Genetics 9: 241-276 (1927)
The Origin of New Forms in Rubus, I
M. B. Crane and C. D. Darlington
John Innes Horticultural Institution, Merton, London

pp. 244-253
Hybridisation Experiments

R. ruticanus inermis x R. thyrsiger.

R. ruticanus inermis is the only blackberry we know totally devoid of prickles or, indeed, armature of any kind. It has been crossed extensively with many species with varying success. The plant used in these experiments was totally self-sterile, but plants raised from natural seeds of a R. rusticanus inermis from Kew were all without prickles and had the typical hairiness of the species. The seedlings however were of two kinds, the larger group consisting of plants growing six or more feet high, and the smaller of plants that never exceeded eighteen inches. The general resemblance of these individuals to R. rusticanus inermis especially the complete absence of spines shows that they were either crosses with another inermis individual or, more probably, selfs of the plant from which the seeds were taken, in which case this individual must have been self-fertile. The plant used in these experiments has 14 chromosomes in somatic divisions (fig. 4).

FIG. 1. R. thyrsiger. Type and selfed offspring. A. one of the two seedlings having broader leaflets and shorter pedicels than the type. B. one of the eight seedlings with elongated pedicels and leaflets of the type form. A scale of inches is shown here and in the following photographs.

Our plant of R. thyrsiger was self-fertile. A number of selfed seedlings were raised from it, but only ten have been kept and grown to maturity. Eight of these are identical with the parent. The other two differ in having shorter peduncles and pedicels, and broader leaves; in consequence their leaflets overlap more than in the type, and their panicles are less lax (fig. 1) their glands, acicles, hairs and prickles are the same in form and distribution as those of the type. It should be mentioned here that the inflorescence of the compact type is still much more prolonged than that of R. rusticanus (fig. 2). This compact form is apparently the expression of a simple recessive factor, although held to be of great importance in systematic work. The parent plant has 28 chromosomes in somatic divisions (fig. 4); it is, relatively to R. rusticanus, a tetraploid.

FIG. 2. R. thyrsiger showing the typical long and lax inflorescence; pedicels and peduncle elongated. R. rusticanus: inflorescence compact; pedicels and peduncle short.

The following table summarizes the difference between these forms.

Stem:furrowed, moderately pigmented.   bluntly angled, moderately pigmented.
Hairs; stellate, forming a down.   numerous, partly patent, partly appressed.
Prickles: none
(in the type: confined to the angles and only of the large, broad-based kind, of variable form, mainly declining or patent; no acicles or pricklets).
  unequal, declining or falcate, not entirely confined to the angles many short gland-tipped acicles short stout-based pricklets.
Stalked glands; none.   numerous.
Leaves: dark green, nearly glabrous above, close whitish felt beneath, leaflets rather close together. Coriaceous, convex.   paler green, strigose above, thinly hairy and quite green beneath, leaflets widely separated. Not coriaceous, flat.
Flowers; petals imbricate, deep pink, sepals reflexed, broad and fairly short.   petals stellate, pale pink, sepals reflexed and very elongated.
Panicle; very dense, pedicels and peduncles short, flowers ultra-axillary, branches mostly 7-flowered.   lax, very long, pyramidal, pedicels and peduncle elongated, branches mostly 2-flowered.
Time of flowering: late and continuous.   early and more limited.

In 1922, 20 flowers of R. rusticanus inermis were emasculated and crossed with R. thyrsiger. They yielded a few seeds of which five germinated but only three survived. The three F1 plants are subsequently referred to as RT2, RT3 and RT4. The first two have not yet produced flowers, but in their vegetative characters they are alike and strongly resemble R. thyrsiger (fig. 3), that is to say, they have the same type of growth, their glands, hairs, and acicles are of the same kind and have the same distribution; the prickles are also like those of R. thyrsiger, but are even more abundant; and as in that species, the lower surface of the leaves is green on account of the reduction in hairiness.

FIG. 3. R. rusticanus type and var. inermis, R. thyrsiger, R.T. 4 (tetraploid) and R.T.2 (triploid) F x seedlings. Showing form and distribution of prickles.

The other seedling, RT4, differs widely from RT2 and RT3 and strongly resembles R. rusticanus type (the prickled form) except that it is larger growing (fig. 3). It has no stalked glands or acicles, even at the growing points; the prickles are all of one kind, large-based, nearly patent and confined to the angles of the stems; the inflorescence is compact, the petals are deeply pigmented and the period of flowering is late and continuous; in all these respects the seedling is indistinguishable from R. rusticanus. The stems and under-surfaces of the leaves are felted with stellate hairs but rather less densely than in R. rusticanus. This reduction in hairiness, the thinner and less buckled form of the leaves, as well as the presence of prickles, show the influence of R. thyrsiger.

