Genetics 42 (2): 93-103 (1957)
The Connecticut Agricultural Experiment Station, New Haven, Connecticut
Received August 24, 1956

THE immediate superiority of crossbred offspring over their inbred parents or their inbred offspring, heterosis, is a manifestation of gene action. Since information concerning the nature of genes and their effects is accumulating rapidly it is understandable that knowledge of heterosis is in a formative stage. There are two general interpretations of heterosis: the accumulation of the effects of favorable dominant genes at different loci, and the interaction of different alleles at the same locus. The first hypothesis is generally understood as dominance, the second as superdominance or overdominance.

The important difference between these two hypotheses should be clearly understood. They both assume interaction between alleles. In the dominance hypothesis the interaction is between alleles at different loci. These effects can be accumulated in the heterozygous condition and fixed in the homozygous condition. In overdominance the reaction is between alleles at the same locus. These effects can be accumulated in the heterozygous condition but cannot be fixed in the homozygous condition. Therefore, the important difference in these two concepts is whether or not favorable effects can be recombined in one true breeding strain which will be relatively uninfluenced by the type of mating.

Mendelism and linkage provided a simple explanation for an accumulation of favorable effects in hybrid offspring and overcame the objections that had been raised against dominance shortly after Mendel's principles first became known. Further evidence from population genetics has raised additional objections as outlined by HULL (1952), CROW (1952), LERNER (1954), ROBINSON, COMSTOCK, HARVEY (1955), and others, to the dominance hypothesis that need careful consideration. Additional experimental evidence is needed to arrive at a decision in this matter.

The published report of the Iowa Conference on Heterosis, edited by GOWEN (1952), and of a similar discussion in England under the leadership of MATHER (1955), give a comprehensive review of investigations bearing on this subject and much new evidence. EAST (1936) gave a theoretical interpretation of overdominance resulting from the accumulation of dominant effects of alleles at the same locus. Compound genes with multiple effects and completely linked, or very closely linked, genes in the repulsion or transphase give the same result. Additional evidence has been presented by several investigators by correlating the amount of heterosis with the degree of heteroxygosity and from mutant genes originating in inbred lines. New evidence is presented here from mutant genes extracted from crosses showing heterosis and from heterozygous alleles in backcrossed progenies rendered highly homozygous at all other loci. Convincing evidence of dominant genes important in heterosis has been given by EMERSON (1952), ROBBINS (1952), and WHALEY (1952).


One of the unanswered questions is how much of the dominance effect is due to nonallelic gene interaction. There is abundant evidence showing the relation between the degree of heterozygosis and the expression of heterosis. STRINGFIELD (1950) compared four stages of heterozygosity in maize in yield of grain, time of maturity, and height of stalk, as a measure of the amount of heterosis. Inbreds from different varieties after ten or more generations of inbreeding intercrossed among themselves were used to represent 0 degree of heterozygosis; the F1 generation from crosses of two inbreds, intercrossed among themselves, and the backcrosses of F1 by their parental inbreds, represented 50 percent heterozygosis; the combination of two single crosses (A X B) X (A X C) having one inbred in common represented 75 percent heterozygosis, and three combinations of two, three, and four inbreds represented 100 percent heterozygosis. STRINGFIELD concluded that "the expression of hybrid vigor does not follow a linear trend as equal increments of new dominants are added. The deviation from linearity was slight when yield was measured, but the other measures strongly suggest that the regression is curvilinear". SENTZ, ROBINSON and COMSTOCK (1954) compared five degrees of heterozygosis in a similar manner and obtained a proportionately higher expression of heterosis at the 25 percent heterozygosis level. This amount of hybridity was obtained by backcrossing twice in succession to each inbred parent that made up the original cross. At the other degrees of heterozygosis the relations were not essentially linear. The 75 percent point was frequently too low and never significantly too high. The authors consider this evidence to indicate an interaction of genes at different loci.

