Genetica Vol. 48, 3: 171-183, 1978
Department of Agricultural Botany, University College of Wales, Aberystwyth, U.K.
Received July 27, 1977/Accepted October 25, 1977
*Present address: Department of Botany, Rajshahi University, Rajshahi, Bangladesh

Reciprocal crosses were made between large and small genotrophs differing in the capsule character H-h (hairy-hairless septa) of the variety Stormont Cirrus, thus bringing together two levels of regulation, one determining H and the other h in the Hh heterozygotes. HH, Hh and hh types appear in all generations from F2 to F5 but the Hh plants can be grouped in up to four classes with about 20, 30, 40 and 50 hairs per septum. A proportion of F2 and F3 Hh plants are switched to HH in some environments and to hh in others, but in later generations more stable Hh heterozygotes appear which are unaffected by the environment and give normal ratios undisturbed by the Hh class from which they are descended, either because they have been selected for, or because they have been built up, over generations. HH plants in F2 are enhanced, i.e., they have a higher mean hair number than HH parents, and give more Hh plants in the highest hair number class when backcrossed to hh than do Hh parents crossed with hh.


Two types of plants, the large genotroph (L) and the small genotroph (S), induced by growing plants of the flax variety Stormont Cirrus in specific environments (Durrant, 1962) differ in several characters as well as in plant size. L has 16 per cent more nuclear DNA, due to alterations in a range of repeated gene sequences (Cullis, 1973; Timmis & Ingle, 1974), and about 60 per cent more ribosomal genes (Timmis & Ingle, 1973, 1975; Cullis, 1976). They differ in isozymes (Cullis & Kolodynska, 1975), and in a capsule character, hairless septa, h (Fig. 1), S having about 60 or more hairs per false septum and L none (Durrant & Nicholas, 1970; McLellan & Durrant, 1973). L and S are stable and breed true in most environments in all characters, as far as they have been studied, except total amount of DNA which can be prescribed and altered up to certain inducible limits by altering the environment (Durrant & Jones, 1971; Joarder et al, 1975). The initial changes in DNA induced by the environment in the original variety occur gradually and can be measured during the first five weeks of induction (Evans, 1968), confirming that they are environmentally induced changes.

Other large and small types have been induced at various times and all are referred to as L and S genotrophs, but they vary to some extent, i.e., the L genotrophs are not identical in all characters, nor are the S genotrophs. Another pair used in these experiments, for example, had only a 10 per cent difference in DNA amount, and the capsule character was in the other phase, i.e., L had hairy septa and S was hairless.

The differences between L and S must be due to differences in gene activity and regulation, induced by the environments, which are maintained thereafter through mitosis and meiosis, and they present opportunities for studying gene regulation by crossing and selection which are more usually the prerogatives of studies on classical gene differences: The capsule character is particularly suitable because, paradoxically, H-h behaves more like a mendelian factor, often giving ratios close to expectation (Durrant & Nicholas, 1970; McLellan & Durrant, 1973). Scoring can be made on large numbers of plants crowded in a greenhouse allowing more than one generation a year, and Hh heterozygotes, if they are indeed heterozygotes, can be distinguished from HH homozygotes by the number of hairs per septum. The studies also showed that in the F2, Hh plants give a distinct trimodal distribution (Fig. 2a) with modes of 30, 40 and 50 hairs per septum. The differences in hair number between septa within capsules, or between capsules on the same plant, are relatively small. F2 Hh plants with 40 hairs give F3 families with approximately normal 3:1 ratios, but those, with 50 hairs give a large excess of H plants in F3 due to H←h changes. It became apparent that the different ratios were part of a much wider pattern which has since been studied more intensively, and the results are described here.

Material and methods

Two sets of large and small genotrophs were used. One set was Lh and SH, where the large genotroph was homozygous hairless hh and the small genotroph homozygous hairy HH, and the other set was LH and Sh, i.e., the large genotroph was homozygous hairy Hh and the small homozygous hairless hh. Reciprocal crosses between Lh and SH, and between LH and Sh, were made in 1968 and seed for the initial experiments was collected from F1 plants in 1969. The parent and F1 plants were grown in 18 cm pots with John Innes compost in a warm greenhouse for the first five weeks and then placed outside. Other plants were similarly reared for additional crosses and backcrosses later.

Fig. 1. Flax capsules cut in half to show left, hairless false septum (hh) of the L genotroph and, centre, hairy false septum (HH) of the S genotroph with about 60 hairs. False septa are dissected out for hair counting and the one on right is from a (Hh)4 heterozygote with 42 hairs. The hairs are approximately one mm long.

Larger populations for scoring hair number in the F2 were grown from sowing to maturity in 13 cm pots with John Innes compost, ten plants to a pot, in an unheated greenhouse in the summer, moderately heated during the rest of the year. In these conditions the plants are tall, thin, unbranched with a few capsules at the top. The F3 was similarly grown, and so on to F5. A larger number of F1 plants was grown in 13 cm pots in a later year. There were no specified sowing times but about three months elapsed between collecting ripe seed and sowing. General cultural conditions have been described (Durrant, 1961, 1970; Joarder et al, 1975).

