unless a perfect state of homozygosity is present in these species, in which case normal self-fertilisation would have the same result. If one of the lower cells should persist and form an embryo-sac an interesting result might follow in the formation of a diploid species by fertilisation with a 7 chromosome male gamete. Such an occurrence is remote, however, since the small sacs would be rapidly displaced by the larger ones, unless by some abnormality only chalazal cells continued. The subsequent development of the embryo-sacs is the same as in the regular species except that in the irregular species there are consistently two equal-sized and equally well-formed 8-nucleate sacs in each ovule, and often more, as many as five having been counted in one ovule. Many 4-nucleate sacs occur around the main ones as in the regular species (fig. 38, Plate 15), which eventually degenerate. The fact that at least two complete embryo-sacs are present in each ovule may be significant, since one may be sexual and the other apomictical as in Hieracium (Rosenberg, 1906), though in the case of the apomictical sacs are of nucellar origin. Various counts taken in the developing embryo-sacs of Rosa all show the reduced number of chromosomes and, so far, none with the full somatic number have been observed. The occurrence of nucellar or integumental embryo-sacs with the somatic number of chromosomes as found by Rosenberg (1906) in Hieracium, and by Chiarugi (1926, 1927) in Artemisia, has not been observed which the most probable source of apomictical embryo-sacs appears to be the equational divisions of embryo-sac mother cells, leading straightway to the formation of apomictical embryo-sacs with the full somatic number of chromosomes, the normal first reduction division being omitted. Later cuts in several species show in all cases a perfectly normal embryo (fig. 39, Plate 15) in one of the embryo-sacs, the others degenerating after persisting for a time. These embryos are apparently formed from an egg-cell in the usual way and are surrounded by endosperm as in the regular species. Divisions in the endosperm are difficult to find, and although a great deal of material has been cut, no divisions showing that triple fusion has taken place have been met with, although the adhesion of the two polar nuclei has been observed (fig. 38). If the embryo-sac is apomictical with a double fusion of polar nuclei instead of the sexual triple fusion, there should be 7 more chromosomes present in the endosperm than in a fertilised one, since the male gamete would only bring in 7 chromosomes while the two apomictical polar nuclei would each have 7  chromosomes more than those in sexual sacs. (Apomictical double fusion, 35 + 35 = 70. Sexual triple fusion, 28 + 28 + 7 = 63). Since there is always a large percentage of bad pollen in these species, the pollen-grains on the stigmas are not so numerous as in the regular species, but a few grains may be observed germinating on the stigmas and their tubes penetrate freely down the styles. The actual stage of fertilisation has been missed, but in some cases the tubes have been seen entering the micropyle. In R. glaucophylla var. jurassica, collected by the author at Schuls in the Lower Engadine, a pollen-tube was seen approaching the embryo-sac. In view of the fact that hybrids are easily produced in crossing these irregular species with regular and other irregular species, and that the progeny show the expected number of chromosomes from a reduced sexual embryo-sac, it is evident that the sexual sacs persist, and it may be that the apomictical sacs are also present and function in the absence of fertilisation. That the own pollen of the irregular species should fail to function is remarkable, but the absence of any segregation or variation in "selfed" progenies shows that they must either be produced apomictically or that there is a definite and unusual homozygosity within the varieties of the species. Self-fertility in these species has not been demonstrated, and self-sterility leading to the functioning of the apomictical sacs in the absence of foreign pollen may be the explanation, since the pollen of the irregular species put on to other irregulars and on to regular species functions normally, though naturally not so freely as that of species which have 100 per cent, perfect pollen. In those species in which very young embryos were seen, and in a few cases where the endosperm was forming while the egg-cell still remained dormant, there was no apparent sign of fertilisation having taken place, such as degenerate synergids, pollen tubes and triple fusion, the sacs certainly having the appearance of continuing without fertilisation, although everything else was carried on apparently as normally as if fertilisation had actually taken place.