How accurately degrees of heterozygosity can be related to the amount and source of heterosis is difficult to measure. Interactions between genes at different loci occur in the homozygous as well as in the heterozygous condition. In the heterozygous condition there are simply more different genes involved. With a large number of genes, each contributing a small amount to the total effect, the accumulation would be gradual. Duplicate genes make their largest contribution at the lower degrees of heterozygosity while complementary genes add more at higher concentrations. Thus, the two types of gene action would tend to balance each other.


Cases of heterosis resulting from assumed single gene differences have been reported. These have been summarized by HAGBERG (1953). Spontaneous or induced mutations occurring in inbred and relatively homozygous material have been crossed back with their normal homozygous parental lines and the assumed monofactorial heterozygote has been compared with the homozygous normal and homozygous mutant. Where the homozygous mutant is lethal or severely handicapped, the comparison, of course, is made between the homozygous and heterozygous normal. In many cases, in a wide diversity of plant and animal material, the heterozygote is superior in some measurable character.

The writer (JONES 1945) studied five spontaneous mutants in maize and found the heterozygous plants to be significantly superior to the original normal line in one or more respects. The mutations were all of a degenerative nature. They were of the type that would normally beat a distinct reproductive disadvantage under conditions of natural selection. These mutations affected leaf width, plant height, chlorophyll content, and time of flowering, and were considered to be a somewhat random sample of the degenerative changes that occur regularly in this species.

Variations of this kind occurring in a naturally cross-fertilized species like maize may be maintained in the heterozygous condition indefinitely but are rapidly eliminated by self-fertilization. Since crosses of the surviving normal lines, after these deleterious recessive genes are eliminated, are as vigorous and productive as the original population, or more so, and again show the same reduction on further inbreeding, obviously this type of variation can have little or no effect on heterosis as generally expressed in crosses of homozygous lines. They are of considerable importance in reproductive ability and adaptation in naturally cross-fertilized species. The appearance of these deleterious recessive characters in individuals when naturally cross-fertilized species are first inbred contributes directly to their reduced viability and reproductive efficiency, and usually furnish the basis for much of the apparently unfavorable effects of inbreeding.

One of the mutant characters studied, crooked stalk, was also noted to be accompanied by an obscure chlorophyll change. At early stages of growth the young leaves at the top of the plants are lighter in color. This condition, called pale top, was later found to be separable from the crooked stalk. Therefore, more than one mutation had occurred spontaneously in this inbred line. For this reason it was suspected that the heterosis apparent in the original cross of mutant and normal was not the result of a single gene difference. From this cross recessive pale crooked plants were recovered as well as homozygous normal plants. A comparison was made in height of stalk, and in weight of grain between the original mutant pale crooked and the recovered pale crooked plants as segregants from the cross with the normal line in which these mutations occurred, and with the cross of original and recovered lines. In this comparison all three lots were genetically alike for all visible characters. They were all mutant. The question to be answered is do they differ in growth and productiveness. If so, could this be due to genes obtained from the normal line that were not present in the original mutant line? The results are given in table 1. The comparison was made in a 3 X 3 Latin square replicated three times. The letters identify the different entries.

In all comparisons shown in tables 1 to 3 seed from several hand pollinated plants in each lot was bulked and planted in rows ten feet long and three feet between rows. An excess of seed was planted and thinned at an early seedling stage to ten plants per plot. The plants were measured for height to the tip of the unbroken tassel as soon as all pollen was shed. Ears were harvested at maturity and dried to uniform moisture content. Growing conditions were favorable and a good stand of plants was obtained in nearly all plots. In some cases the complete number of plants was not measured because of broken tassels. In these cases there was no unevenness in stand. All of the results shown in the tables are analyzed for significance by the standard methods. The results are given with each table.

