The American Naturalist, 92(867): 321-328 (Nov-Dec, 1958)

The Connecticut Agricultural Experiment Station, New Haven, Connecticut

The concept of crossing homozygous inbred strains anew each year to produce vigorous high-yielding combinations of hybrid corn was first proposed by George H. Shull in 1908. Theoretically it had many advantages over the older methods of seed selection. It made possible a much more complete control of the heredity in a naturally highly cross-fertilized plant. This finally resulted in the selection and use of superior germplasm. It also made possible the exploitation of the highest degree of hybrid vigor. This natural phenomenon was known to be important in many domesticated animals and cultivated plants, and particularly so in maize.

Notwithstanding these demonstrated advantages, the method as outlined by Shull and tested by numerous agronomists was not put into production due to very serious handicaps. Chief of these was the small size and variable shape of seeds produced on weak inbred plants. The cost of the seed was high and the performance was erratic, due to the handicap the plants had at the beginning of the growing season. Other factors also worked against the use of the uniform crosses of inbred strains as will be explained later.

E. M. East had started an inbreeding experiment with maize in Illinois in 1905, the same year that Shull began his experiments on Long Island. These experiments were continued when he transferred to the Connecticut Agricultural Experiment Station later the same year. Although no crosses had been grown between his two-generation self-fertilized lines, East immediately recognized the importance of the new method when Shull presented his results at a meeting of the American Breeders Association at Washington, D. C., in January, 1908.

At first both East and his student and successor, H. K. Hayes, made no headway with the commercial utilization of crosses of inbred strains and reverted to a method of crossing varieties originally advocated by W. J. Beal at the Michigan Experiment Station many years earlier. East had been employed as a chemist at the Illinois Station to make the analyses of the maize material being selected for chemical composition. In Connecticut he continued a similar selection program but in inbred lines of maize instead of in naturally pollinated field plantings as in Illinois. Hayes developed this program further and added a number of new lines from the Illinois high and low protein selections of the Burr White variety and similar selections from locally adapted varieties of Leaming and Canada Yellow Flint. These experiments led directly to a method of maize breeding known as "Selection in Self-fertilized lines'' described by the writer (Jones, 1920). This method, with various modifications by many corn breeders, produced the highly selected inbred strains now used for the commercial production of hybrid corn in all parts of the world.

The many inbred strains out of several distinct types of corn available at the Connecticut Station made possible a new method of combining selffertilized lines. Tests had shown that crosses between varieties of corn of different type were usually more vigorous and productive than crosses made between varieties of the same type. This also proved to be true for the inbred strains derived from these varieties. Crosses of Leaming by Burr White inbreds or of inbreds out of High and Low protein selections of Burr White were usually noticeably more productive than crosses of inbreds within these varieties. Crosses of inbred flint or flour lines with each other or with dent inbreds were also sturdy in stalk growth and high yielding for grain. Fortunately crosses in all combinations of inbreds out of the available strains of Leaming and Burr White had been grown. From all of this evidence it seemed quite possible that a second crossing of a first generation hybrid of two Burr White inbreds by another first generation hybrid of two Learning inbreds would be equally productive or even more so.

It was realized that this second crossing would be quite different from the first. All of the plants of the first cross are genetically alike where the inbred strains have been reduced to complete or nearly complete homozygosity. In the second cross all of the plants would be genetically unlike. Many of these individuals would be combinations of genes that had never been tested or even seen before. But if this second crossing would be approximately as productive, it might well be worth the extra effort since it would overcome most of the objections that had been found with the first cross. This second cross combining four inbreds, for convenience, was called a "Double Cross" to distinguish it from the first, or Single Cross, of two inbreds.