Table II.—Irregular Polyploid Species.

List of Irregular Polyploid Species of the Genus Rosa in which Embryo-sac  Formation has been examined by the author, with the names of the varieties in each species and the number of male and female gametic chromosomes and the somatic chromosomes found in each. The species are arranged in their cytogenetical groups and septet species, each letter in the specific formulae representing a septet of chromosomes and characters corresponding with the gametic septets of the five Differential Diploid Species of Rosa from which the Irregular Polyploid Species have been derived (vide Hurst, 1928). Provisional oldest names have been given to each Septet Species, but further research may lead to the discovery of still older legitimate names.

Groups Chromosomes
Gametic Somatic
Male Female
Genus Rosa, L.

Sectio Caninae (sensu latissimo).
AA Septet Group.                  
  Pentaploid Species.      
    AABDE Septet Species—          
            R. canina, L., 1753 em.      
               "       " var. lutetiana (Lém.) 7 28 35
               "       " var. sphaerica (Gren.) 7 28 35
               "       " var. nemophila (Desegl. et Ozan.) 7 28 35
               "       " var. dumetorum (Thuill.) 7 28 35
               "       " var. andegavensis (Bast.) 7 28 35
    AACDE Septet Species—      
      R. mollissima, Willd., 1787 non Fries (R. tomentosa, Sm.)      
                  "             " var. semitalis (Rouy) 7 28 35
                  "             " var. intromissa (Crep.) 7 28 35
                  "             " var. insidiosa (Rouy) 7 28 35
                  "             " var. alsatica (Rouy) 7 28 35
                  "             " var. lanuginosa (Rouy ) 7 28 35
      Undetermined Septet Species—      
      R. stylosa, Desv., 1809.*      
               "           " var. Desvauxiana, Ser. in D.C. 7 28 35
               "           " var. systyla (Bast.) 7 28 35
* The septet formula of this peculiar species has not yet been determined. It produces an abundance of hips which usually contain only a few good seeds (1-3) and is confined for the most part to the lowlands of Western Europe. Déséglise placed the species in the section Synstylae (whose species are all AA diploids), Christ placed it in the section Caninae (which includes all the other pentaploid species), while Crépin created a new section for it called Stylosae and  suggested that it might be a hybrid between R. arvensis, Huds., of the Synstylae and R. canina, L., of the Caninae. Its taxonomic septet formula is approximately 4A + 1C + 1D + 1E, while cytologically it behaves as if it were 2A + 1C + 1D + 1E, having usually 7 bivalent and 21 univalent chromosomes in both p.m.c. and e.m.c. with 7 chromosomes in its male and 28 in its female gametes. That the bivalents are AA is evident from experimental genetical tests which show them to be pure and probably homozygous AA. The genetical tests of the remaining 3 septets, involving at least 4 independent hybridisations, are not yet completed.
    AABCD Septet Species—      
      R. agrestis, Savi, 1798      
               "           " var. Bernardii (Rouy) 7 28 35
               "           " var. sepium (Thuill.) 7 28 35
      X R. bigeneris, Duff, ex Rouy (eglanteria♀ X micrantha♂) 7 28 35
    AABCE Septet Species—      
      R. micrantha, Sm. 1800      
                   "             " var. typica (Christ) 7 28 35
                   "             " var. microcarpa (R. Kell.) 7 28 35
                   "             " var. operta (Puget) 7 28 35
BB Septet Group.      
  Pentaploid Species.      