Thus, while RT2 and RT3 show a general dominance of R. thyrsiger characters, RT 4, on the other hand, shows a dominance of the characters of the R. rusticanus inermis parent; at the same time the form of the leaves, the presence of prickles and the reduced density of the felted hairs prove that there is no question of apogamy. It must be noted here that even in regard to the prickles, although their presence is determined by the R. thyrsiger parent, their form and distribution are typically those of R. rusticanus. Hence the R. rusticanus inermis parent must be supposed to carry the typical form and distribution of prickles of the normal type, although itself incapable of producing them.

1) (a) the felted under-surface of the leaf.
    (b) the absence of stalked glands
    (c) the distribution of the prickles
    (d) the absence of acicles
2) In the Bull. of Appl. Bot. XVII (3) two analogous cases are recorded (Karpechenko's Raphanus-Brassica cross and Eghis and Rybin's Nicotiana Tabacum x N. rustica); in both the same differential genetic effects in tetraploid and triploid are observable and the same assumption is made with regard to their constitution.

The seedlings RT2 and RT3 have the somatic chromosome number 21 (fig. 4) ; they are triploid. A root of RT2 was found with the hexaploid number, the result of the somatic chromosome-doubling so often described. RT4 on the other hand is tetraploid (2n = 28). Summing up the genetical evidence, we find that the two triploid seedlings resemble most closely the tetraploid parent; they are strongly patroclinal. This is what we should be led to expect from the fusion of a diploid paternal gamete with a haploid maternal gamete. The tetraploid on the other hand shows four clear vegetative characters derived from the diploid mother plant that are not shown in the triploids 1). In view of this we must regard the tetraploid as having derived its extra chromosome set from the maternal side, the product therefore of the fusion of a diploid paternal gamete with a diploid, instead of a haploid, maternal gamete — an unreduced ovum 2). Accessory evidence for the occurrence of such unreduced gametes, already described in other genera, in Rubus is provided by another experiment to be quoted later. This view will also be shown to be in accordance with the results of segregation in F2.

We may denote the chromosome complement of R. rusticanus inermis by the letters AA (as we are unaware of any genetical differences between the chromosome sets furnished by R. rusticanus and any difference between them must be very slight) and that of R. thyrsiger by the letters BBCC (as, for reasons we shall show, we cannot regard the two R. thyrsiger sets as wholly identical); the history of the family is thus represented in the scheme outlined in the diagram (fig. 4).

FIG. 4. History of the family derived from crossing R. rusticanus inermis and R. thyrsiger. Drawings of somatic divisions of the plants concerned (x 4100, reduced to 1/2). The division of RT4 is an anaphase, the rest, metaphases.

Family RT4 Selfed

It has been mentioned that RT2 and RT3 have not yet produced flowers. RT4, however, has flowered and fruited abundantly, and it is noteworthy that it has a higher proportion of good pollen than either of its parents (fig. 5). Attempts to back-cross the mother plant, R. rusticanus, on to it, failed.

FIG. 5. Mature pollen. A. R. rusticanus inermis. C. R. thyrsiger and B. The tetraploid F1, R.T.4.


1) Sown the previous September.

It is however perfectly self-fertile, the seeds resulting from self-pollination were well-developed and over 90% germinated — for Rubi, an exceptionally high proportion. This year a selfed family of 490 individuals was raised 1). Shortly after germination and before prickles had formed on any of the seedlings this F2 family was classified according to whether the plants possessed stalked glands or not. There were 471 with glands and I9 without.

Some weeks later the seedlings were examined for prickles and it was found that the plants which were originally glandular had all developed prickles, whilst the eglandular plants, like the female parent of RT4, were totally devoid of prickles.

At an early stage, only stalked glands formed on the seedlings which were subsequently to develop prickles. A little later, occasional gland-tipped prickles occurred among the stalked glands, and the plants soon reached a stage of development when prickles appeared without any glandular structure. The parent seedling, RT4, was not examined so soon after germination but it presumably passed through the same stages in developing to the same mature form. At the present time the growth of all the plants is entirely eglandular. It is therefore evident that glands and prickles are morphologically equivalent, and the classification of the family in this respect is:

Primarily, stalked glands only, secondarily, stalked glands and gland-tipped prickles, and finally eglandular prickles only . . . . . 471 plants   Eglandular and without prickles throughout growth . . . . . . . . .19 plants.

In July the prickled seedlings were again examined and it was found that all but eight had abundant prickles. These eight exceptional plants were very sparsely prickled, having usually only one prickle to an internode instead of eight or nine. A further examination in October showed that only four of the seedlings retained this reduced prickling (fig. 6B), the other four having produced growth which approached the normal prickled condition.

All the plants have prickles of the R. rusticanus and RT4 type although the frequency of the prickles varies and a few are exceptionally heavily armed. The totally unprickled form and four degrees of prickling — only arbitrarily distinguished from one another are illustrated in fig. 6 A —E.