The first Double Crosses were produced in a small isolated field crossing plot grown at the Connecticut Agricultural Experiment Station farm at Mount Carmel in 1917. One of the best combinations of two Leaming inbreds was used as the pollen parent, being planted in every other row and in both outside rows. In the alternating rows were arranged 27 small plots. These seed parents, which were detasseled at the proper time, included various combinations of inbreds out of Burr White High and Low Protein selections and single crosses of other Learning inbreds differing from the two inbreds used for the pollinator single cross. In addition, there were several F1 crosses of dent x flour, dent x flint, flour x flint, flint x flint, and a few dent inbreds alone. The last items when crossed would give combinations of three inbreds instead of four. Usually a hybrid of this kind, called a three-way cross, is made with the single cross as the seed parent to gain the advantage of high seed yields and large seed kernel. The object here was to compare multiple crosses of inbreds out of the same variety, some with a hybrid, others with an inbred seed parent.

All of these multiple hybrids were grown in a randomized yield test the following year. Most of them grew well and gave high yields. One of the combinations of two Burr White inbreds by two Leaming inbreds seemed to be most desirable on account of high grain yield, strong sturdy stalks and freedom from smut, aphis, and mold damage on the stalks and ears. This combination was called Double Crossed Burr Leaming. After testing for five years in Connecticut in comparison with all of the best adapted varieties commonly grown in the State and in many other places including Virginia, Ohio, Iowa, and Spain, it was put into commercial production by George S. Carter, at Carter Hill Farm, Clinton, Connecticut, beginning in 1921.

Later, another earlier maturing Double Cross called Canada Leaming was developed from inbred strains out of Canada Yellow Flint and the same Leaming pollinator. This with some modification was produced commercially by the Eastern States Farmers' Exchange beginning about 1930, and widely grown throughout New England and other northeastern states. In time both varieties were replaced by better double crosses as soon as these became available.

The greater genotypic variability of the double cross which, at first, was thought to be a serious objection, proved to be no handicap but, in fact, a further advantage. As measured by the coefficient of variation, the number of kernel rows on the ear and nodes per plant were somewhat more variable in the double cross than in the average of the two single-cross parents. Also, in weight of ear per plant the variability was slightly more but not significantly so. Height of plant and length of ear were less variable (Jones, 1922a). In every character measured, except rows of grain, the variability of the double cross was much less than the F2 selfed generation of the parental single crosses.

In all details of plant structure the double cross was also much less variable than the original naturally pollinated varieties. All of the plants were uniformly productive (Jones, 1922b). The Burr Leaming hybrid showed no barren stalks and nearly all ears were about the same size and shape, and all of the plants were free from disease and insect infestation. All of the ears had a mixture of yellow and white kernels showing the genic segregation that was going on in these mixed populations.

Since all of the individuals in these double-crossed populations are uniformly vigorous and productive, all of the many diverse combinations of genes are heterotic. This is an important fact that I called attention to in the publication previously cited. This means that there are an enormous number of genes on all of the chromosomes involved in the heterosis shown by maize and other plants. If this were not the case the variability in productiveness of the double cross would approach that of the F2 self-fertilized generation from the same inbred parents.

Heterosis may be defined as the tendency for crossbred organisms to surpass both their inbred parents and their inbred offspring in some respect. The early hybridizers, Kölreuter, Gärtner, Focke, and many others, gave numerous examples of hybrid vigor in plants which is manifest in many different ways. They showed that the result of cross-fertilization may not always be beneficial to the organism. There is hybrid weakness as well as hybrid vigor. Within the species the result of crossing individuals that differ genically is usually an improvement in some respect. In general there are two different categories of effects. One that includes an increase in size or number of parts, and in rapidity of growth. This result is much the same as an improvement in environmental conditions such as better nourishment or better protection from unfavorable environmental factors. This effect has been called luxuriance. The second classification of effects includes all the factors that help an organism and its offspring to survive, for example, greater reproductive efficiency, disease and insect resistance, tolerance of unfavorable temperatures and other external influences of all kinds. This has been called true heterosis in contrast to mere growth luxuriance.