The recovered homozygous pale crooked and the first generation cross of the original and recovered mutant lines are both significantly taller than the original mutant line. The same comparison was made between the original normal and the recovered normal and between crosses of original normal by the original mutant line and the recovered normal and recovered mutant line, and also with the cross of original normal and recovered normal. In this last cross the mutant genes are not involved. The results are given in table 2. The comparison is made in a 6 X 6 Latin square arrangement of plots. Just as the recovered mutant line came out of the cross taller than it went in, so also did the recovered homozygous normal line. The cross of original normal and pale crooked gave the same increase in height (six percent) as when previously crossed several generations earlier, as reported in 1945. The cross of recovered normal by recovered pale crooked also showed the same increase in height. Likewise, the cross of original normal by recovered normal, where the mutant genes were not involved, gave the same increase in height.

A comparison in plant height of original and recovered mutant lines of maize,
pale and crooked, in self-fertilized and intercrossed progenies

Field arrangement and average height
P 51.0 Q 58.9 R 58.8 Q 57.5 R 58.0 P 54.3 R 59.0 P 54.8 Q 58.6
Q 54.9 R 58.1 P 54.9 P 53.9 Q S5.7 R 59.9 P 55.8 Q 57.2 R 57.3
R 52.3 P 53.1 Q 57.6 R 56.2 P 50.5 Q 58.4 Q 54.4 R 56.7 P 56.0


Code Progenies Average height of
plant in inches
P Original Pale Crooked Selfed 53.8
Q Recovered Pale Crooked from Cross 57.0
R Original X Recovered F1 Cross 57.4


Analysis of Variance
Rows 2 15.30 7.65 4.93 <.05
Columns 8 54.63 6.83 4.40 <.01
Varieties 2 69.22 34.61 22.31 <.001
    Inbred vs.crosses (1) (68.69) (68.69) (44.32) <.001
    Within crosses (1) (0.53) (0.53) (0.34) >.5
Error 14 21.80 1.55 1  
Q and R taller than P at probability .95 (from tR test).
Q and R (crosses as a group) taller than P (inbred) at probability .999 (from analysis of variance).

Two normal lines (O and J in table 2) from the same source have been continued by self-fertilization. These do not differ in any measurable degree. Crosses between normal lines were compared in the original report. These showed no increases in crosses. A further test is needed.

Yield of grain of all of these progenies was also compared. The differences are in the same direction but no significant differences were obtained and the results are not reported here. Yield of grain is very difficult to measure in inbred strains of maize. Small environmental effects may determine whether or not a plant produces seeds obscuring small genetic differences.

A comparison in plant height of original normal inbred lines of maize and normal lines recovered
from a cross with mutant lines, pale and crooked, out of the same inbred line

Field arrangement and average height
J 73.3 K 75.6 L 78.6 M 79.0 N 79.7 O 72.3
M 78.6 J 71.3 O 75.4 K 79.6 L 79.6 N 78.6
K 76.2 L 77.1 N 78.2 J 72.0 O 76.3 M 77.5
O 71.4 M 78.1 K 78.9 N 78.6 J 74.3 L 79.0
N 76.5 O 72.9 J 75.9 L 79.0 M 78.3 K 76.5
L 75.0 N 74.8 M 78.8 O 72.2 K 75.7 J 69.5


Code Progenies Average height
of plant in inches
O Original Normal 73.4
J Original Normal 72.7
K Recovered Normal from Cross 77.1
L Original Normal X Original Pale Crooked, F1 Cross 78.1
M Recovered Normal X Recovered Pale Crooked, F1 Cross 78.4
N Original Normal X Recovered Normal, F1 Cross 77.7


Analysis of variance
Rows 5 29.16 L 5.83 5.00 <.005
Columns 5 39.42 7.88 6.76 <.001
Varieties 5 187.17 37.43 32.08 <.001
Inhreds vs. crosses (1) (180.19) (180.19) (154.01) <.001
Within inbreds and crosses (4) (6.99) (1.75) (1.50) >.2
Error 20 23.34 1.17 1  
K, L, M, and N (aller than O and J at probability .95 (from tR test).
K to N (crosses as a group) taller than O and J (inbreds as a group) at probability .999 (from analysis of variance).