Fig. 2. Examples of hair number distributions. (a) F2 trimodal Hh distribution and bimodal HH distribution in 1971. (b) F4 quadrimodal Hh distribution and bimodal HH distribution in 1972. (c) F4 distribution in 1971 showing no separation into heterozygous or homozygous classes. In (a), (b), (c), hh plants are not shown. In (a), (b), plants with more than 54 hairs are homozygous HH; both sets of homozygotes are enhanced with a main peak at 60 and a smaller one at about 70 hairs per septum. (d) HH parents. (e) Enhanced HH plants in F3 from F2 (Hh)5 showing extended tail containing higher proportion of plants with about 70 hairs per septum.

F1 to F5 and backcrosses were sown in different years and months (Table 1). The F2 was grown three times, each in a different year but all were derived from a common stock of F1 seed collected in 1969. The F3 was grown in three different years from seed from F2 plants grown in 1970. The F4 was grown in 1971 and 1972 from F3 plants grown in 1970, and also in 1972 from 1971 F3 plants. The F5 was grown in 1972 from 1971 F4 plants. The F1 was grown once for scoring in 1972 from an additional set of crosses in 1971 and the F2, F4 and HH backcrosses had their separate crossing programmes.

Plants were classified homozygous hairless (hh), homozygous hairy (HH) or one of the heterozygous classes (Hh) described later, any plant with more than 54 hairs per septum being judged HH (Fig. 2a, b). Germination was over 95 per cent. Individual and joint χ2s testing the heterogeneity of the reciprocal crosses and the types of crosses, Lh and SH on the one hand, and LH and Sh on the other, with regard to the numbers of plants in the homozygous and heterozygous classes were not significant, allowing for deviations expected in a distribution of nearly 700χ2s. Consequently reciprocals and types of crosses are combined in tables and figures and reference is made only to crosses between HH and hh parents, accepting at the same time that this difference is associated with other differences in phenotype and gene regulation.

Table 1
Sowing times of generations and backcrosses. The number of Hh classes in each is given in brackets.

  1970 1971 1972
F1     May (4)
F2 April (3) July (3) May (4)
F3 October (4) July (4) May (4)
F4   July (1) July (4)
F5     June (4)
F2 Backcrosses     May (4)
F4 Backcrosses     May (4)
HH Backcrosses     October (3)


In previous studies, Hh plants fell into three distinct classes (Fig. 2a) having modes of approximately 30, 40 and 50 hairs per septum which were symbolised I, II and III. In the generations and backcross generations grown here quadrimodal distributions were more common, with the appearance of an extra class with 20 hairs (Fig. 2b). The environment in which the generation is grown determines the kind of distribution because F2 plants grown from a common seed stock have a 4-modal distribution in 1972 and 3-modal in the two previous years (Table 2). To accommodate the extra class, and to allow for the possible future appearance of a fifth class with 10 hairs, the Hh classes with approximately 20, 30, 40 and 50 modes are symbolised here as (Hh)2, (Hh)3, (Hh)4 and (Hh)5 respectively.

The frequencies with which the different Hh classes occur can be used to measure the average H potential of the Hh plants in each generation. A generation containing more Hh plants in the (Hh)5 class and fewer in the (Hh)3 class, for example, is considered to have a higher H potential, i.e., it has a higher level of regulation for H, than one containing more Hh plants in the (Hh)3 class and fewer in (Hh)5. This was measured by k, which is calculated from the regression of the frequencies of (Hh)3, (Hh)4, (Hh)5 onto χ values 1, 2, 3 respectively, or the frequencies of (Hh)2, (Hh)3, (Hh)4, (Hh)5 onto 1, 2, 3, 4, multiplying the regression coefficient by 100, and dividing by the total number of Hh plants so that families containing different numbers of individuals could be compared. A family with a positive k value has more Hh plants in the higher hair number class and is said to have a high H potential. One with a negative k value has more Hh plants in the lower hair number classes and has a low H potential.

Generations F1 to F5

The heterozygous classes of the F2 generations (Table 2) have typical mean hair numbers of approximately 30, 40, 50 in 1971 and 20, 30, 40, 50 in 1972 (mean ≅ mode), but in 1970 the environment has exceptionally increased these values by approximately six hairs per septum. The phenotypic ratio H/h the heterozygote/homozygote ratio Hh/(HH+Hh), the homozygote ratio HH/hh, and the k value are given for each year.

In 1970 and 1971 there are significantly more heterozygotes than homozygotes, giving H/h ratios with significant excess of H plants. In 1972 there is a significant excess of both homozygotes, presumably due to the dispersion of heterozygotes in both directions due to Hh HH and Hh hh changes.

