Hereditas 13: 342-356 (1930)
Inheritance of variegation and of black flower colour in Viola tricolor L.
J. Clausen


IN a previous paper (CLAUSEN 1927a) the author described three cases of albomaculatio in Viola, viz. one in V. tricolor, another in V. arvensis and the third one in a cross between these two species. Here the inheritance of the case in tricolor shall be dealt with.

In the paper mentioned the variegation was described as due to vegetative segregation for a plastid character, the leucoplasts in the white parts are defective and do not turn green. By the vegetative segregation pure green branches are formed and these give only green offspring, while variegated branches by self-fertilization give both green and variegated plants in various proportions. White seedlings were never observed, although a careful lookout was kept for them. All germinating plants have green cotyledons, variegation does not appear before later. The case was described as being in accordance with WINGE'S scheme, category II, A (WINGE 1919, p. 17), but it was not known, if the character was transmitted through both parents or not. The type of variegation is shown in fig. 1.

Later experiments have strengthened the view as to the type of inheritance. In order to track the expected white seedlings, a number of seeds from variegated branches were sown on tissue paper spread out on sand in flower pots. The paper was kept constantly moist by the capillary action of the sand, which took up moisture from below. In this manner record could be kept on all seeds. As will be seen from table 1 some seeds in all plots were empty, but not all the seeds with embryo germinated either. Apparently nothing was wrong with the embryos in the non-germinating seeds. They were swollen and white just as in all seeds, but it cannot be told what they would look like, if they had happened to germinate. All germinating seeds (a total of 162) had green cotyledons, none were white. A number of these green seedlings later on would turn variegated as shown by their sister sowings from the same lots of seeds (table 2). If genetically white embryos exist, they must be in the non-germinating seeds, but a reasonable explanation might also be that ovules on totally white parts of the plants are non-viable; in accordance herewith many seeds from strongly variegated branches are empty.

Fig. 1. Variegated V. tricolor, F2 ex variegated X green.

TABLE 1. Germination of Viola seeds from variegated branches sown on tissue paper (compare table 2).

  number of seeds in the sowings
V. 1207 V. 1208 V. 1212 total
germinated seeds 29 36 24 89
non-germinated, but with embryo 13 21 22 56
empty seeds 8 7 2 I 17
total 50 64 48 162

In order to investigate the mode of inheritance of the variegation of leaves two plants of the variegated type V. 1078 were crossed with a non-variegated, V. 1083-1. The variegated type was a V. tricolor alba with erect stems (pedigree shown in table 2), the non-variegated was a Viola tricolor of the maritima-type, not the original one (CLAUSEN 1926, p. 4), because this one suffered so much by inbreeding that it after some generations dropped out of culture. The new maritima, which not suffers from inbreeding, was extracted from the cross tricolor hortensis X maritima rosea (CLAUSEN 1926, Cross VI, p. 27), and V. 1083 was F4 of this cross. In its total appearance it is very similar to the var. maritima, having small, fleshy leaves and many prostrate, dark anthocyanous stems with long internodes. Then, furthermore, it has reddish coloured, narrow petals, the upper ones and the spurred one being velutina; the last named character is an inheritance from tricolor hortensis, which it also resembles in the character of not suffering from inbreeding. In the diagram, fig. 2, the gene symbols for the two types are given according to earlier investigations (CLAUSEN 1926, p. 12-14).

TABLE 2. Pedigree of variegated V. tricolor by self-fertilization.

V. 887-1 variegated (cf. CLAUSEN 1927 a, table 2)
V. 1078: (86 green: 7 variegated; total 93)
mother plant no. V. 1078-1 V. 1078-2 V. 1078-3
green strongly
no of sowing V. 1207 V. 1208 V. 1209 V. 1210 V. 1211 V. 1212
green 20  9 1 16 25
variegated 9 2 8 9
total 14 22 9 1 24 34

Two flowers of V. 1078-1 were used for pollination on V. 1083-1, one strongly variegated with many stripes on sepals and ovary and another more faintly variegated. V. 1078-2 was used as a female plant and pollinated by the green V. 1083-1; also here a strongly variegated and a less variegated flower were used (fig. 2). The F1 sowings, which had variegated as the female parent (V. 1213 and 1214), gave variegated and green plants as by self-fertilization of variegated (to a total of four green and six variegated plants). The two F1's with variegated as the male parent (V. 1215 and V. 1216), on the other hand, contained only green plants, namely 22 green plants in the two sowings together. The other characters of F1 were as expected from earlier crossings, namely violet non-velutina flower colour, and prostrate, anthocyanous coloured stems.