    ABBCD Septet Species—      
      R. eglanteria, L., 1753 (R . rubiginosa, L. 1771)      
                "            " var. rotundifolia (Rau) 7 28 35
                "            " var. heteropoda (Rouy) 7 28 35
                "            " var. apricorum (Rip.) 7 28 35
                "            " var. comosa (Rip.) 7 28 35
    ABBCE Septet Species—      
      R. elliptica, Tausch., 1819 (R. graveolens, Gren. et Godr.)      
               "             " var. typica (Rouy) 7 28 35
      Hexaploid Species.      
    ABBCDE Septet Species—      
      R. inodora, Fries, 1814 (non Borr. nec. auct. Britt.)      
               "           " var. typica 7 35 42
CC Septet Group.      
  Tetraploid Species—      
    CCDE Septet Species—      
      R. pomifera, Herrm., 1762 (R. villosa, L. p.p.)*      
                 "            "  var. Grenierii (Desegl.) 7 21 28
                 "            "  var. recondita (Puget) 7 21 28
                 "            "  var. murana (Rouy) 7 21 28
                 "            "  var. arvernensis (Rouy) 7 21 28
                 "            "  var. meridionalis (Rouy) 7 21 28
                 "            "  vra. iserana (Rouy) 7 21 28
                 "            "  var. vosgesiaca (Rouy) 7 21 28
* Formerly the tall, smooth, cany and almost prickleless forms of R. pomifera, often found in botanic gardens and in the lower regions of the Swiss Alps, were placed in the septet species CDEE (Hurst, 1928), but recent genetical experiments at Cambridge have demonstrated that these ecological forms more properly belong to the septet species CCDE and are not specifically distinct from the dwarfer and more prickly forms of the High Alps.
  Pentaploid Species.      
    ACCDE Septet Species—      
      R. uriensis, Lagg. et Pug. ex Crep. 1869      
               "           " var. pubescens (R. Kell.) 7 28 35
DD Septet Group.      
  Tetraploid Species.      
    CDDE Septet Species—      
      R. mollis, Smith, 1812 (R. mollissima, Fries, non Willd.)*      
               "         " var. typica W-Dod 7 21 28
* This northern and western species is not usually found in Britain south and east of Derby,  and since only one colony has been found in the Cambridge district (Lime Pit Hill nr. Fulbourn) it is possible that seeds from the north have been transported by migratory birds. Incidentally this introduction of R. mollis, Sm., to Cambridgeshire has given rise to a new natural hybrid with the native R. eglanteria, L. 1753, with which it grows (R. eglanteria ♀ X mollis ♂) and well combines the characters of these two species. R. eglanteria, L. 1753, has the septet formula ABBCD, producing female gametes ABCD, while R. mollis, Sm., has the septet formula CDDE, producing male gametes D, the hybrid is therefore ABCDD with its female gametes identical with those of its mother parent R. eglanteria and its male gametes identical with those of its father parent R. mollis. In its characters and septet formula this natural hybrid is identical with the septet species R. pseudo-mollis, Ley, and provides a further illustration of the origin of a species by hybridisation, in this case in a new district (cf. X R. bigeneris, under R. agrestis, Savi). During the last eight years this natural hybrid has established itself as a species by apomictical seeds without variation.
    ADDE Septet Species—      
      R. glauca, Pourret (non Vill.). (R. rubrifolia, Vill.)      
               "            " var. pseudo-glauca (Rouy) 7 21 28
               "            " var. dispersa (Rouy) 7 21 28
  Pentaploid Species.      
    ABDDE Septet Species—      