The three F3 generations grown on separate occasions (Table 1) were not significantly different and their class frequencies were combined. About 750 progeny of each of the three F2 (1970) Hh classes were grown (Table 3), plus smaller numbers of progeny of HH and hh F2 plants which bred true. There were highly significant differences between F3 families from the different F2 Hh classes, and their ratios and k values were correlated with the hair numbers of the F2 Hh classes. F2 (Hh)3 plants give an F3 H/h ratio of 1.79 with significant excess of h plants, and F3 heterozygotes with a k value of -7.35 (low H potential). F2 (Hh)5 plants give an F3 H/h ratio of 4.70 with significant excess of H plants and k of 4.75 (high H potential). F2 (Hh)4 plants give intermediate values in F3. These differences are highly significant. Hence F2 Hh plants with low hair number give rise to frequent H ® h changes and heterozygotes with more plants in the lower hair number classes. F2 Hh plants with high hair number give H¬h changes and more Hh plants with high hair number. There are significantly more heterozygotes than homozygotes in the F3 families from each F2 Hh class, which was consistent over the three years combined in Table 3. Overall they are shifted towards a low H potential with H/h ratios of 1.8, 2.4 and 4.7 respectively for the descendants of the F2 (Hh)3, (Hh)4 and (Hh)5 classes. In one previous experiment (Durrant & Nicholas, 1970), they were shifted towards a higher H potential with F3 Hh ratios of 3.3, 3.5 and 6.8 respectively; in another they were more intermediate with ratios 2.3, 3.0 and 5.0.

Table 2
Numbers of plants and mean hair numbers in F2 generations grown in three different years. Ratios and k values are given below.

of plants
Mean hair
of plants
Mean hair
of  plants
Mean hair
Hh 79 0 86 0 99 0
(Hh)2         32 21.5
(Hh)3 22 37.2 46 30.5 34 30.7
(Hh)4 99 46.5 102 41.2 42 40.7
(Hh)5 81 55.7 82 50.5 40 49.5
HH 79 60+ 94 65.0 123 66.5
H/h 3.56   3.77   2.74  
Hh/(HH+hh) 1.28   1.28   0.67  
HH/hh 1.00   1.09   1.24  
k 14.60   7.26   2.16  

Table 3
Numbers of plants and mean hair numbers in F3 generations grown from the three F2 Hh classes.
Combined results over three years. Ratios and k values are given below.

F2 (Hh)3Number of plants
Mean hair number From
F2 (Hh)4Number of plants
Mean hair number From
F2 (Hh)5Number of  plants
Mean hair number
Hh 268 0 215 0 161 0
(Hh)2 157 19.1 113 19.1 78 19.4
(Hh)3 117 30.6 112 30.7 94 30.2
(Hh)4 83 40.6 104 41.2 116 40.3
(Hh)5 65 49.7 95 50.1 138 49.9
HH 59 62.0 103 64.5 169 65.9
H/h 1.79   2.45   4.70  
Hh/(HH+hh) 1.29   1.33   1.29  
HH/hh 0.22   0.48   1.05  
k -7.35   -0.46   4.74  

F4 descendants of the 12 combinations of the three F2 Hh classes and four F3 Hh classes, were grown in 1971 and 1972. One F4 generation in 1971 from 1970 F3 plants gave no heterozygous classes, nor could a separation be made between heterozygotes and HH homozygotes (Fig. 2c). This is due to the 1971 environment because a second F4 generation grown in 1972 from the same 1970 F3 seed stock gave a 4-modal Hh distribution (Table 4). The unimodal distribution of 1971 probably arises from a contraction and merging of the Hh and HH classes because it extends over a wide range and, in other experiments, plants with different hair numbers taken from a similar type of distribution in F2 gave F3 H/h ratios correlated with the F2 hair numbers. A third F4 generation from 1971 F3 seed gave virtually the same class frequencies as the second F4 generation from 1970 F3 seed with which it was grown in 1972, and the data of these two generations were combined. There were no significant interactions between F2 and F3 contributions to any of the three F4 generations so that only the main effects of F2 and F3 and their F4 descendants are given in Table 4.

Although the Hh. classes are blurred in the 1971 F4 generation the H potential of their progeny can be assessed from their mean hair number because this would be greater, for example, if there were more plants in the higher hair number classes, i.e., a larger positive k value. Mean hair numbers for 1971 are therefore given in Table 4, and the k values and H/h ratios for the two generations combined in 1972. The F2 classes have little effect on F4. The F3 classes have more effect but the differences between the k values and H/h ratios are still small compared with the differences in F3 due to the F2 classes (Table 3). Table 4 gives the overall frequencies for the two F4 generations in 1972; the ratios are mendelian, and there are practically the same number of plants in each of the Hh classes, but the class means, or modes, are the same as in F2 and F3. The F4 appears to be a more stable generation than the F2 or F3.

Table 4a
Mean hair numbers of F4 generation grown in 1971, k values and H/h ratios of F4 generations grown in 1972,
from F2 Hh classes summed over F3 Hh classes, and from F3 Hh classes summed over F2 Hh classes.