It is obvious that genetic differences other than the visible mutations are involved in these comparisons of self-fertilized and intercrossed lines. Both the mutant and the normal lines came out of the first cross taller than they went in. These effects are dominant in crosses since the taller growth is shown by all of the intercrosses.


Additional evidence has been sought by a somewhat different experiment in which an attempt was made to compare homozygous and heterozygous single gene differences in backcrossed lines. For this experiment two inbred strains were selected that differed in easily visible gene markers. The genes involved are yellow and white endosperm color, red and white cob color (alleles of the P pericarp color gene), and normal and glossy seedlings. One of the two inbreds showing different alleles of these three genes is Connecticut 243, one of the original Learning inbreds (designated 1-6 in earlier publication). This line had been continuously self-fertilized for over 35 generations when the experiment was started. The other line is Connecticut 20 out of Burr White, and continuously self-fertilized more than 20 generations. C243 is YY, PP, gl gl in composition, and C20 is yy, pp, Gl Gl. The first generation hybrid plants of this combination show the usual increase in height and productiveness characteristic of crosses of inbred lines out of varieties of different type. The first generation hybrid was backcrossed by both parental lines and plants showing each dominant gene marker were continuously backcrossed in successive generations by the line that was recessive for the marker gene in each case. Three lines were established, each approaching the recurrent recessive parent in all visible characters except for the one dominant gene marker that was selected in each generation and maintained in the heterozygous condition. The yellow endosperm and red cob lines were each backcrossed continuously by the C20 line, recessive for these alleles. The non-glossy line was also backcrossed continuously by the C243 line, recessive for the glossy leaf character determiner.

The plan of the experiment was to backcross long enough to restore the visible characteristics and level of vigor of the recurrent parent, and then make a careful comparison in measurable characters between individuals or progenies heterozygous dominant and homozygous recessive from the same parents.

These three gene markers were selected because they have no differential effect on growth or productiveness as far as known in the dominant or recessive condition. They are, therefore, suitable for comparing the effect of the homozygous and heterozygous states. Yellow seeded and white seeded varieties of maize are about equal in numbers throughout the world. This gene controls the amount of vitamin A precursor stored in the seed, but has no effect on the growth of the plant as far as it can be measured. Likewise, no differences in yield or other measurable characters have been established for red cob or white cob varieties, or between normal and glossy seedlings. The glossy character affects the water shedding ability of the leaf surface and might be considered to be an unfavorable character. Normal plants are pubescent and bluish green in color of the seedling leaves and shed water readily. Glossy plants lack this pubescence in the early seedling stage. However, many species of grasses are normally glossy and a number of standard inbreds widely used in the production of hybrid corn are glossy. Glossy plants become normally pubescent over most of the leaf surface after the early seedling stage but can be easily distinguished in small areas until flowering.

These three genes in their alternate alleles can be assumed to have little or no effect on growth and productiveness. The evidence presented here bears this out. They also represent a type of gene differences that characterizes varieties. It is just such varietal differences that were assumed to contribute to heterosis by EAST and by SHULL in their original hypothesis of a stimulus from heterozygosity itself.

While the three genes studied comprise a very small sample of such gene varietal differences, they are thought to represent a characteristic random sample of such differences. All three genes show clear-cut Mendelian segregation and recombination and have been located on chromosome maps.

An additional source of the Y endosperm color gene was obtained from a third unrelated inbred designated as Indiana P8. This line was selected as a source of the dominant Y gene as it had a darker yellow color. The cross of PS X C20 was also backcrossed by C20 and carried along in the same way as the cross of C243 and C20. In both backcrosses the yellow color was diluted rapidly and after a few generations the difference in color between Yy and yy seeds on the same ear was quite faint but could be easily distinguished on a blue background in strong daylight. The color separation was always checked by different persons and, since only yellow seeds were used for further backcrossing, faulty classifications or errors due to heterofertilization were eliminated in each generation if they occurred, and would have no bearing on the results obtained. The darker color of the P8 source was evidently due to modifying genes which were quickly lost as both backcrossed series appeared to be the same in depth of color.