There is thus a failure, on the mature growth of all the seedlings, to maintain the development of the stalked glands and acicular structures — in other words the R. thyrsiger type, in which the juvenile condition continues in maturity, never reappears.

Variation occurs in the degree of the anthocyanin pigmentation of the stems and leaves — in which the parent species differ little — most of the plants being very heavily pigmented, and in the height of the plants. Other differences are also evident, but the seedlings are as yet too young to enable these differences to be recorded with confidence.

FIG. 6. F2 seedlings from R.T.4 (R. rusticanus inermis x R. thyrsiger) A. totally without prickles. B. very sparsely prickled, only one present on the shoot figured. C. Slightly prickled, a type which in early stages was very sparsely prickled. D. moderately prickled, the commonest type.

The proportions of segregation in this family should enable us to determine the conditions governing the pairing of the four chromosomes (AABC) of particular types in the parent. In this constitution AABC we have an entirely new condition, for tetraploids previously studied have been the result of doubling, somatic or germinal, in a species — more or less true-breeding — in which case differential affinity is not to be expected, or the result of doubling in an interspecific hybrid F1, in which case pairing should be simply between chromosomes from the same parent. In the tetraploid Datura for example (Blakeslee, Belling and Farnham, 1923) no significant departure yet is found from the ratio expected with random assortment.

In the tetraploid Primula kewensis, on the other hand, (PELLEW, 1927) constant pairing within the chromosome complements of the parental species prevents any general segregation of their characters. Now, in this tetraploid, we are dealing with chromosomes of three kinds, A, B and C; the differences between B and C are hypothetical, based only on general considerations, but they cannot be assumed to be absent and the possibilities in regard to assortment are therefore various.

We may consider for simplicity three possibilities in regard to any one member of the chromosome set: first, pairing of A with A and of B with C (autosyndesis) which will never yield the pure form AAAA; secondly, pairing of A with B and of A with C (allosyndesis) yielding 1 in 16 of AAAA, and, thirdly, an ideal condition of "random assortment" (Muller, 1914), where the chances are equal amongst the three possible assortments of each of the four chromosomes; from this we should expect 1 in 6 of AA gametes and 1 in 36 AAAA zygotes. The following table shows the observed proportions of segregation and expectations of Unprickled individuals in the F2:

  prickled unprickled
Observed numbers 471 19
Autosyndesis expectation 490 0
Random assortment expectation 476.4 ± 2.5 13.6 ± 2.5
Allosyndesis expectation 459.4 ± 3.5 30.6 ± 3.5

It will be seen that the observed numbers depart from the free-pairing expectation by 2.2 times the probable error and from the allosyndesis expectation by 3.3 times the probable error. Of course, the two expectations are not alternative. If we admit the existence of a differential affinity between homologous chromosomes, free pairing becomes a theoretical ideal; relative frequency of different possible associations being determined by the relative attraction, free pairing presupposes perfect equality in the attractions between the four homologous chromosomes in the tetraploid; this condition is barely conceivable where they are of three distinct kinds. The results indicate therefore the possibility of the departure from perfect free pairing being in the direction of allosyndesis. A larger family and reciprocal F2 back-crosses to test for the possibility of differential fertilisation will no doubt show whether there is any such preference of two identical chromosomes for pairing, not with one another, but with two dissimilar chromosomes — whether, indeed, the A chromosome serves to bridge the gap between the B and the C. It may be remarked that there is no evidence of differential zygotic viability in favour of inermis individuals or otherwise.

To consider a second character, the type of prickling of all the seedlings is that of R. rusticanus. The differences in habit of prickling of the two species are considerable; their development and their genetical basis may be complex, but the entire absence of acicles and stalked glands on the mature growth of the seedlings means that every one of them has, in respect of at least one chromosome, two or more representatives from R. rusticanus, that is, has the constitution AABC, for the R. thyrsiger characters are partially dominant, appearing with the triploid A : BC balance. The only regular limitation of pairing that can account for this is absolute autosyndesis — the pairing of A with A so that B must pair with C — giving each of the 980 gametes concerned in the production of the family the constitution, again in respect of this chromosome only, A B or A C and, every one of the zygotes, the constitution AABB, AABC or AACC.

It may be remarked that if pairing is absolutely determined to take place only between B and C in the F1, this tendency can scarcely be weakened in later generations, because the dissimilarity of B and C cannot be increased by crossing over: on the contrary in-breeding will tend to reduce it and make still sharper the differentiation between this pair of chromosomes and their homologues in the A set; autosyndesis will be established. In this way it is possible that the sharp distinctions usually evident between different chromosome sets in polyploid species may arise partly before, and partly after, hybridisation.