R. glaucophylla, Winch, 1816 (R. glauca, Vill.)

                   "             " var. jurassica (Rouy) 7 28 35
                   "             " var. platysepala (Rouy) 7 28 35
                   "             " var. falcata (Rouy) 7 28 35
                   "             " var. brevipes (Rouy) 7 28 35
    ACDDE Septet Species—      
      R. caesia, Smith, 1811. (R. coriifolia, Fries)      
               "          " var. venosa (Swartz) 7 28 35
    ABCDD Septet Species—      
      R. pseudo-mollis, Ley, 1907. (R. tomentosa pseudomollis, E. G. Baker)      
                  "              " var. typica 7 28 35
EE Septet Group.      
  Pentaploid Species.      
    ACDEE Septet Species—      
      R. Froebelii, Christ ex Grav., 1902 (R. laxa, Hort. non Retz. nec Lindl.)      
                 "            " var. laxa 7 28 35


  1. Embryo-sac formation bas been investigated in 193 varieties of 104  Linnean species in 3 genera of tbe Tribe Roseae.
  2. A description is given of tbe various stages and processes of embryo-sac formation in tbe diploid and regular polyploid species of tbe Tribe.
  3. The first and second meiotic divisions in these species are carried through in a regular and normal manner. The first division is tbe reduction division in which an equal number of chromosomes is segregated to each pole. The second division is equational, which leads to tbe formation of regular quartets arranged in an axial row. The micropylar cell of tbe quartet usually develops in to tbe normal 8-celled embryo-sac. Triple fusion takes place and after fertilisation normal endosperm is formed which, with tbe development of tbe embryo, is gradually reduced to a single layer. Minor variations occur from species to species and in some varieties, but tbe general mode of female meiosis and gametogenesis is normal and uniform throughout tbe diploid and regular polyploid species of tbe Tribe.
  4. Throughout tbe Tribe several embryo-sacs are produced in each ovule, one of which ultimately survives in tbe regular species.
  5. No evidence of the occurrence of apomixis was found in the diploid or regular polyploid species of Rosa.
  6. A description is given of embryo-sac formation in tbe irregular polyploid species of tbe genus Rosa which is entirely different in mechanism and process from that of tbe diploid and regular polyploid species.
  7. The reduction divisions of tbe embryo-sac mother cells are regular but unequal, giving rise to a mechanism of gametogenesis so far unparalleled in plants or animals.
         This mechanism of female gametogenesis differs from that of tbe male gametogenesis in the same species as widely as both differ from the normal formation in the regular species.
         In the first meiotic division, the 7 bivalent chromosomes alone reduce while the whole of the univalent chromosomes collect at tbe micropylar pole without splitting, thus giving rise to an unequal reduction division of chromosomes of 21 and 7 in the tetraploid species, 28 and 7 in the pentaploids and 35 and 7 in the hexaploid species.
         The second meiotic division is equational and regular, resulting in the formation of unequal quartets placed in an axial row of two large micropylar and two small chalazal cells. The upper micropylar cell develops into an 8-celled embryo-sac, the egg-cells of which contain three, four or five times as many chromosomes as the male gamete.
  8. The female heritage of these irregular polyploid species is consequently that of a regular polyploid species higher than itself, i.e., an irregular tetraploid produces the female gametes of a regular hexaploid species, a pentaploid those of an octoploid and a hexaploid those of a decaploid species, while the male heritage of all these irregular polyploid species is that of a simple diploid species.
  9. Experimental genetical evidence shows that these irregular polyploid species have an alternative method of reproduction by apomictical seeds. When "selfed" or agamised there is no varietal segregation, the progeny being completely identical with the mother parent. Self-fertility has not been demonstrated, but the species are faculatively sexual and hybridise freely when experimentally cross-fertilised with other varieties or other species.
  10. Cytological evidence of this apomixis is scanty, although occasional divisions are found in the embryo-sac mother cells in which the reduction division is suppressed and these cells contain the full somatic number of chromosomes. So far, however, the few good counts available in embryo-sacs show only the reduced number of chromosomes.
  11. The constant presence of two or more mature embryo-sacs in each ovule, peculiar to these species, may be significant, since one may be sexual and the other apomictical, the latter functioning in the absence of fertilisation.
  12. Notwithstanding the large amount of material examined, no traces of the formation of embryo-sacs from nucellar or integumental tissue have been found in Rosa.


Hurst Bibliography


Embryo-sac and Embryo Formation in Diploid and Regular Polyploid Species of Roseae.


FIG. 1.—Diakinesis in the Hexaploid species Rosa Moyesii Hemsl. et Wils. (21 pairs of chromosomes).