The backcrossed lines approached their recurrent parents closely in size, time of maturity, and other visible characters in the sixth to eighth generation. About 15 plants were grown in each backcrossed progeny in each successive generation. The homozygous inbreds were used as the pollen parent. Plants used for cross pollination were selected at random in the field as far as this can be done. Weak plants produce no ear shoots and naturally cannot be used for propagation as in any inbreeding or backcrossing experiment. The inbred lines used for backcrossing were always propagated by self pollinated plants.

Selection of heterozygous plants or seeds in each generation enforced heterozygosity upon the one locus involved in each case in each generation, and presumably this locus also carried with it a section of chromosome on each side of the marker gene maintained also in the heterozygous condition. Without this enforced heterozygosity random mating would reduce the expected heterozygosity to less than two percent in six generations of backcrossing on the average. Since all of the backcrosses were visibly larger and different from their recurrent inbred parents until after six generations this enforced heterozygosity had some effect, and genes affecting heterosis are presumably present in these regions of the chromosomes.

Backcrossing was continued for 11 or more generations, until no visible differences were apparent between the original self-fertilized recurrent parental line and the two backcrossed progenies in each case, one heterozygous dominant and the other homozygous recessive for the gene marker used in each case. Homozygous progenies were also included in the comparison grown from homozygous recessive plants self-fertilized from a previous backcrossed generation.

For the comparison of the glossy and normal plants backcrossed by C243 three progenies were grown in a 3 X 3 Latin square replicated five times. The three lots of seed planted came from several ears in each case and were as follows: the original glossy inbred, C243, continuously self-fertilized; glossy plants from the 12th generation of backcrossing, backcrossed once more by glossy C243; normal plants in the 12th generation backcrossed once more by the glossy inbred. This last progeny gave an equal number of normal and glossy plants which could only be separated after planting. Dominant and recessive plants were marked in the field and measured for height. No significant differences were found. The parental, continuously selfed glossy inbred measured 78.2 inches in average height at maturity. The homozygous backcrossed glossy progeny was slightly taller measuring 80.6 inches in height. The segregating backcrossed progeny averaged 81.4 for both dominant and recessive plants. When separated into two groups the homozygous recessive glossy plants averaged 82.1 and the heterozygous dominant normal plants averaged 80.3 inches in height. These differences are not significant and show no superiority of the plants heterozygous Gl gl compared with the plants homozygous recessive at this locus. In this last comparison the homozygous and heterozygous seeds were produced on the same plants and in equal numbers, thus avoiding any differences in germination or vitality of seed from any source.

Yield of grain was also compared in this series. Many of these long inbred plants produced no ears and no significant differences were recorded. All of the backcrossed lines yielded more than the continuously self-fertilized line. Although slightly shorter in height the heterozygous normal plants yielded considerable more than the homozygous glossy plants from the seed grown on the same plants. However, the average yield for both kinds of plants in this segregating progeny was only 73.2 grams per plant compared to 71.2 grams for the homozygous glossy plants equally backcrossed. Plot yields varied from 31.7 to 114.0 in grams per plant, and these results cannot be relied upon. Some other measure of productiveness is needed in a comparison of inbred plants such as these.

A similar comparison is made between plants heterozygous and homozygous for the gene markers Yy and Pp. Nine different progenies were grown in a 9 X 9 Latin square. The results are given in table 3. In this series the other parent, C20, of the original cross was used as the recessive recurrent parent. Eight different lots of backcrossed plants are compared with the original inbred parental line. This parental line was propagated by a single selfed ear in each generation from the same progeny as used for backcrossing.