Three F1 plants (two variegated and one green) with variegated as the female parent were selfed, and also of the green F1 plants with a variegated male parent were three selfed. One of the variegated plants (V. 1213-1) had some totally green branches and these were selfed independently (see diagram, fig. 2).

Fig. 2. Diagram of the crossing variegated X green and reciprocally. The arrows indicate the direction, in which the pollen was carried. In some cases (V. 1078-1, V. 1078-2 and V. 1213-1) different branches of the same plant have been used in the experiments, of V. 1213-1 variegated and green branches and also some branches without any description at all. — varieg. = variegated, variegation. For explanation of gene symbols, see text pp. 349-350 and pp. 353-354.

Variegated plants and branches (V. 1330, V. 1332 and V. 1333, fig. 2) gave in F2 variegated and green plants to a total of 462 green and 50 variegated, while green F1 plants and green branches with variegated as the female (V. 1331 and V. 1334) yielded 563 plants, all green (see table 3). V. 1330—V. 1332 show difference between branches from the same F1 plant.

Only from two F1 plants with variegated as the male parent were seeds sown, namely as the sowings V. 1335 and V. 1336. The first one had 302 plants, all green, while the other, V. 1336, contained 658 green and 3 variegated plants. One of these variegated plants had a totally white main branch and green side branches, the other two had a white sector on only one or two of their leaves. The sowing, which contained these three variegated plants, happened to originate from pollen of a faintly variegated flower of V. 1078-1.

TABLE 3. Segregation in F2 ex variegated X green and reciprocally.

  variegated X green total green X variegated
(green F1)
variegated F1 green F1 V. 1335 V. 1336
green 462 563 1025 302 658 960
variegated 50 50 3 3
total 512 563 1075 302 661 963

This case seems to be an interesting transition between the cases of exclusively maternal inheritance (as in Antirrhinum, BAUR 1911) and those, where the variegation is transmitted through both eggs and pollen (Pelargonium, BAUR 1909). It is unexpected, indeed, that it should take more than one plant generation, before the defective plastids have been multiplied so much as to enable them to form cells with white chromatophores alone, but, on the other hand, it also seems unprobable that new mutation should take place at the same time in three plants of one sowing. A third possibility cannot be neglected, however, namely that a mutation took place already in one of the plastids of the mother plant of V. 1336 (V. 1216-1) and that, after multiplying of this, at least three ovules of V. 1216-1 received such defective plastids.

At the present, a decision between these three alternatives cannot be taken; the author thinks it most probable that the three variegated plants in V. 1336 really inherited the disposition for variegation from their male grandparent through a pollen grain, but this disposition is transmitted much more rarely through the pollen than through the ovules.

The mechanism in the Viola case then acts similar to what NOACK (1924) found in Pelargonium zonale albotunicata. In BAUR'S experiments, (1909), variegation was transmitted considerably from the male parent, (12,1 % variegated plus 1,9 % purely white F1 seedlings with variegated as the male parent; as contrasted with 27,4 % variegated seedlings with variegated as mother). NOACK (1924, pp. 56-57), on the other hand, found only 1,7 % variegated seedlings with variegated or white as the male parent, while variegated or white pollinated by green yielded 61,5 % variegated F1 seedlings, a considerable difference between crossings in the two directions. Just as in Viola, NOACK did not find any white seedlings in his Pelargonium hybrids.

There are, nevertheless, two or three points of difference between Pelargonium and Viola. In BAUR'S experiments, and in. NOACK'S as well, both of the reciprocal F1's contained green and variegated plants, but in Viola only the cross variegated X green showed segregation in F1, while the reciprocal cross, green X variegated, yielded a purely green F1 and only extremely few variegated plants in one of the F2's.

Furthermore, in the Pelargonium, according to BAUR, F1 plants of the crosses white X green and reciprocally really all are variegated, the cotyledons being mottled, while in Viola all F1 plants have purely green cotyledons, variegation appearing first later.

