The American Rose Annual, 1953, pp. 111-122.

Max Graf and Its Progeny, with Special Reference to Rosa Kordesii

One of the important functions of the American Rose Annual is to record scientific research on rose problems. Through the good auspices of Wilhelm Kordes we are able to publish this excellent paper by Dr. Wulff on a fundamental research project which will undoubtedly become standard reference work on the subject.

Max Graf
IN 1919 a new trailing rose was registered under the name Rosa hybrida, Max Graf (American Rose Annual, Vol. 5, 1920, p. 166). This rose had been discovered by the foreman, Max Graf, in the nursery at Pomfret Center, Conn., owned byJames H. Bowditch. According to Egan (1, 2) it was a chance seedling which presumably arose from a cross between the two species Rosa Wichuraiana Crép. and Rosa rugosa Thunb. My guess is that the latter rose was the mother of the new hybrid. Some details on the morphological characters of Max Graf and its qualities as a garden plant are given in the American Rose Annual 1920, and it pictured on plate X in Volume 9, 1924. A very remarkable fact is that Max Graf only rarely produces seed.

Nevertheless, the German rose breeder Wilhelm Kordes (Sparrieshoop in Holstein) after repeatedly selfing it for years found some hips on his plants of Max Graf in 1940 and raised two seedlings from them. These were different in their morphology as well as in their hardiness. They both were planted in the open field in the autumn 1941; one of them, more resembling R. rugosa, died during the severe winter 1941-42; the other, more similar to R. Wichuraiana, survived and is known under the name Rosa Kordesii. Max Graf and its seedling R. Kordesii show striking differences especially as to their foliage, their flowers (which are double and deeper colored in R. Kordesii) and chiefly their fertility. Max Graf is almost totally sterile, while R. Kordesii produces seed abundantly after selfing or crossing. It was the latter fact which encouraged me to undertake a cytological investigation of Max Graf, R. Kordesii, and two more seedlings of the first mentioned rose grown by Mr. Kordes from seeds of the years 1950 and 1951. The results are reported in two earlier papers (7, 8) and shall be briefly repeated and somewhat completed here.

The first result was to find that Max Graf has 14 chromosomes in its body cells (FIG. 1). As 7 is the basic chromosome number of the genus Rosa, Max Graf thus proved to be diploid like its supposed parents, R. rugosa and R. Wichuraiana. R. Kordesii (FIG. 2) and both the seedlings from 1950 and 1951 are tetraploid and show the double chromosome number, 2n = 28, in the cells of their root tips or n = 14 chromosomes (FIG. 3) after the reduction divisions and in the sex cells.

I cannot give an exact explanation of how the doubling of the chromosome number in R. Kordesii and the other two seedlings of Max Graf occurred. The reduction division of this hybrid is highly irregular. It shows multivalent configurations of chromosomes and univalents instead of 7 bivalents which would be necessary for the normal course of the reduction division. As a consequence the distribution of the chromosomes to the two poles is extremely disturbed in Max Graf and ends with the formation of a very high number of non-viable pollen grains and egg cells, and results in the high degree of sterility of the plant.

It may be conjectured that the distribution of the chromosomes to the two poles is sometimes totally inhibited; they all may be enclosed again into a single, viable nucleus, a so-called restitution nucleus. This, of course, would have the unreduced chromosome number, possessing 14 instead of the expected 7 chromosomes. Fertilization of an unreduced egg cell by an equally unreduced sperm cell may give, then, a seedling with 28 chromosomes in its somatic tissues.

FIG. 1. Mitosis in the root tip of Max Graf with 14 chromosomes. FIG. 2. Reduced chromosome plate (x = 14) from anaphase I of the reduction division of R. Kordesii. FIG. 4. Reduction division of R. Kordesii (diplotene) showing some bivalents and a ring of four (quadrivalent). FIG. 5. Ring-shaped quadrivalent and some bivalents in the diakinese of the reduction division of R. Kordesii. FIG. 6. Chain of 4 chromosomes and 12 bivalents in the diakinese of the reduction division of R. Kordesii. FIG. 7. Metaphase I (side view) of the reduction division of R. Kordesii with chain of 4 chromosomes and some separating bivalents. FIG. 8. Anaphase I (side view) of R . Kordesii. At each pole two chromosomes in advance. FIG. 9. Metaphase I of R. Kordesii with a trivalent group (left), some bivalents, and a univalent (right). FIG. 10. Anaphase I of the reduction division of R. Kordesii with two lagging and splitting chromosomes. FIG. 11. Late anaphase I of the reduction division of R. Kordesii with normal chromosome distribution. FIG. 12. Interkinesis of the reduction division of R. Kordesii with 2 nuclei and some univalents not enclosed with them. FIG. 13. Anaphase of the reduction division of R. Kordesii. Some chromosomes advance. FIG. 14. Telophase of the reduction division of R. Kordesii with 4 nuclei and some univalents free in the plasm. FIG. 15. Normal anaphase II of the reduction division of R. Kordesii with 14 chromosomes in side view (above) and polar view (below). FIG. 16. Four young pollen grains of R. Kordesii in a regular "tetrade." FIG. 17. Irregular "pentade" of R. Kordesii showing 5 young pollen grains of different size.