F2 (Hh)3
F2 (Hh)4
F2 (Hh)5
F3 (Hh)2
F3 (Hh)3
F3 (Hh)4
F3 (Hh)5
Mean hair number 1971 41.8 41.8 44.7 39.7 43.5 42.8 45.1
k 1972 -0.73 -0.38 -0.45 0.03 -0.86 1.21 0.21
H/h 2.96 3.16 3.01 2.53 2.95 3.19 3.45

Table 4b
Numbers of plants, mean hair numbers and ratios of all 1972 F4 plants summed over F2 Hh and over F3 Hh classes

  F4 classes
  Hh (Hh)2 (Hh)3 (Hh)4 (Hh)5 HH
Numbers of plants 720 396 385 380 371 710
Mean hair number 0 19.7 29.9 40.4 49.9 62.8
H/h = 3.11 Hh/(HH+hh) = 1.07 HH/hh = 0.99 k = -0.52

The F5 generation was grown from four groups of F4 plants with approximately 20, 30,40 and 50 hairs per septum respectively, taken from each of the twelve 1971 distributions descended from the twelve combinations of the three F2 and four F3 heterozygous classes. F5 progeny of another group of F4 plants with 60 hairs per septum bred true and were HH.

There were no trends or significant differences in the F5 due to the four F4 groups, nor were there significant interactions between the F2 and F3 contributions, and the main effects only of F2 and F3 contributions to F5 are given in Table 5. There is a trend in the k values due to the F3 classes but the differences are small compared with those in F2 and F3 generations (Tables 2 and 3), otherwise the F2 and F3 classes have no significant effect on F5. Summing over the data, all ratios are mendelian, there are equal numbers of plants in the four Hh classes and the mean hair numbers are the same as before. F4 and F5 give similar results.

327 F1 plants grown in 1972 gave a quadrimodal Hh distribution (Table 6) with a low average H potential (k = -11.22). The mean hair numbers of the classes are compressed with a range of 23 instead of 30 in other Hh distributions, which is probably due to the environment rather than to a characteristic of the F1.

F2 and F4 backcrosses

Hair number can reliably be scored only in the ripe capsule so that an excess of F2 and F4 plants were backcrossed in 1971 reciprocally to ensure that all homozygous and heterozygous classes were represented. Backcrosses were subsequently grown from F2 and F4 plants selected according to their homozygous or heterozygous class as judged by hair number. Checks were made on the F2 and F4 plants which contributed to the backcross generations by growing their progeny from selfing. F2 and F4 plants from reciprocal crosses between Lh and SH only were used because of the large excess of backcrossing required. 4-modal Hh distributions were obtained in F2 and F4 backcrosses to both parents and in all reciprocals. Chromosomes or gametes from F2 and F4 are denoted H' and h' to distinguish them from H and h chromosomes and gametes from the HH and hh parents to which they were backcrossed.

Table 5a
H/h ratios and k values of F5 generations, summed over all F4 Hh classes, from F2 Hh classes summed over F3 Hh classes, and from F3 Hh classes summed over F2 Hh classes.

F2 (Hh)3
F2 (Hh)4
F3 (Hh)5
F3 (Hh)2
F3 (Hh)3
F3 (Hh)4
F3 (Hh)5
H/h ratio 2.72 3.03 3.00 2.54 3.15 3.14 2.85
k -0.44 1.76 0.35 -1.50 -0.13 0.27 2.65

Table 5b
Numbers of plants, mean hair numbers and ratios of all F5 plants summed over F2, F3 and F4 Hh classes

  F5 classes
hh (Hh)2 (Hh)3 (Hh)4 (Hh)5 HH
Numbers of plants  434 205 221 220 221 396
Mean hair number 0 19.57 30.37 40.84 49.96 65.48
H/h = 2.91   Hh/(HH+hh) = 1.04   HH/hh = 0.91   k = 0.54


In the F2 backcrosses to HH parents (Table 7) the three F2 H'h' classes, (H'h')3, (H'h')4, (H'h')5, gave Hh'/HH' ratios of 1.43, 1.15 and 0.86 respectively, and k values of -5.8, -3.8 and 7.4, i.e., there is a close correlation between the hair number of the F2 (H'h') plants, the frequencies of Hh' and HH' plants given by their backcrosses, and the frequencies of plants in the heterozygous classes of the Hh' plants. For example, F2 (H'h')3 plants, with the lowest hair numbers, give more Hh' than HH' plants, and among the Hh' plants there are more plants in the lower hair number classes.

The F2 backcrosses to hh parents give a similar picture. Here, for example, F2 (H'H')3 plants, with the lowest hair numbers, give more h'h than H'h plants, and again among the H'h plants more plants occur in the lower hair numbers classes.