Finally, in Pelargonium the variegated or albotunicata plants sooner or later give either green or white branches, and especially in F2 many revert to either the white state and die, or to the green state. In Viola the recurrence to white is very rarely met with, and seeds on such branches generally are empty and not germinable, while it is not so uncommon to find purely green branches on variegated plants. In this way, if no selection were performed, the variegated type after some generations, no doubt, would disappear.

From this it appears that the vegetative segregation is effected much more slowly in Viola than in Pelargonium. Nothing is known about the approximate number of plastids in cells of the two genera. One should think that Viola with the more slow segregation should have a rather high number of plastids.

There is only scarce information regarding transmission of plastids by the male germ cells. BLISS (1912) studied the fertilization process in Viola cucullata and observed the fusion of the nuclei of the two male cells with the egg nucleus and the endosperm nucleus, but she does not tell, whether the sperm cells carry any cytoplasm at all. MADGE (1929) studied Viola odorata var. praecox GREGORY and found that the sperm cells before the discharging from the pollen tube are surrounded by a sheath of cytoplasm. This sheath probably is shed before the male nuclei enter the egg and the endosperm nuclei. A small portion of dense cytoplasm is told to be seen behind each of the male nuclei as they move towards the female nuclei. Apparently only the nuclei enter the female cells, but it is told that the end of the pollen tube is pressed against the egg, and a part of stainable material (from degenerated tube nuclei?) is discharged from the pollen tube during fertilization and surrounds the egg.

RANDOLPH (1922) stated that the proplastids of Maize sometimes were extremely small, almost beyond visibility and such small granules might escape observation, even if transmitted. MANN (1924) observed a great number of plastids in The microspores of Ginkgo biloba, about 140 are figured. CHITTENDEN (1927) more closely discusses the evidences for transmission of plastids by male germ cells but dares not take it for proven in any case.

In F2 of the crossing just described (V. 1330—V. 1336, a total of 2038 individuals) segregation took place also for a number of other genes, but all were ordinary Mendelian and in accordance with the general scheme for inheritance in Viola tricolor earlier described (CLAUSEN 1926). The types segregated were the following:

15 anthocyanine coloured: 1 alba,
3 non-velutina : 1 velutina (1 + 2),
3 violet flowered : 1 rose, and
3 prostrate : 1 erect.

In the earlier paper (CLAUSEN 1926, Cross III, pp. 20-22) an apparent case of repulsion between n and prostrate (P1 or P2) was found, although the genes for these went in together. The case corresponded to an apparent cross-over percentage of about 85, but no stress was put on this isolated case, also because this crossing was complicated by two polymeric genes for prostrate stems being present. I have not been able to make the reciprocal crossing: violet prostrate X rose erect because the type violet prostrate had lost its power of germination. The present crossing apparently should be a repetition of the first one: alba (with gene R for violet) erect X rose prostrate, but only one of the two polymeric genes for prostrate stems was found to be in action, and no correlation at all was found between the flower colour and the direction of the stem. Accordingly, the finding has not been verified, but not disproved either. I should estimate the case as uncertain, however. WELLENSIEK (1928) mentions a case of apparently more than 50 per cent of overcrossing in Pisum, but no stress can be laid on this case either, taking in consideration the highly varying percentage of overcrossing, together with other complications, in this species.


The darkest flower colour of any pansy is a dark velvety black. It is present in the so called V. tricolor maxima nigra (tricolor maxima hort. = V. Wittrockiana GAMS, a stabilized species hybrid, cf. CLAUSEN 1927 b, pp. 694-696). This large flowered type, which is used in gardens, is constant as regards flower colour.

In 1923 seeds were received from Cambridge Botanical Gardens of a Viola tricolor, which appeared to be such a black velutina, small flowered tricolor proper with n = 13 chromosomes (while V. Wittrockiana has about n = 24 and is irregular). The black flower colour was constant in the offspring from the first beginning. The darkest Viola type investigated before was the V. tricolor hortensis (CLAUSEN, 1926, Line 519, pp. 5-6) found to be recessive for three modifying and bleaching genes, which are present in the wild Viola tricolor. The pattern of tricolor hortensis is similar to the pattern of the flower shown as number four from the left, in the middle row of fig. 4.

