Mendel’s work on Geum hybrids


by David Allen

Return to main article
Works Cited
Jacobus Heimans is not alone in seeking to marginalize Mendel’s work on constant hybrids. Ivo Cetl, for example, wrote that the results of Mendel’s experiments on plants such as Geum and Hieracium were invariably the same: constant hybrids 'were not observed at all' (2002/3 p. 15). But Cetl distorts the evidence.

Geum urbanum × G. rivale was one of the crosses that Gärtner had said produced constant hybrids ('the primary hybrids [i.e. F1 hybrids] maintain a hybrid appearance in the following generation, and no individuals appear of a pure parental type'—cited Prywer 1932, p. 87).1 In other words: these were instantly constant, with an F2 generation which did not segregate. Others had found the same—Thomas Bell Salter, for example, reported in 1852 that the vast majority of hybrids he produced

were precisely like each other, with the exception that one or two were slightly more like one of the pure parents; and so also with the subsequent generations: they remained like each other, and like the first hybrids; and to this day the same form continues to propagate itself, by seed, in my garden.

We have, then, in this latter case, not only an instance of two species considerably different being capable of forming permanent and fertile hybrids, but also the proof of such actually existing in the wild state! (Salter 1852, p. 740.)

In 1869, Mendel sent Nägeli some specimens of the F2 Geum hybrids he had produced. He included the following notes on the hybrids:

No. 21. Several hybrids from the same fertilization varied somewhat in flower size and were of unequal fertility. Nos. 22 and 23 are the two parental types Nos. 24, 25, 26 will flower for the first time this year. Nos. 27 and 28 produced some incomplete flowers last year. According to Gärtner, progeny of this Geum hybrid show no variation. (OG, pp. 84-5)

Cetl speculates that Mendel’s sample probably contained several specimens 'that were different at first sight' (2002/3, p. 15); and thus, Nägeli would instantly see that the hybrids were variable, and not constant as Gärtner claimed. But this assumes that the hybrids were different; but the only variation Mendel noted, in fact, was in flower size (and this was only 'somewhat' variable).

Mendel made some notes on Geum on the flyleaf of his copy of Gärtner's book. He listed some traits where G. urbanum and G. rivale were different; those for rivale were marked a, b, c, d, and e (the lower-case letter indicating recessive).2 There were also formulae for G. intermedium: 'ABcDEe', and 'ABcdEe'. Cetl concludes that these formulae must refer to Mendel's hybrid crosses, and show that he had 'obtained variability at least in two trait pairs (D,d, and E,e) in the F2' (2002/3, p. 15). However, if these were the results of hybrid crosses, and they followed the Pisum pattern, we would expect to see segregation in all traits, and not just in one or two.

Mendel noted (in a letter to Nägeli) that Gärtner had found two types of hybrid: the majority had 'large flowers approaching those of rivale'; only a few had 'small yellow flowers approaching closer to those of urbanum' (Gärtner 1849, p. 302). Mendel, in contrast, found only one type in the F1 generation—in his case, with small yellow flowers like urbanum (leading him to speculate that this type had become the rule, not the exception; see OG, p. 69). All other traits corresponded to G. intermedium Ehrhart.3

Geum intermedium is a wild-growing form, first identified by Ehrhart. Its status was disputed—it was seen by some as a distinct form, and by others as a hybrid of G. urbanum and G. rivale. (For example, in 1832, one writer objected to the plant's designation as an 'intermediate form' and insisted it was a real 'planta hybrida'—Hornung, p. 196.) It seems likely that, in his notes in Gärtner's book, Mendel was outlining the different traits of G. rivale, G. urbanum and G. intermedium, perhaps prior to his experiments—and not describing the results of his hybrid crosses. He might have been planning to compare the results of his own hybridisation experiments with the G. intermedium form. (He certainly analysed the different forms—he told Nägeli that G. intermedium cannot be described 'without qualification' as an intermediate type between the parental species—OG, p. 59.)

Mendel does not record any other details about the results of his G. urbanum × G. rivale crosses.4 There is no evidence, then, to support Cetl's claim that these experiments proved that constant hybrids do not exist.

Cetl goes on to argue that, in the early years of the 20th century, it was conclusively demonstrated that Geum plants (and Aquilega atropurpurea-canadensis, another cross which Mendel tested) do not, in fact, produce constant hybrids, but segregate like Pisum; and this—together with the discovery of apomixis in Hieracium—meant, according to Cetl, that the whole theory of constant hybrids had collapsed. (Cetl 2002/3, p. 34)

Actually, hybridists found different and even contradictory results in rivale × urbanum crosses. Frederick Weiss, D. Rosen and Eric Marsden-Jones all claimed that the hybrid, on selfing, did not breed true, but that segregation took place. (However, Weiss' plants in the F3 generation showed little or no variation from their F2 progenitors.) In 1919, in contrast, Louis Blaringhem saw no segregation or 'deviation, nor any return to the parental forms. This hybrid behaves like a good, well-fixed species' (p. 112). Czesława Prywer put Blaringhem's results down to his small sample. Prywer’s own experiments showed both ordinary segregation of dominant and recessive traits; and independent assortment (1932, p. 106). Thus, it seemed as if the case was closed: Geum hybrids obeyed Pisum laws after all. But closer examination of Marsden-Jones' results, in particular, suggest the case is more complex than it may appear.

Marsden-Jones took a specimen from the wild which he termed G. intermedium × rivale —i.e., he assumed it was the result of a spontaneous backcross between intermedium and rivale. He found that, when this plant was selfed, no segregation took place 'in the ordinary sense of the word, the plants were like their parent, only minor fluctuations being observed, such as are to be found within the range of a species' (1930, p. 390). It was widely held among florists that there were two types of Geum intermedium—a sub-rivale and a sub-urbanum. This was rejected by Rosen, on the basis that segregation would make this impossible (1916, p. 165); but Marsden-Jones's results lend some credence to the idea of a stable sub-rivale form.