After 13 generations of backcrossing white cob plants in this segregating generation were selfed and grown as lot B in table 3; other white cob plants were again backcrossed and grown as lot D. The seed from red cob plants also backcrossed were grown as lot C. In each case seed from several plants of the same genotype was mixed. The backcrossed red cob plants produced red and white cob plants in equal numbers. These could not be separated in the field at the time the plants were measured for height. Many plants produced no ears. These three lots averaged B 73.3, C 74.2 and D 75.0 inches in height compared with A 72.3 for their continuously self-fertilized parental line. The difference between the backcrosses as a group and the inbred parent is highly significant. The differences between the various backcrossed progenies are not significant. As in the other series, the backcrossed progenies are slightly taller than the inbred but there is no difference between the plants heterozygous dominant and homozygous recessive for the Pp marker gene.

The most critical comparison is between the plants grown from yellow and white seeds produced on the same ears. These seeds were separated before planting and the comparison can be made between progenies in which the plants are either all heterozygous Yy, E and G, or all homozygous yy, F and H. As stated previously, the dominant Y gene was derived originally from two different sources, inbred lines C243 and P8. In the series from C243 the two lots, E and F, averaged exactly the same in height. In the two lots from P8, G and H, the plants from yellow seeds are slightly taller but not significantly so. Plants from white seeds again backcrossed by the white seeded parent, I, also show no significant differences from the yellow seeded plants. In every case the backcrossed lines are taller than the original inbred line. As a group the differences are highly significant. This difference cannot be attributed to the heterozygosity of the visible marker genes involved.

A comparison in plant height of an original self-fertilized line with recovered backcrossed lines,
heterozygous and homozygous for the marker genes
Pp and Yy, derived from unrelated lines

Field arrangement and average height
A 69.0 B 72.3 C 71.8 D 74.5 E 74.2 F 74.8 G 75.0 H 72.1 I 70.0
B 71.8 C 74.6 E 73.7 G 71.2 D 76.4 I 72.7 F 75.4 A 73.3 H 70.5
C 73.0 D 75.6 F 73.8 A 72.4 H 70.9 G 72.7 I 75.8 E 72.3 B 71.8
D 70.6 H 72.4 A 70.7 B 71.5 F 73.5 E 74.7 C 73.3 I 74.5 G 73.1
E 70.3 G 77.4 B 74.4 I 75.6 C 76.3 H 76.1 D 76.1 F 75.2 A 69.3
F 69.3 I 76.8 H 75.9 E 75.3 B 71.9 D 75.9 A 76.9 G 76.8 C 69.1
G 73.0 F 76.4 I 74.7 C 77.8 A 71.5 B 74.1 H 76.6 D 76.4 E 73.4
H 69.8 E 77.3 G 78.2 F 76.1 I 73.9 A 72.7 B 75.0 C 76.0 D 74.2
I 71.0 A 75.3 D 75.6 H 76.3 G 73.1 C 75.6 E 77.8 B 77.2 F 74.0


Code Progenies Average height of
plant in inches
A Original inbred line C20, yy pp 72.3
B Recovered selfed line from backcross, white cob, pp 13.3
C Segregating backcrossed line, red and white cob, Pp and pp 74.2
D Recovered hackcrossed line, white cob, pp X pp 75.0
E Heterozygous backcrossed plants from yellow seeds, C243, Yy 74.3
F Homozygous backcrossed plants from white seeds, C243, yy 74.3
G Heterozygous backcrossed plants from yellow seeds, P8, Yy 74.6
H Homozygous backcrossed plants from white seeds, P8, yy 73.4
I Recovered backcrossed line, white seeds yy X yy 73.9


Analysis of variance
Rows 8 70.53 8.82 3.65 <.005
Columns 8 192.26 24.03 9.94 <.005
Varieties 8 46.03 5.75 2.38 <.05
Inbred vs. crosses (1) (25.36) (25.36) (10.48) <.005
Within crosses (7) (20.67) (2.95) (1.22) > .2
Error 56 135.42 2.42    
D taller than A at probability .95 (from tR test).
B to I (crosses as a group) taller than A (inbred) at probability .995 (from analysis of variance).