The F2 backcrosses therefore give similar results to selfing F2, for F2 H'h' heterozygotes with low hair number give H' h' changes in the backcrosses to both parents and more H'h or Hh' plants in the lower hair number classes. F2 H'h' heterozygotes with high hair number give H' h' changes and more H'h or Hh' plants in the higher hair number classes.

The backcross genotypes are shown (Table 7) assuming that only H' and h' gametic contributions of F2 H'h' change, i.e., H' h' or H' h'. In some cases the H or h gametic contributions of the parents, HH or hh, could have been changed by the contributions of the H' or h' gametes on formation of the heterozygote instead of H' or h' themselves changing. Hence excess heterozygotes in the backcross of (H'h')3 to HH parents may be due to H'H H'h rather than H'H h'H given above, and excess heterozygotes in the backcross of (H'h')5 to hh parents due to hh' Hh' rather than hh' hH'. Since the frequencies of these alternative events cannot be determined the genotypes are left in the form given. Excess homozygotes can only arise by changes in the contributions of the H' and h' gametes from F2 (H'h') plants to give H'h h'h, and Hh' HH'.

Table 6
Numbers of plants in the Hh classes, and the mean hair number of each class, in F1

  (Hh)2 (Hh)3 (Hh)4 (Hh)5
Number of plants 110 145 54 18
Mean hair number 22.7 30.9 39.1 47.4

Table 7
Numbers of plants in the F2 backcrosses to hh and HH parents, with k values and ratios. Contributions from F2 plants denoted H', h'

F2 backcrosses to hh
F2 plants back-crossed h'h (H'h)2 (H'h)3 (H'h)4 (H'h)5 k H'h/h'h
(H'h')3 81 24 16 13 10 -7.1 0.78
(H'h')4 33 12 18 13 7 -4.0 1.52
(H'h')5 24 6 7 12 19 10.0 1.83
H'H' - 26 32 47 41 4.1 -
F2 backcrosses to HH
F2 plants back-crossed (Hh')2 (Hh')3 (Hh')4 (Hh')5 HH' k Hh'/HH'
h'h' 70 88 40 27 - -7.9 -
(H'h')3 28 20 19 13 56 -5.8 1.43
(H'h')4 11 12 9 7 34 -3.8 1.15
(H'h')5 5 6 8 12 35 7.4 0.86

The backcrosses of F4 plants descended from F2 (Hh)3 and F3 (Hh)2 (heterozygous ancestors with lowest H potential) and from F2 (Hh)5 and F3 (Hh)5 (heterozygous ancestors with highest H potential) were grown together with the F2 backcrosses. There were no significant differences between the numbers of plants in the backcross Hh classes due to the F4 Hh hair number classes, or to their ancestral F2 and F3 classes, nor were there any significant correlations, and the frequencies of the backcross Hh classes were summed over all F2, F3 and F4 classes (Table 8).

In addition to the lack of correlation, the F4 backcrosses differ from the F2 backcrosses in another respect. F4 H'h' plants backcrossed to hh parents give ratios with excess H'h or Hh' plants over mendelian expectation, and this could only occur by h'h → H'h or h'h → h'H changes, i.e., h' gametes from F4 H'h' plants have an associated positive, or H determining, potential because they either change to H' or they cause h from the hh parents to change to H. Similarly in backcrosses to Hh parents, F4 H'h' plants give ratios with excess Hh' or H'h plants which could occur only by H'H → h'H or H'H → H'h changes, i.e.,

H' gametes from F4 H'h' plants have an associated negative, or h determining, potential because they either change to h' or cause H from HH parents to change to h. The relative frequencies of H → h and h → H changes on the one hand and of H' - h' and h' -+B' on the other are unknown and the Hh genotypes in Table 8 are written as normal backcross expectations, i.e., in the present instance, H'h in the backcrosses to hh, and Hh' in the backcrosses to HH. The F4 backcrosses to both parents together give 30 per cent more H'h or Hh' heterozygotes over mendelian expectation (P <0.1 per cent) whereas there is no excess of Hh plants in F5 (Table 5).

The F2 and F4 backcrosses are alike in one respect. H' gametes from F2 and F4 H'H' plants give H'h heterozygotes with positive k values (high H potential) when backcrossed to hh parents, and h' gametes from F2 and F4 h'h' plants give Hh' heterozygotes with negative k values (low H potential) when backcrossed to HH parents, although the values are larger in the F2 backcrosses (4.1, -7.9) than in the F4 backcrosses (2.0, -4.4).

HH homozygotes

The mean hair number per septum of HH parent plants, LH and SH, varies with the environment from about 60 to 63. HH homozygotes in the F2 and F3 of crosses between HH and hh have significantly higher, or enhanced, values up to 66 or 67 (Table 9). Selfing of enhanced HH plants gives HH progeny with lower hair number, and after three generations of selfing the hair number returns to approximately the original HH parental value (Table 9). There are less data on HH plants in later generations of crosses between HH and hh but the trend appears to be the same.