The flower colour of Viola tricolor is determined by the cooperation of a number of bleaching and modifying genes, the darkest colours being the most recessive or hypostatic ones. Starting from tricolor hortensis (fig. 4, middle row, fourth flower from the left), the addition of one modifying gene removes the velvety colour from the two side petals (number three from the left), another gene removes the triangular velvety spot on the lower petal (number two, where only the two upper petals are velvety violet) and a third one removes velvety colour altogether, giving the wild type violet flowered Viola tricolor (the first flower from the left). This last gene is the most superior one in its effect; if it is present, the flower is non-velvety, irrespecting of the presence or absence of the two more inferior or hypostatic modifiers.

The addition of a lutea-gene in homozygous condition (LL) changes the violet flowered type to yellow (fig. 4, number one from the left in the lower row); the lower petal in this type is bright yellow and the four upper petals are bleached to yellowish white or (in older flowers) to a very light mauve tinge. A single dose (Ll) of this gene changes a violet flower to whitish or, in older flowers, to a pale violet. Finally, the yellow LL-type can be bleached to yellowish white, also on the lower petal, by the addition of a still more superior gene, (w), always present in Viola arvensis, rare in Viola tricolor proper, but often present in Viola tricolor subspec. alpestris.

In the absence of the modifiers for velvet, the L-gene cannot bleach the dark velvety colours; the flowers number two to four from the left in the lower row of fig. 4 have bright yellow lower petal, and still they show the velvety pattern, although it has been a little more sharply restricted than in the corresponding types of the row above. The types of the series in some way intergrade and the classification of them has not always been reliable, but the removal of the velvety colour from the petals follows a definite order. In some way it can be said that the three modifiers are cumulative in their effect, but they cannot well substitute one another. The absence of a reaction gene (R) changes all these colours towards different shades of reddish (rr).

Fig. 3. Diagram of the crossing V. tricolor, alba-lutea X nigra; for explanation of gene symbols, see text.

All these types of flower pattern cannot be realized if not any of a series of basic genes (A1, A2 or A3) for anthocyanine colours are present. In the absence of all these (a1a1a2a2a3a3) the colour is purely white, alba, (fig. 4, the flower in the right corner below); or it is alba-lutea (the upper flower to the right) if the lutea-gene is homozygously present. Thus the basic genes are basic for the anthocyanines only (red or violet), not for the flavones.

When these facts are borne in mind, it should be expected that the black flowered Viola tricolor nigra (fig. 4, upper flower to the left) should be extremely recessive, a Viola undressed for all the modifiers, genetically speaking exhibiting the innermost nucleus of the flower colours in Viola. From one point of view it is startling to expect a colour so dark to be very recessive to a light shade of violet; but the expectations were answered to, it proved to be extraordinarily recessive.

Fig. 4. Flowers of parents and F2 types ex V. tricolor, alba-lutea X nigra. Reading from the left, the types run as follows:

upper row: nigra, alba-lutea;

middle row: non lutea types, viz. non velutina, velutina 1, velutina 2, velutina 3 and velutina 4;

lower row: lutea types (LL), viz. lutea non velutina, velutina 1, velutina 2, velutina 3 (?), and alba (without lutea gene).

Viola tricolor alba-lutea (fig. 4, upper right flower) was selected to be crossed with nigra. This alba-lutea type was F4 ex tricolor lutea X tricolor alba (CLAUSEN 1926, p. 22, Cross IV) and a constant one. A diagram of the crossing is shown in fig. 3. The F1, V. 1217, consisted of only six plants and all these were anthocyanous and of a faintly violet flower colour with a whitish lower petal (owing to the bleaching effect of the lutea gene in single dose); the upper petals were also faintly velutina, showing that not enough modifiers were present for a total suppression of the velvet colour. The type of F1 is shown in the left flower of the upper row of fig. 5. Both of the parent types had n = 13 chromosomes, and the conjugation of the 13 pairs of chromosomes in meiosis of the hybrid was normal.

Three F1 plants were selfed and also used for backcrossing to nigra. A total of 1547 F2 plants and 317 plants of the backcrossings were grown. It was remarkable that among the 1547 F2 plants no real nigra plant appeared, the darkest type in F2 was still violet in the centre part of the flower (the right flower in the middle row of fig. 4).