Naturally, applied geneticists have been more interested in the manifestations of heterosis that improve the organism itself, not only in its ability to grow and survive unfavorable environments but to take full advantage of all favorable environmental conditions that can be provided to both animals and plants so that their economic value can be enhanced. Students of evolution, understandably, are more concerned with the manifestations of heterozygosity and heterosis that enable organisms to compete and to survive under natural conditions. Many times these different results of crossing are antagonistic. Most cultivated plants are so highly selected for the production of food and other special purposes that they can seldom survive more than one or two generations in the wild. Domestic animals are also so highly specialized for meat, milk or egg production for food or for other purposes that they are severely handicapped in reproduction and survival even under carefully controlled conditions. Cross-fertilization among different types of these highly specialized domesticated forms usually results in a strong tendency to revert to a more nearly wild-type condition. They may produce offspring that are less productive but are usually better able to survive. This is particularly noticeable in crosses of the dairy breeds of cattle and of egg-laying strains of chickens. It was not until crosses were made of strains specially selected for productiveness in hybrid combination that the full advantage of heterosis was realized in both animals and plants.

The term heterosis was first used by Shull to designate the stimulus usually accompanying heterozygosity without any implications as to how this invigorating effect was brought about. There are two general interpretations of heterosis. One, the accumulation of dominant favorable genes at different loci. This accounts for much of the superior performance of hybrids. In addition to this there is also an interaction between different alleles at the same locus such that the final result is more favorable to the organism than the action of either allele in the homozygous condition. Direct evidence for intragenic heterosis is difficult to obtain. In view of the complexity of gene loci in bacteria and fungi, where it has been shown that there are numerous sites capable of cumulative action but cannot be recombined, it is difficult to distinguish between allelic and non-allelic interaction. For all practical purposes heterosis is an accumulation of favorable dominant effects. The position effect of cistrons is undoubtedly a factor in hybrid weakness but cistrons are relatively infrequent.

Heterozygosis is the usual result of amphimixis and may or may not be accompanied by heterosis. In most cross-breeding organisms the amount of heterosis remains at a fairly constant level and the organism is dependent upon it for survival. The genetic variability brought about by continued heterozygosity has long been recognized as one of the important factors in sexual reproduction. The ability of organisms to regulate and to utilize genic variability has been called genetic homeostasis by Lerner (1953). In 1944 I proposed the term genic equilibrium (Jones, 1944) to include somewhat the same idea. Darlington and Mather (1949) used the term genetic inertia. Whatever it is called, the concept of a regulatory mechanism that holds variability in bounds and enables an organism to survive is extremely useful and important. It is one of the most important factors in the success of hybrid corn and offers great promise for the further improvement of many naturally self-fertilized crop plants.

To illustrate the importance of homeostasis in hybrid corn let us compare a large number of single crosses made by crossing in all combinations a series of highly selected inbreds with a series of double crosses made from these single crosses. Such a comparison is available from field trials grown in various places in Iowa in 1951. These tests were carried out by G. F. Sprague, P. A. Miller, and L. H. Penny. The data are given in a mimeographed report entitled: "Iowa Experimental Corn Trials" and sent to the corn breeders in the various Corn Conferences in different parts of the country. The data in this report are particularly useful for this purpose as the tests were made in different parts of the State all in the same year. The fertility levels and weather conditions differed at the various locations but the yields obtained, ranging from 38 to 103 bushels per acre, indicate about average production for this most favorable part of the corn-growing region.

The inbreds used are the standard lines widely grown in the north central corn-growing area, selected from many thousands of inbred lines over many years of testing and observation under many different conditions. Along with these older lines are a number of new lines for comparison that were considered promising in preliminary trials. In other words, this series of inbreds is a very highly selected group and represents probably some of the most valuable maize germplasm that has been assembled to date.

From the performance of a series of inbreds crossed in all combinations it is possible to predict quite accurately the yield and other characters of the many double crosses that can be made by crossing again these various single crosses. In the Iowa tests these predictions had been calculated and only a small proportion of the best of the many possible double crosses were actually made and grown. The comparison in yield of the two series of 317 single and 483 double crosses is shown in table 1.