Table 8
Numbers of plants in the F4 backcrosses to hh and HH parents, with k values and ratios.
Contributions from F4 plants denoted H', h'. The ratios show excess heterozygotes in both backcrosses

F4 backcrosses to hh
F4 plants back-crossed h'h (H'h)2 (H'h)3 (H'h)4 (H'h)5 k H'h/h'h
H'h' 86 20 26 33 32 (3.9) 1.29
H'H' - 23 24 30 28 (2.0) -
F4 backcrosses to HH
F4 plants back-crosses (Hh')2 (Hh')3 (Hh')4 (Hh')5 HH' k Hh'/HH'
h'h' 42 47 33 27 - (-4.0) -
H'h' 38 45 21 27 79 (-4.4) 1.66

The distribution of hair numbers among HH parents is skew with a mode of 60 and a tail towards higher hair numbers (Fig. 2d). Enhancement does not shift the mode but increases the relative frequencies of plants in the tail (Fig. 2e). In larger samples of enhanced HH plants a second less distinct mode appears at about 70 hairs per septum, but it is variable and perhaps may be vaguely discerned in large samples of HH parental populations in some environments. The subdivision of the RH distribution into 60 and 70 hair number classes is more subjective than the subdivision of Hh plants into classes and therefore the mean hair number of all plants in each HH distribution is given in Table 9, although this is a less sensitive estimate of the amount of enhancement. A mode of about 70 is visible in the distributions in Fig. 2a, b, c.

Table 9a

Hair numbers of enhanced HH plants. Mean hair numbers of HH parents, (HH)0, and of HH plants homozygous for 1, 2, 3 and 4 generations (HH)1. (HH)2, (HH)3, (HH)4, after 1, 2, 3 or 4 generations heterozygous Hh. Homozygotes in F2 (i.e., after one generation heterozygous in F1) were selfed for two, or three, generations on three occasions, (a), (b), (c).
Number of previous
generations heterozygous
(HH)1 (HH)2 (HH)3 (HH)4 (HH)0
1 (a) 65.10 63.38 61.69 - 61.88
(b) 66.38 65.95 64.21 63.87 62.96
(c) - 66.35 65.25 61.79 62.90
2 65.20 64.30 63.48 - 62.96
3 64.49 64.44 - - 62.96
4 65.86 - - - 62.96

Table 9b

Mean hair numbers of HH plants in F3 descended from the three F2 Hh classes and grown on three occasions
F2 (Hh)4
F2 (Hh)5
1970 61.5 62.7 64.1 60.2
1971 61.1 64.0 65.4 61.8
1972 63.5 66.9 66.0 62.9
Mean 62.0 64.5 65.2 61.6

The enhancement of HH in F3 is significantly influenced by the F2 Hh class from which the F3 is descended (Table 9), progeny of F2 (Hh)5 having the greatest amount and those of F2 (Hh)3 the least, or none. Selection of plants with about 60, 60-70 and 70 hairs per septum gave similar progeny with a principle mode of 60 and no significant differences between means, so that the reversion on selfing enhanced HH is unlikely to be due to selection.

HH parents, denoted (HH)0, and HH plants homozygous for 2, 3 and 4 generations, (HH)2, (HH)3,(HH)4, after one generation heterozygous in F1, were crossed with hh parents, (hh)0, shown in the first column of Table 10. They were also crossed at the same time to hh plants homozygous for 2 and 3 generations, (hh)2, (hh)3, after one generation heterozygous in F1 (columns 3 and 4, Table 10a). All the progeny were grown together, and all were heterozygous as judged by their hair number, each family being trimodal with mean hair numbers of approximately 30, 40 and 50 hairs per septum.

Table 10a

Backcrosses of enhanced HH plants, k values of heterozygotes obtained from crossing HH parents, (HH)0, and HH plants homozygous for 2, 3 and 4 generations, (HH)2, (HH)3, (HR)4, after one generation heterozygous in F1 , with hh parents, (hh)0, and hh plants homozygous for 2 and 3 generations, (hh)2, (hh)3, after one generation heterozygous in F1. Backcrossed F2 HH plants, (HH)1, to (hh)0, and backcrossed F2 (hh)1 plants to (HH)0, are fitted into the table (in italics) from Table 7.
  (hh)0 (hh)1 (hh)2 (hh)3 Mean
(HH)0 -4.7 -7.9 -7.5 -7.9 -6.7
(HH)1 4.1 - - - -
(HH)2 17.4 - 19.6 17.7 18.2
(HH)3 12.1 - 16.3 17.6 15.4
(HH)4 -4.6 - -8.3 -6.9 -6.6
Mean 5.1 - 5.0 5.1  

Table 10b
Hh class frequencies summing over all progeny of (HH)0 and (HH)4 combined, and all progeny of (HH)2 and (HH)3 combined

Class frequencies

  (Hh)3 (Hh)4 (Hh)5
114 219 70
65 85 165

The k values have a simple pattern. The different origins of hh have no effect. HH parents give progeny with negative k values of about -7.0, which are comparable with the F1 k value of -11.2 (Table 6). HH plants homozygous for two generations have a high positive k value of 18.2, after which there is a fall with continued selfing to return to a value similar to that given by HH parents. k values of F2 HH and F2 hh, backcrossed to hh and HH parents respectively, calculated in the F2 backcross experiment (Table 7), can be fitted into Table 10, and are shown in italics.