Thus, the generally observed sterility of Max Graf depends on meiotic disturbances which, in their turn, are caused by structural changes (translocations) of the chromosomes and not, as I supposed in my earlier papers, by a lack of homology between the R. rugosa and R. Wichuraiana chromosomes which have come together in this hybrid. The American cytologist Miss Erlanson (Mrs. MacFarlane) was the first to show the presence of such structurally changed chromosomes in some roses (3, 4). In these plants as well as in other genera they usually lead to the formation of multivalents in the reduction division and to a more or less pronounced sterility which in Max Graf nearly reaches completeness. As soon as the chromosomes are doubled in number, as is the case in our three roses with 28 chromosomes, they will find better possibilities for a normal pairing in twos, the reduction division proceeds more regularly and, therefore, the fertility of R. Kordesii is rather high. The percentage of good pollen grains is, for instance, as low as 16 in Max Graf, but in R. Kordesii 85 percent of the pollen grains are good.

Because R. Kordesii possesses in its body cells (28 chromosomes) all the chromosomes of R. rugosa (14) and all the chromosomes of R. Wichuraiana (14), which probably do not pair mutually but only inter se in the reduction divisions; it does not show any Mendelian splitting up as fertile, numerical hybrids normally do. R. Kordesii represents a so-called amphidiploid plant; it is a new species rather than a hybrid, and was named as a new species in honor of its discoverer, Wilhelm Kordes.

It is a well known fact that amphidiploid plants generally are true breeding unless chromosomal irregularities during the reduction divisions occur. Though the progeny (F1) of R. Kordesii, as referred to above, does not show Mendelian segregation it is, nevertheless, not quite uniform. A cytological investigation revealed the reason for this unexpected behaviour. Like its mother, Max Graf, R. Kordesii shows multivalent and univalent chromosome configurations in the reduction division, but they are found in a minor degree and do not disturb very much either the course of the reduction division or the fertility. They do, however, effect the formation of sex cells which are genetically unequal and, by the various possibilities of their combination during fertilization, lead to a segregation in the F1.

Some drawings will give a better impression of the course of the reduction division in R. Kordesii. Generally 14 pairs of chromosomes or bivalents are formed resulting in a quite regular reduction division and distribution of the chromosomes (figs. 3, 11, and 15) and at the end of the division four pollen grains are seen, enclosed in the old membrane of the pollen mother cell (FIG. 16). Sometimes single chromosomes are wandering to the poles in advance (figs. 8 and 13). Only a small portion of the microspore mother cells 12 bivalents are formed instead of 14, and besides the 12 bivalents a quadrivalent in the shape of a ring (figs. 4 and 5) or an open chain (figs. 6 and 7) is to be seen. Figure 9 shows a trivalent and a univalent, and figure 10 two lagging univalents which have split. Such lagging univalents are often not enclosed in the nuclei (figs. 12 and 14) and finally may degenerate or cause the formation of "pentades" of 5 pollen grains of unequal size (FIG. 17). In the figures 4 to 9, all the 12 bivalents are not always shown.

* Ed. note: There is some question of the advisability of giving all amphidiploids arising from the same species hybrid botanical names because it would eventually result in multiplying the number of species in the genus beyond reason.

Because the two seedlings of 1950 and 1951 did not flower until now, only the chromosome number of their root tips could be traced with 28. But we may expect that they will, on closer examination, show cytological behaviour similar to R. Kordesii and that they will be as fertile as this plant is. They both must be regarded as new amphidiploid species, too, and they will be named* when their vegetative development is more advanced and when they have come into bloom.

An increased chromosome number is, in many cases, accompanied by an enlargement of the cell volume. Pollen grains and the stomata of the green leaves are cells which can easily be measured to detect variations in size. The results of such measurements are given in Table I.

Comparison of the Size of Stomata and Pollen Grains of Max Graf, its Parents, and its Offspring

Name of plant Average length
of stomata
(in 1/1000 mm.)
Average diameter
of pollen grains
(in 1/1000 mm.)
R. Wichuraiana 32.32 ± .042 41.70 ± 0.50
R. rugosa 20.80 ± 0.29 39.20 ± 0.78
Max Graf 27.52 ± .026 39.20 ± 0.90
R. Kordesii 39.04 ± 1.22 56.00 ± 0.95
Max Graf seedling 1950 30.50 ± 0.50 .
Max Graf seedling 1951 31.00 ± 0.29 .


The first three values of Table I (first column) clearly show an intermediate inheritance of the length of stomata. A comparison of the last four values, on the other hand, reveals that the doubling of the chromosome number has caused an enlargement of the stomata in R. Kordesii as well as in the two later seedlings. The fact that R. Kordesii grew in the open field, the other two seedlings in the greenhouse, may perhaps be responsible for the different degrees of stomatic enlargement in the last three plants.

With regard to the diameter of the pollen grains we are able to state (Table I, second column) that the two species roses and Max Graf have pollen grains of practically the same size, whereas the pollen grains of R. Kordesii show a remarkable enlargement.