According to the principle of transformation, backcrossing eventually makes a hybrid revert to the original parental form. (Gärtner claimed that G. urbanum could be transformed through backcrossing into G. rivale in four generations—1849, p. 464.) But Marsden-Jones had found what appeared to be an at least partially stabilized, sub-rivale hybrid form; and he found that, in a locality where he conducted a field study, this new form was even 'very largely replacing rivale, an example of a wild population of a new type arising from inter-specific hybridisation' (p. 395). (Focke observed that, in the wild, there is usually a full range of intermediates between the two parental species; but he also saw that, in many places, two types—which he termed a per-rivale and a per-urbanum—could be distinguished; 1881, p. 124.)

Marsden-Jones claimed that 'on selfing G. intermedium, segregation always takes place' (p. 394). However, he himself collected a specimen of G. intermedium from the wild, and on selfing it, found that four distinct types appeared in the F2 generation. Two types were like rivale in appearance, and two nearer to urbanum. (Blaringhem had found two different forms in the F2 generation—1920, p. 1286.) These results break the laws of segregation and independent assortment. One possible explanation—as suggested by Gajewski in his study of Geum—may lie in the fact that, in interspecific hybrids, the elimination of zygotes may occur at various phases of development in the F2 generation, and this 'may favour the plants more related with the F1 hybrids and parental species' (1957, p. 364). A mechanism such as this would promote the more rapid establishment of certain distinct forms; and it may account, in part at least, for some of the unusual results which were found by Marsden-Jones and Blaringhem—and perhaps, also, by Gärtner and Bell Salter.

Thus, despite asserting that his results proved segregation, it is surprising how many of Marsden-Jones' results showed constancy of form. They indicate how a hybrid form may become stabilized, despite being surrounded by the parental types—to the point when it may (a) supplant the parental type, and (b) acquire the status of a new type, if not a new species.


Cetl, Ivo (2002/3). 'Mendel's Hybridization Experiments With Other Plants Than Pisum', in Folia Mendeliana 37-38, pp. 5-36

1. Prywer claims that he has taken this quote from Gärtner (1849). However, I have been unable to locate the quote in this source.

2. Olby transcribes these notes (1985 p. 212), but mistakenly suggests that it was the G. urbanum characters that were marked as recessive, rather than G. rivale.

3. The traits specified by Mendel are listed in this table:

Geum urbanum Geum rivale
A: lower joint of awn glabrous.
B: lower joint of awn four times
as long as upper joint.
C: upper joint pubescent at the base.
D: flowers erect.
E: fruit-bearing calyx reflexed.
F: carpophore lacking.
a: hairy at base.
b: almost as long as the upper joint.

c: pubescent to just under base.
d: flowers drooping.
e: erect.
f: carpophore almost as long as calyx.
Note: The letter labels in the left-hand column appear in square brackets because they were not present the original (i.e., they are inferred from the lowercase letters in the right-hand column).

Prywer claimed that dominant traits in G. urbanum included an erect floral peduncle. He found that, while some traits were intermediate, there was no trait in which G. rivale was wholly dominant (pp. 92-3). Rosen, in contrast, specified several traits in which G. rivale was dominant, including drooping flowers (cited Prywer p. 93). Gajewski's results were different again: he claimed that G. rivale was partially dominant in the drooping of the floral peduncle and in the erect position of the calyx (1957 p. 248). There is widespread hybridisation and introgression between G. rivale and G. urbanum in nature, resulting in a complex web of different forms—and this may explain some of the different results achieved in hybridization experiments.

Mendel defined G. intermedium as 'ABcDEe' and 'ABcdEe' (He made no reference here to the sixth trait he had identified, 'F/f'.). Three traits (ABc), then, are the same in both forms; one trait (D/d) is different. This refers to the flowers being erect (D) or drooping (d). Mendel made no reference to this trait in his letters to Nägeli. In English Botany (1864), it was stated that: 'This plant is readily distinguished from G. urbanum by the drooping flowers' (Syme p. 199); but in J.C. Röhlings Deutschlands Flora (1832), G. intermedium is described as having flowers which are either 'drooping or erect' (Mertens & Koch p. 551); Mendel may have observed the same. Louis Blaringhem found that, while some of his urbanum-rivale hybrids had drooping flowers, the majority were drooping at first, but later became erect (1920 p. 1286).

As for trait 'E/e', this refers to the fruit-bearing calyx (reflexed in G. urbanum and erect in G. rivale). The fact that Mendel used both letters together ('Ee') for G. intermedium implies that the relevant locus had two alleles. In other words, there would be four forms of G. intermedium, with nine possible genetic formulae thus:


DD and Dd would of course have the same phenotype since D is dominant over d (as would EE and Ee). Thus, the two forms of G. intermedium differing with respect to D/d (DD and Dd/dd) would each be constant with respect to all traits except one (E/e). (Mendel could only know this if he observed the plants over several generations.)

4. In a previous letter, dated 6 November 1867, Mendel had told Nägeli that the progeny of the G. urbanum × G. rivale cross looked like G. intermedium Ehrh. (OG p. 75.)

Most shared on

Human Origins: Are we hybrids?

On the Origins of New Forms of Life

Mammalian Hybrids

Cat-rabbit Hybrids: Fact or fiction?

Famous Biologists

Dog-cow Hybrids

Georges Cuvier: A Biography

Prothero: A Rebuttal

Branches of Biology

Dog-fox Hybrids