Because of the origin and behaviour of h in the flax genotrophs and their crosses, the difference between HH and hh in the parents must be due to a difference in regulation, and it is inferred that h is H repressed. Yet in many respects H-h behaves as a mendelian gene. HH and hh parents breed true; all F1 plants are heterozygous as judged by their hair number; F1 plants give HH, Hh and hh plants; Hh x HH backcrosses give only HH and Hh plants; Hh x hh backcrosses give only Hh and Hh plants. In earlier generations of crosses between HH and Hh, however, the ratios are frequently non-mendelian, signifying that in a proportion of the plants H h or H h changes occur, whereas in later generations the ratios are mendelian. This trend is summarised in Fig. 3a. It argues for an early disruption of regulation followed by a return to stability.

Three or four heterozygous classes appear in all generations which, with Hh and two HH classes, give seven phenotypes, indicating several levels of gene activity and the compound nature, or multiplicity, of the factors involved. The distribution of plants among the Hh classes changes from a skewed normal distribution in earlier generations (F1 heterozygotes agree with (0.7 + 0.3) and (0.75 + 0.25) binomials, but not with a Poisson) to a more linear relationship in F3, and to equal frequencies in F4 and F5 (Fig. 3a). A random distribution of events may determine the distribution of plants among the Hh classes in the earlier generations whereas in later generations each class has an equal probability of occurrence. The hair numbers of the F2 Hh classes are correlated with the H : h ratios of their F3 families, but these parent/offspring correlations practically disappear in later generations (Fig. 3a). The Hh classes themselves cannot therefore be entirely responsible for the H h changes in the next generation, rather their frequencies in the earlier generations reflect the possibly random distribution of events producing the genetic changes in the next. F2 (Hh)2 plants, for example, with low hair number may be nearer one tail of the distribution of controlling elements, or regulatory patterns which eventually switch H h. The apparently random distribution in the earlier generations of the cross could be due to the disruption of the pre-existing regulations in the parents, which appear to stabilise once more in later generations and become uniformly distributed at certain levels among the heterozygotes, and at other levels respectively in the homozygotes. In nearly all F2 and F3 generations (11 out of 12) there are significantly more heterozygotes (21 per cent) than homozygotes, which suggests that the intermediate regulation of the F1 tends to persist.

Fig. 3. (a) Selection of Hh heterozygotes. The F2 generation and plants of F3, F4 and F5 generations from Hh heterozygotes only of each previous generation. --Mean k values of Hh plants ignoring sign. --- Mean deviations x 10 from mendelian ratios ignoring direction. (b) Selection of HH homozygotes. HH parents, F2 HH plants, and F3, F4 and F5 HH plants from HH homozygotes only of each previous generation, shown as the number of generations homozygous after one generation heterozygous in F1 (HH)0 are the HR parents. -.-.-. Mean hair number. —Mean k values of Hh plants in crosses and backcrosses of the HH plants to h plants. (HH)1 k value obtained from a separate experiment.

It is unlikely that the trend in Fig. 3a is fortuitously produced by the environments because some generations were grown in different years, and different generations were grown in the same years, without disturbing the pattern. In an independent study with another set of crosses in other years and conditions a suspected similar trend emerged (McLellan & Durrant, 1973). It is possible that the gradual changes in amount of DNA which occur in L and S plants grown in the restricted conditions of 13 cm pots could have an effect. But large changes in DNA amount can occur in L and S in other environments without there being any observed associated changes, and the independent study was not done in this restricted type of environment. The environment however probably alters the rate at which the changes over generations occur, and the amount of variation between F4 families.

In these experiments, the environment has the greatest influence on the inheritance of H-h in the earlier, F2 and F3, generations, particularly in the F2 generation where the differences between the ratios and k values in different years are highly significant. This is not due to selection because the seed samples came from the same batch of seed, and germination was virtually 100 per cent. In other experiments (Durrant & Nicholas, 1970) different F2 ratios were obtained. Furthermore, the correlation between the Hh classes of one generation and the Hh ratios of the next is highest between the F2 and F3. Hence the changes in the frequencies of H and h plants in these populations are due to the effect of the environment on the regulation of H-h in the earlier generations of the cross, not to selection for H or h, and they show that the environment can induce heritable changes not only in the otherwise true breeding original variety but also in the heterozygotes of crosses between types environmentally induced from it.

In other characters, genetic backgrounds and environments, regulation in the heterozygotes might conceivably swing completely in one direction and switch all plants to one or other parental value, either to HH or to hh, giving a paramutation-like change. There is an indication that plant weight behaves thus in L and S when these are outcrossed to some varieties (Durrant, 1972) the magnitude or rate being dependent upon the interaction between the genetic background and environment.