In each group there is a high level of both heterozygosity and heterosis. In average yield the two groups do not differ significantly. The interesting difference is in the variability of yield. Both the two highest and the two lowest classes are in single crosses in which categories the double crosses are not represented at all. Not only is the range in yield wider in the single crosses but the frequency is distinctly bimodal. The double crosses form a more normal frequency distribution, but with a decided skewness toward higher yields (figure 1).


Class centers in
bushels per acre
38 43 48 53 58 63 68 73 78 83 88 93 98 103 Total Mean S.D.
Single crosses 6 16 30 27 5 7 16 53 63 50 28 13 1 2 317 71.03 15.15
Double crosses     4 18 32 55 89 89 119 52 24 1     483 72.09 8.90

All corn breeders are familiar with the fact that the highest yields are usually obtained from single crosses when any large number are compared with multiple crosses. However, almost never is it the same single cross in any one year or in any one location that gives this high yield. The double crosses are more consistently high in yield and desirable in other respects as well, as compared to the single crosses which are more erratic in their performance. While genetically more variable in composition they are more stable and consistent in performance. That is why there are many double-crossed hybrids that are grown over very wide areas, some from Maine to Virginia and west to the Missouri River valley. There are varieties that grow well in the Southern states, in the San Joaquin Valley of California, in Spain, Italy, Israel and India.

The first double crosses were designed to overcome the handicaps that the single crosses had in seed production. It actually turned out that the genic equilibrium, genetic inertia, genetic homeostasis, or whatever it may be called, is by far the more important. It is the gyroscope that holds the ship steady in a surging sea.

Most of the genetic improvement that has been made with self-fertilized plants is based on the selection of pure lines as studied first by Hansen with yeast and Johannsen with beans. These genetically uniform pure line varieties are very productive and highly desirable when environmental conditions are favorable and the varieties are well protected from pests of all kinds. When these external factors are not all favorable the result can be disastrous. Mixtures of pure lines have been proposed and tested by many agronomists (see review by Hayes, Immer and Smith, 1955). These mixtures usually do not give the highest yields at any time but may average higher over a period of years. They are a form of insurance against very low yields or even complete loss due to some new virulent parasite. Silviculturists have found that mixed plantings of forest trees are usually safer than pure stands of one species.

FIGURE 1. Frequency distribution of single- and double-crossed maize hybrids for yield.
Solid line = single crosses, broken line = double crosses.

Harlan, Martin and Stephens (1940), Sunneson and Stevens (1953), Sunneson (1956), and others have tested the crossing of pure lines of cereals and have found this to be advantageous. Starting with a small amount of intercrossed plants, the seed supply is built up rapidly in a few generations. This hybrid mixture has the advantage of both heterozygosity and heterosis, as well as a mixture of homozygous lines. Heterosis is lost rapidly in successive self-fertilized generations but enough of this invigorating effect remains after several generations to be worthwhile. The heterotic plants are more productive and tend to increase faster than the more homozygous individuals. Enforced heterozygosity tends to persist in all self-fertilized organisms whether naturally or artificially inbred.

Many ecological and genetic factors are involved in plant mixtures and agronomists have given considerable attention to this matter. Appreciable increases in yield can be obtained by hybrid mixtures, making the initial intercrossed seed each year and increasing this as fast as possible for field planting while considerable heterosis remains, using this seed for general field planting for only one generation. However, the mixture of genetically different types persists for many generations and can be used in several ways:

  1. As a mixed population subject to natural selection in different locations, with the possibility of significant gains in yields from better adaptation over many generations.
  2. As material for repeated hybrid recombination after periods of natural self-fertilization and consequent reduction in heterozygosity.
  3. As a source of progeny selections. These new selections, after testing, may be used as pure line varieties or as material for a new mixed population with or without intercrossing.

Hybrid mixtures can do as much for naturally self-fertilized crops as crosses of inbred strains have done for cross-fertilized plants. There is every reason to believe that the importance of homeostasis will be recognized and utilized in applied genetics as well as it is recognized as a factor in evolution and survival under natural conditions.