FIG. 2.—Telophase of first meiotic division in the Diploid R. gymnocarpa Nutt. (7 pairs of chromosomes reduced).

FIG. 3.—Preparation for the second meiotic division in the Tetraploid R. altaica, Willd. (14 pairs of chromosomes reduced).

FIG. 4 .—Second meiotic division, in the Octoploid species R. acicularis Lindl. (p.p. cum icon.) with 28 chromosomes (the reduced number) in each cell.


FIG. 5.—Quartet in the Hexaploid species R. nutkana Presl in which the chalazal cell is enlarging to form the embryo-sac (rare in Rosa).

FIG. 6.—First division of the micropylar megaspore to form the embryo-sac in the Diploid R. cabulica, Boiss. The normal process in Rosa. (Telophase with 7 chromosomes at each pole.)

FIG. 7.—Two-nucleate embryo-sac in the Tetraploid species R. centifolia, L.

FIG. 8.—Second division in the embryo-sac of the Diploid R. xanthina, Lindl. (7 chromosomes).

FIG. 9.—Second division (chalazal nucleus) in the embryo-sac of the Tetraploid species R. pendulina, L.

FIG. 10.—Second division of the embryo-sac (7 chromosomes) in Platyrhodon microphylla, Decn.; on the right the same chromosomes more highly magnified.

FIG. 11.—Four-nucleate embryo-sac in the Diploid Rosa pisocarpa, A. Gray, showing usual appearance.


FIG. 12.—Second type of 4-nucleate embryo-sac, found occasionally, in which the nuclei lie in a row near to one another, in the Diploid R. Hugonis, Hemsl.

FIG. 13.—Third and last division in the embryo-sac of the Tetraploid R. damascena, Blackw., leading to the formation of 4 antipodal nuclei, each with 14 chromosomes (two cuts).

FIG. 14.—Cytoplasm furrowing round three antipodal nuclei to form three cells, in the Diploid species R. rugosa, Thunb.

FIG. 15.—Antipodal cells formed in a less usual manner, one above the other, in the Diploid R. Hugonis, Hemsl.

FIG. 16.—Eight-nucleate embryo-sac in the Diploid species R. rugosa, Thunb., polar nuclei approaching one another, egg-cell and two synergids above, three antipodal cells below.

FIG. 17.—Later stage in an 8-nucleate embryo-sac, the two polar nuclei lying pressed together, in the Diploid species R. rugosa, Thunb.

FIG. 18.—Pollen tube, with its male nuclei, penetrating the micropyle in the Diploid R. nitida, Willd.


FIG. 19.—Fertilised egg-cell in prophase for its first division, in the Tetraploid species R. pendulina, L., one synergid degenerating and the remains of the pollen tube above, a lining of free endosperm nuclei already formed.

FIG. 20.—Second division in the young embryo in the embryo-sac on the left, with lining of endosperm. On the right, a second embryo-sac degenerating, the two polar nuclei already degenerate, in the Tetraploid species R. pendulina, L.

FIG. 21.—Young multicellular embryo in the Diploid R. palustris, Marsh, the endosperm commencing cell formation.

FIG. 22.—Later stage in which the endosperm is forming cells in the upper and central parts of the embryo-sac, in the Tetraploid R. Kotschyana, Hort. Grav. (non Boiss.), an undescribed Linnean species of the Septet species BBDD.

FIG. 23.—Endosperm cells of the preceding more highly magnified.

FIG. 24.—Older embryo in the Hexaploid species R. Sayi, Schwein.

FIG. 25.—Divisions in the Enneaploid endosperm of the Hexaploid species R. Sayi, Schwein., showing 63 chromosomes as the result of the triple fusion of the first polar nucleus (21), the second polar nucleus (21), and the male nucleus (21).

Embryo-sac Formation in the Irregular Polyploid Species of Rosa, L.