The trend over generations implies that Hh heterozygotes in the earlier generations are not the same as Hh heterozygotes in later generations. This is borne out by the backcrosses. In the F2, differences associated with the Hh classes are transmitted in the backcrosses to both parents, giving H/h ratios correlated with the F2 Hh classes. In the F4 backcrosses there are little or no differences ascribable to the Hh classes of previous generations; instead about 30 per cent of the homozygotes are converted to heterozygotes. Either H and h collaborate in the F4 backcrosses with other factors to restore the balanced, stable regulation of the heterozygotes previously built up in F4, implying that selection of heterozygotes over generations selects for a balanced Hh regulation; or there is a return in the F4 backcrosses to the instability of the F2 and F3 generations and an excess of Hh heterozygotes which characterises them. On either interpretation, H and h gametes from F4 Hh heterozygotes are not genetically the same as those from F2 Hh heterozygotes.

HH plants in F2 and F3, and probably in later generations, have a mean hair number significantly higher (enhancement) than that of HH parents due to the presence of relatively more plants with higher hair numbers. Fig. 3b shows that enhanced HH plants produce Hh families with higher k values when backcrossed to hh, i.e., they have a higher H potential with more plants in the higher hair number class.

The F2 Hh class influences the amount of enhancement in F3. The lowest class, F2 (Hh)3 in 1970 gives F3 HH plants with a mean hair number which is not significantly greater than HH parents, and the distribution is similar to that of HH parents (Fig. 2d). Table 3 shows that in these F3 families there are more than three times as many Hh as HH plants, and more than four times as many Hh, indicating that not only are these F3 HH plants unable to produce enhancement but they also are unable to maintain in many cases the HH class with 60 hairs per septum and are switched to the (Hh)5 class with 50 hairs per septum or lower. Hence enhancement of HH homozygotes is an extension of the pattern displayed by the heterozygotes and is presumably due to the same causes.

Taking these several results together, one interpretation is that H and h are stable in the homozygotes and co-exist in the heterozygote, preserving their identity, and giving mendelian segregations apart from deviations due to instability at the locus and the influence of the environment. Structural and regulator genes would be closely linked, or adjacent, and the instability due to interaction between the two homologous regions.

Another is that H and h are stable in the homozygotes but do not co-exist in the 'heterozygote', the 'segregations' arising from statistically random events. HH and hh, and their adjacent regulators, could be controlled by genes elsewhere on the chromosomes; in the 'heterozygote', the two regulations are disrupted giving an unstable situation where, in a given environment, there is an equal probability of repression or de-repression occurring in each nucleus, yielding on average equal numbers of H and h gametes, and mendelian ratios. Other environments may shift the balance, giving unequal probabilities and deviations from mendelian ratios. This would mean that Hh 'heterozygotes' never exist as two allelic entities, H and h, and the range of phenotypes of the 'heterozygotes' is some justification for this view.

A third interpretation is that a reasonably stable, linked structural/regulator gene association, which may or may not suffer a moderate amount of relatively independent internal disruption when heterozygous, is further regulated by factors elsewhere on the chromosomes. Differences between the Hh classes in the earlier generations are reflection, or a result, of other events in the nuclei; they are not due to H or h per se, but are an indication of probable Hh changes to come.

There are other types of change, which would also require regulation of some kind. These are, the attachment of controlling elements, alteration of gene position due to interpolation of additional DNA, or its loss, heterochromatisation, gene redundancy with activation of different numbers of repeats or, deserving further comment, gene amplification. The mainly arithmetic progression of hair number in the seven phenotypes, hh with none, Hh with 20, 30,40 and 50 hairs, and HH with 60 and 70, may be due to additive increments, or a geometric progression (Amaldi, et al, 1973), of reiterated sequences. There are substantial differences in reiteration of DNA sequences between L and S, and there are similarities with the bobbed character in Drosophila melanogaster. Flies with the bobbed phenotype contain fewer ribosomal genes than wild type flies. In certain heterozygotes the number of genes is amplified, some times in excess of the wild type number, and progeny carrying the amplified chromosome revert, or are magnified, to the wild type phenotype (Boncinelli, et al, 1972: Graziani, et al, 1974). There is no correlation between HH and hh and the number of ribosomal genes in the parents here. Both SH and Sh have about 1500, and LH and Lh have about 2500 (Timmis & Ingle, 1973, 1975; Cullis, 1976). Although therefore the h locus is apparently not the rDNA locus it may nevertheless be amplified; when the different numbers of gene copies of the HH and hh parents are brought together in the F1 they could give variable populations of free and integrated copies in the earlier generations of the cross, where a substantial proportion of the plants would have intermediate numbers (i.e., those designated Hh), eventually to resolve into stable heterozygous and homozygous combinations by their integration into the chromosomes in later generations.