The assumption that naturally occurring visible mutations in highly homozygous inbred lines of maize involve only single gene changes is not valid as indicated by the writer (1952). The evidence for this conclusion is presented here. Additional critical evidence has been given by SCHULER (1954) JINKS (1955), and SCHULER and SPRAGUE (1956). Other gene changes occur at or near the time the visible mutations appear or all heterozygosity of genes affecting growth cannot be eliminated by continuous self-fertilization.

Although there is evidence for an overdominance effect from a study of variability in cross-fertilized organisms as presented by HULL (1952), CROW (1952), LERNER (1954), and others, specific cases of single gene superiority or overdominance are remarkably few and even these cases may be due to multiple gene effects as shown by the cases given here that have been more critically analyzed.

The mutant tobacco described by the writer (1921) is a good illustration of single gene superiority. The change involved the reaction to length of day. The mutant plants were unable to flower in the normal summer day but flowered and produced seed in a shortened day. The visible change was due to a spontaneous mutation from a single dominant to a recessive condition. The recessive mutant was wholly unable to survive under natural conditions in the long day at northern latitudes. The heterozygote of normal and mutant alleles was better able to survive than the homozygous normal since the plants produced more leaves, flowers, and seeds under the same conditions. This is a clear case of single gene superiority but the heterozygote shows no acceleration of growth rate. It merely extends the growing period longer to produce a larger amount of material and greater reproductive ability. Whether other genes are involved in the larger growth and greater reproductive ability of the heterozygous plants is not known. This has not been examined by the recovery tests as done here, but whether or not other genes are involved has little bearing on the question at issue, since number of leaves in tobacco is not influenced by heterosis to the extent shown in this case. Whether this is a case of heterosis is a matter of definition.

A similar situation has been reported by QUINBY and KARPER (1946) in sorghum. A single gene for time of maturity in the heterozygous condition when acting with other genes, also regulating maturity, greatly increases the total amount of growth. As in tobacco there was no increase in the rate of growth. Theoretically, many genes could act in this way. It is surprising that so few instances have been found (see LERNER 1954; PONTECORVO 1955).

Many genes show dosage effects where the productiveness of the gene is correlated with the numbers of dominant alleles present. In any case where the diploid or higher number of normal alleles at the same locus produces an excess of essential ingredients, thereby reducing the effectiveness of other genes involved in a reaction system, the hemizygote or heterozygote would be superior. There is every reason to expect that conditions of this kind would be selected for in cross-fertilized organisms. Apparently the genes of most importance in physiological processes cannot be easily identified in the higher plants and animals where heterosis has been mainly studied.


A sample of two different types of genes in maize (1) unfavorable degenerative mutants, and (2) varietal differences with no effect upon growth and productiveness have been compared in the homozygous and heterozygous condition. These two types of genes provide most of the visible variation found in this naturally crossfertilized plant. Neither type gives any evidence for overdominance.

Both types of genes show dominance. The heterozygous normal is equal to the homozygous normal in all visible characters in the degenerative mutant series (1) and the presence of color is dominant over no color, and pubescence over nonpubescence in the varietal difference series (2).

The evidence shows that other genes or some unknown factors are involved in producing heterosis, since all backcrossed and intercrossed progenies are taller than the corresponding continuous self-fertilized progenies.

If genes are involved in this heterosis they have no visible effects by which they can be identified by the techniques that have been used so far.

It is possible that these unknown genes may show overdominance but this has not been proved.


Valuable assistance has been given by DR. C. I. BLISS and MR. A. P. MUNSON in planning the field arrangement of the experiments and calculating the statistical significance of the results.