On the other hand, this type was considerably darker than the tricolor hortensis (to the left for it, velutina 3) previously the darkest types analyzed.

In the backcrosses eight totally black flowered plants appeared among a total of 317 individuals. the segregated nigra type is shown in fig. 5, the bottom row to the right Apparently no difficulty was involved in distinguishing it from the type nearest to it. A comparison between fig. 5 and the parental type in fig. 4 shows how identical they look in colour. Nigra appeared in both directions of the backcrossings. Eight out of 317 is very near to one out of 32, which means that nigra in comparison with aiba-lutea is recessive for five modifying genes.

Also the alba type was rare, although by far not so rare as nigra. In F2 28 alba of 1547 individuals appeared, which is nearly one of 64 (table 4). Accordingly, nigra contains three pairs of polymeric basic genes for development of anthocyanine, a condition similar to V. tricolor hortensis, Line 519, and different from several wild growing varieties of V. tricolor, which have only two polymeric basic genes for anthocyanine.

Fig. 5. Flowers of types of the backcross tricolor, [alba-lutea X nigra] X nigra. From the left:. upper row: faintly velutina (similar to F1), velutina 1, velutina 2; lower row: velutina 3, velutina 4 and nigra. The flowers in the upper row are Ll, in the lower likely ll.

One fourth of the F2 individuals were homozygous for the lutea gene, having bright yellow lower petal (the four first flowers in the bottom row of fig. 4). Due to the bleaching effect of the LL-gene these cannot become nigra. As the expectation for nigra in F2 is as one among 1024 non-lutea and non-alba individuals, it is not so surprising that it was not found among 1084 of that category. In the backcrosses no plants can become homozygously lutea.

While one of the parent types did not occur at all in F2, one fourth of the alba individuals, should become alba-lutea. the other parent type, that is, it should occur once among each 256 plants. Only three were found, which is three below the expectation, but this type is not very vigorous in any case. In this way both of the parent types were extremely rare in F2 of this variety crossing.

The classification of the different shades of velutina was difficult in the present case, what is not to wonder about, keeping in mind how many modifying genes are in operation. Furthermore, the individuals, which have one lutea gene present (Ll-individuals), are of a lighter, more whitish shade on the lower petal than the true violet types (ll) without this gene. The types presented in figs. 4 and 5, of the F2 and the backcross respectively, must not be taken for more than approximate representations of types with one, two, three, four and five modifiers, of which the three superior ones have been analyzed in a number of earlier crossings.

The increased number of modifiers make a change in the terminology of these appropriate. In the earlier paper (CLAUSEN 1926) they were termined M1 to M3; M3 being the most superior or epistatic one. In view of the fact that the two new ones here analyzed are more hypostatic than any of the first three ones, it would appear more appropriate to begin with M1 for the most epistatic one, which by itself and in one step can bleach nigra to non-velutina violet, and continue to M5 that bleaches only the innermost part of the three lower petals (the right flower in the middle row of fig. 4); its effect cannot be seen, except when all other modifiers are absent. Beginning with nigra, the series of velutina types then will run as below:

velutina 5: m1m1m2m2m3m3m4m4m5m5


velutina 4: m1m1m2m2m3m3m4m4M5

 middle part of only three lower petals bleached;

velutina 3: m1m1m2m2m3m3M4

 a velutina edge left on three lower petals;

velutina 2: m1m1m2m2M3

 a velutina triangle left on the spur-bearing petal, the two side petals non-velutina;

velutina 1: m1m1M2

 only two upper petals velutina;

non-velutina: M1 

 all five petals non velvety.

TABLE 4. Segregation in the crossings alba-lutea X nigra.

segregating characters F2 F1 x nigra
observed calculated ratio observed calculated ratio
non alba
non lutea (ll + Ll) non velutina 673 |
      307,1 31
velutina 1 262 - - 91 |
velutina 2 102 1141,1 193347 135
velutina 3  37 - - }83
velutina 4  10 - -  
velutina 5
1,1 189 8 9,9 1
lutea (LL) non velutina 319 |
380,7 64512      
velutina 1 70
velutina 2 30
velutina 3-4 16
non lutea (ll+Ll)
lutea (LL)
total 1547 1547,0 262144 317 317,0 32
non nigra 1547 1545,9 4093 309 307,1 31
nigra 1,1 3 8 9,9 1
total 1547 1547,0 4096 317 317,0 32
with spot (on style) (S) 879 933,0 3 126 131,0 1
no spot (ss) 365 311,0 1 136 131,0 1
total 1244 1244,0 4 262 262,0 2
non lutea 1109 1160,2 3
lutea 438 386,7 1
total 1547 1547,0 4
anthocyanous 1519 1522,8 63
alba 28 24,2 1
total 1547 1547,0 64