FIG. 26.—Regular but unequal reduction in an embryo-sac mother cell of the irregular Tetraploid species R. glauca, Pourret, var. dispersa (Rouy) (R. rubrifolia, Vill., var. dispersa, Rouy): first division metaphase with 7 bivalent chromosomes on the equatorial plate and 14 univalents passing undivided to the micropylar pole.

FIG. 27.—Later stage in the irregular Tetraploid species R. pomifera, Herrm., var. arvernensis (Rouy) (R. villosa L. var. arvernensis Rouy): interkinesis with 7 chromosomes in the chalazal nucleus and 21 chromosomes in the micropylar nucleus.

FIG. 28.—First division metaphase in the embryo-sac mother cell of the Pentaploid species R. glaucophylla, Winch, var. platysepala (Rouy) (R. glauca, Vill., var. platysepala Rouy), with 7 bivalent chromosomes reducing on the equatorial plate and 21 univalents going to the micropylar pole.

FIG. 29.—Later stage in the Pentaploid species R. uriensis, Lag. et Pug., var. pubescens, R. Kell., in which the 7 bivalent chromosomes have reduced and one set of seven has gone to the chalazal pole, while the other set of seven has joined the 21 univalents at the micropylar pole, making an unequal reduction of chromosomes of 28 at the micropylar pole and 7 at the chalazal pole.

FIG. 30.—Second division metaphase in the Pentaploid R. stylosa, Desv., var. Desvauxiana, Ser. in D.C., with 7 chromosomes in the chalazal cell and 28 in the micropylar cell.

FIG. 31.—Equational and non-reductional division in an embryo-sac mother cell of the Pentaploid species R. Froebelii, Christ, var. laxa (Hort.), which might give rise to an apomictical embryo-sac (35 chromosomes).


FIG. 32.—Two megaspore quartets in the irregular Tetraploid species R. mollis, Sm., var. typica, W. Dod, showing the smaller chalazal cells and nuclei and the larger micropylar cells and nuclei, the uppermost enlarging to form embryo-sacs, pushing away the lower ones which are beginning to degenerate.

FIG. 33.—Megaspore quartet in the irregular Tetraploid species R. pomifera, Herrm., var. vosgesiaca (Rouy) (R. villosa, var. vosgesiaca, Rouy), in which the lower micropylar cell is enlarging apparently to form an embryo-sac. Note distinction in size between the two upper and lower cells and nuclei.

FIG. 34.—Megaspore quartet in the Pentaploid species R. glaucophylla, Winch, var. brevipes (Rouy) (R. glauca, var. brevipes, Rouy), showing metaphase plates in both of the upper and larger micropylar cells, each with 28 chromosomes, the reduced number.

FIG. 35.—Similar case in the Pentaploid species R. mollissima, Willd., var. semitalis, (Rouy) (R. tomentosa, var. semitalis, Rouy), in which the upper micropylar cell is already 2-nucleate, while the lower one has a metaphase plate with 28 chromosomes, the reduced number.

FIG. 36.—Megaspore quartet in another variety of the Pentaploid species R. mollissima, Willd., var. insidiosa (Rouy) (R. tomentosa, var. insidiosa, Rouy), in which the upper and lower micropylar cells and the upper chalazal cell have carried through one division.

FIG. 37.—Metaphase plate of second embryo-sac division in the Pentaploid species R. mollissima, Willd., var. insidiosa, (Rouy), with 28 chromosomes.

FIG. 38.—Embryo-sacs in the Pentaploid R. agrestis, Savi, var. sepium (Thuill.) (R. sepium, Thuill.): two fully grown 8-nucleate embryo-sacs are present, both showing adhesion of polar nuclei; there are also present one small 8-nucleate and several 4-nucleate and 2-nucleate embryo-sacs.

FIG. 39.—Embryo in the Pentaploid species R. canina, L., var. nemophila (Desegl. et Ozan.) (R. nemophila, Desegl. et Ozan.), with the normal lining of endosperm.