Comparing this segregation in regard to the kind of prickling, with the segregation in regard to the possession of prickles we see that in the two cases the relationships of the chromosomes of the three sets A, B and C are entirely different, in the one, leading to autosyndesis, in the other, favouring allosyndesis. This reminds us that it is perhaps too often assumed, in crosses between polyploid species or between diploids and tetraploids, that conclusions with regard to the pairing relationships amongst the chromosomes of one type are necessarily applicable to the chromosomes of the other types. This experiment demonstrates clearly for the first time that such an assumption is unwarrantable.

It will be noticed that the results of autosyndesis in this cross bear a marked resemblance to the „suppression of characters on crossing” observed by BIFFEN (1916) and to the „shift” detected by ENGLEDOW (1920), both in crosses between tetraploid Triticum species.

1) Here we are dealing, of course, with a main factor,
like Biffen, not with a modifying factor, like Engledow.
2) Nature CXIX, p. 320.

In these hybrids certain characters contributed by one of the parents failed to reappear in the F2 and later generations of the offspring. In our case, also virtually a cross between tetraploids, the explanation offers itself inevitably that pairing has taken place of A with A and, in consequence, of B with C, so that, by pairing internally, the R. thyrsiger complement in respect of a particular chromosome is never recovered 1). This explanation, suggested by Darlington (1927) 2), in the Triticum case, is even more strongly supported by the results of this experiment than by the original evidence. Indeed we can now see that in the production of a true-breeding intermediate hybrid and in the suppression of characters on crossing essentially the same process is involved.


  1. BELLING, J. and BLAKESLEE A. F., 1923. — The Reduction Division in Haploid, Diploid, Triploid and Tetraploid Daturas. Proc. Nat. Acad. Sci. IX.
  2. BLAKESLEE, A. F., BELLING, J. and FARNHAM, M. E., 1923. — Inheritance in Tetraploid Daturas. Bot. Gaz. LXXVI.
  3. BUXTON, B. H. and NEWTON W. C. F., 1928, — The Production of a New Tetraploid Hybrid in Digitalis. Journ. of Gen. XIX. (in press).
  4. CLAUSEN, R. E. and GOODSPEED, T. H. 1925. —Hybridisation in Nicotiana. Genetics X.
  5. CRANE, M. B., 1927. — Studies in Relation to Sterility in Plums, Cherries, Apples and Raspberries. Mem. Hort. Soc. of New York. 3.
  6. DARLINGTON, C. D., 1926. — On the Cytology of the Cherries. Brit. Assoc. Report of the 94th meeting.
  7. DARLINGTON, C. D. 1928. — Studies in Prunus I and II, Journ. of Gen. XIX (in press).
  8. DELAUNAY, L. N., 1926. Phylogenetische Chromosomenverkürzung. Zeitsehrift f. Zellforschung u. mikr. An. 4.
  9. FOCKE, W. O., 1911-14. — Species Ruborum. Bibl. Bot. Stuttgart.
  10. HAGERUP, O., 1927. Konjugierende Chromosomen bei Papaver-Hybriden Sv. Bot. Tidskr. 18.
  11. HUSKINS, C. L., 1927. — On the Genetics and Cytology of Fatuoid or False Wild Oats. Journ. of Gen. XVIII.
  12. LAXTON, W., 1906. — The Cross-Breeding and Hybridisation of Peas and of Hardy Fruits: Rep. of Third Int. Conf. on Gen. London.
  13. LIDFORSS, Bengt, 1914. Résumé seiner Arbeit über Rubus. Zeitschr. f. ind. Abs. u. Vererbungslehre XII.
  14. LONGLEY, A. E., 1924. — Cytological Studies in the Genus Rubus. Am. Journ. Bot. 11.
  15. LONGLEY, A. E. and DARROW, G. M., 1924. — Studies of Diploid and Polyploid Forms in Raspberries. Journ. of Ag. Res. XXVII.
  16. MULLER, H. J., 1914. — A New Mode of Segregation in Gregory's Tetraploid Primulas. Am. Nat. XLVIII.
  17. PELLEW, C, 1927. — The genetic behaviour of Primula kewensis. Zeitschr. f. ind. Abst.- u. Vererb. XXVII.
  18. ROSENBERG, O., 1926. — Die semiheterotypische Teilung und ihre Bedeutung für die Entstehung verdoppelter Chromosomenzahlen. Hereditas VIII.
  19. WILSON, J. H., 1906. — Infertile Hybrids. Rep. of Third Int. Conf. on Gen. London.
  20. WINGE, Ö., 1924. — Zytologische Untersuchungen fiber Speltoide und andere mutanten~hnliche Aberranten beim Weizen. Hereditas V.

Crane & Thomas: Reproductive Versatility in Rubus (1949)

Darlington: Integrated Species Differences (1949)