The types can be identified in this order in the upper and the middle row of fig. 4 and partly also in the lower row, as well as in fig. 5. M1 acts on all five petals, while M2 to M3 act on the three lower petals only.

Table 4 gives the total segregation; also a dominant gene for spot on style (S) was in action in most of the sowings. The proportions between the different types do not always answer very close to the expectations, but in a cross so wide, with no less than nine genes segregating, it cannot be expected that all types are equally viable. The present experiment is an example of an extremely complicated interaction of a number of genes, all determining the colour of the flowers; and although some of the genes have a superior effect, most of them to some extent modify the effect of the other ones or even substitute them, what makes it difficult to classify many of the types with any certainty. Only the classifications of anthocyanous to alba, of non-lutea to lutea, of spot to non-spot and of non-nigra to nigra are reliable in this case.


(1). A case of variegation in leaves of Viola tricolor is inherited as a non-Mendelian character, giving vegetative segregation effected by the plastids. It is transmitted by the ovules giving segregation already in F1, and it is probably also transmitted through pollen, but in this case F1 was green and in F2 among a total of 963 only three individuals were variegated.

(2). The velvety black flower colour of Viola tricolor nigra is extremely recessive as compared with wild growing, non velvety types of Viola tricolor, which have a series of five modifiers or suppressors for the black velvety colour. On the other hand, tricolor nigra contains three dominant, polymeric, basic genes for the anthocyanine colours (violet or reddish). A crossing, involving nine genes, all cooperating in determination of flower colour, is described.

This investigation is part of a series carried out by a grant of the Carlsberg Foundation, for which the author expresses his thanks to the Trustees of the Foundation.


  1. BAUR, E. 1909. Das Wesen und die Erblichkeitsverhaltnisse der 'Varietates albomarginatae' hort. von Pelargonium zonale. - Zeitschr. in Abst.- u. Vererb.-lehre 1, p. 330.
  2. ——1911. Untersuchungen über die Vererbung von Chromatophorensnerkmale bei Melandrium, Antirrhinum und Aquilegia. - Ibid. 4, p. 81.
  3. BLISS, MARY C. 1912. A contribution to the life history of Viola. - Ann. Bot. 26, p. 155.
  4. CHITTENDEN, R. J. 1927. Vegetative segregation. - Bibl. Gen. 3, p. 355.
  5. CLAUSEN, J. 1926. Genetical and cytological investigations on Viola tricolor L. and V. arvensis MURR. - Hereditas VIII, p. 1.
  6. ——1927 a. Non-Mendelian inheritance in Viola. - Hereditas IX (Festskrift för W. JOHANNSEN 19 3/2 27). p. 245.
  7. ——1927 b. Chromosome number and the relationship of species in the genus Viola. - Ann. Bot. 41, p. 677.
  8. MADGE, A. P. 1929. Spermatogenesis and fertilisation in the cleistogamic flower of Viola odorata var. praecox GREGORY. - Ann. Bot. 43, p. 545.
  9. MANN, MARGARET C. 1924. Microsporogenesis of Ginkgo biloba L. with special reference to the distribution of the plastids and to cell wall formation.. Univ. Calif. Publ. Agric. Sci., 2, p. 243.
  10. NOACK, K. L. 1924. Vererbungsversuche mit buntblättrigen Pelargonien. Verh. d. physik.-med. Ges. Würzburg 49, p. 45.
  11. RANDOLPH, L. F. 1922. Cytology of chlorophyll types in Maize. - Bot. Gaz. 73, p. 337.
  12. WELLENSIEK, S. J. 1928. Pisum-crosses III. - Genetica 11 p. 225.
  13. WINGE, O. 1919. On the non-Mendelian inheritance in variegated plants. C. R. Tray. Labor. Carlsberg 14, No. 3.