There are, besides the size of the pollen grains, some other characters of the flower which show changes in connection with chromosome duplication. We know, for instance, that the development of diploid plants is faster than in tetraploids derived from them (6). Thus, the diploid Max Graf is a profuse June bloomer beginning to bloom on the 1st of June 1948, while the tetraploid R. Kordesii began blooming on the 15th of the same month. The single flowers of Max Graf last about 3 days, but a flower of R. Kordesii will remain in good condition for two weeks.

If we now turn to the morphology of our roses little can be said about the two seedlings from 1950 and 1951. Our comparison must be confined, therefore, to Max Graf, its parents and R. Kordesii. Regarding the latter four roses it is very striking that only R. rugosa has a straight, upright growth while the three others are arching to the ground. We can learn from this behaviour that trailing growth is dominant in inheritance in roses. Underground shoots are characteristic for R. rugosa, they are missing in the other three roses, indicating a recessive factor is responsible for them. The internodes at the tip of the aerial shoots are rather short in R. rugosa, its leaves standing more or less crowded. They are long in R. Wichuraiana leaving much space between the leaves. Max Graf shows an intermediate condition and R. Kordesii, probably due to its tetraploidy, has slightly longer internodes than its mother (FIG. 18). The internodes of R. rugosa bear 30 to 50 spines about 10 mm. long, R. Wichuraiana has only 2 to 5 spines 3 to 7 mm. long and Max Graf 3 to 7 spines 4 to 8 mm. From this it seems that the number as well as the length of spines is inherited intermediately. R. Kordesii again demonstrates the effect of tetraploidy: we find up to 20 spines on an internode and their length may reach to about 15 mm. The figure 18 illustrates some important details of the shape and outline of the leaves and stipules of our four roses. Generally speaking, the leaves of R. Kordesii, though they are larger are very similar to those of R. Wichuraiana. This refers to the shape as well as to the color and the glossy surface.

The size of the flowers is inherited intermediately, the flowers of Max Graf being larger than those of R. Wichuraiana but smaller than those of R. rugosa. In this respect R. Kordesii again shows an enlargement as a result of tetraploidy, its flowers approaching or even surpassing those of R. rugosa in size. The most remarkable phenomenon, however, is that R. Kordesii has double flowers whereas Max Graf has simple ones. It is rather difficult to decide whether the doubling of petalage is an effect of tetraploidy, too, or whether it indicates that the R. rugosa plant involved in the original hybridization also possessed double flowers. The latter alternative becomes highly probable when we see that R. Kordesii resembles R. rugosa also in other characters of flower and fruit which are not expressed in Max Graf. I may refer here to the free stamens and the cup-like shape of the flowers. The shape and the slow ripening of the hips and the falling of the sepals are, on the other hand, features in which R. Kordesii behaves like R. Wichuraiana (figs. 19 and 20), while Max Graf closely resembles R. rugosa. It is very likely that this whole extraordinary behaviour in inheritance of the flower and fruit characters must be connected with chromosome translations mentioned above.

A clear example of intermediate inheritance is offered by the color of the flowers. R. Wichuraiana possesses white, R. rugosa red flowers; in Max Graf they are of a bright pink color (with golden centers) which becomes intensified in R. Kordesii.

The stalk of the flowers remains upright in R. Wichuraiana, Max Graf, and R. Kordesii when the hips are ripening; only in R. rugosa does a postfloral downward curving occur (FIG. 19) thus indicating that this growth behaviour in the latter rose depends on a recessive gene. To sum up we may conclude that the morphological investigations, only briefly described here, speak in favor of the opinion that Max Graf is a chance hybrid between R. rugosa and R. Wichuraiana. Egan (1, 2) already has written favorably of its qualities (for instance, its hardiness) and considered it to be a plant of importance for the nurseryman.

Of far greater importance, however, is its tetraploid progeny because of its fertility and the same chromosome number (2m = 28) as our garden roses which allow hybridization with them. Already a good number of hybrids between R. Kordesii and garden roses has been produced by Mr. Kordes, and he speaks of them (5) as the most splendid, most beautifully colored and absolutely hardy climbing roses which ever existed.


  1. Egan, W. C., "The New Trailing Rose, Max Graf." American Rose Annual 5: 55-56, 1920.
  2. ——, "The Trailing Rose, Max Graf." American Rose Annual 9: 81-82, 1924.
  3. Erlanson, E. W., "Chromosome Organization in Rosa." Cytologia, 2: 256-282, 1931.
  4. ————, "Chromosome Pairing, Structural Hybridity, and Fragments in Rosa." Botanical Gazette, 94: 551-566, 1933.
  5. Kordes, W., Das Rosenbuch. Holzminden (Trowitzsch-Verlag), 1951.
  6. Müntzing, A., "The Evolutionary Significance of Autopolyploidy." Hereditas, 21: 263-378, 1936
  7. Wulff, H. D., "Rosa Kordesii, eine neue amphidiploide Rose." Züchter, 21. 122.132, 1951.