these investigations form the basis for the method of improvement by selection in self-fertilized lines. In the inbreeding experiments just described no selection of superior individuals to perpetuate the strain was made. The aim was to take normal plants at random and note the outcome. Nevertheless a great deal of natural selection has taken place. All abnormalities which interfere with or markedly reduce reproductive ability have been automatically eliminated. In this way many chlorophyll deficiencies, endosperm abnormalities and inherited sterility in tassels and ears, unfavorable conditions almost always present in every cross-pollinated

Figure 34. A diagrammatic representation of the actual and theoretical results of inbreeding corn. The solid lines represent strains which have already been obtained, the dotted lines those which may be expected when corn is worked with more extensively.

variety of corn, have been cleaned out. But this outcome of inbreeding, valuable as it may be, is less important than the control over the heredity made possible by hand pollination and the resulting fixity of type.

In common practice, selection with nearly all cross-fertilized plants has been based on the appearances of the plant or upon the performance of the progeny, and no adequate control of the heredity brought in from the pollen parent has been possible. As generally practised, corn breeding has been similar to a system of animal breeding in which selection is carried on only with the dams paying no attention whatever to the sires. The disastrous

result that such a system would have upon purebred live-stock can readily be appreciated. With all cross-fertilized plants it would be theoretically possible to follow the method now used in animal breeding. Certain desirable individuals could be chosen as seed parents and others as pollen parents. Pollination could be made by hand and the progenies compared on the basis of their performance. There is no doubt that this system followed up as carefully as it is in mating farm animals would give equal results. But such a method is wholly impracticable on account of the small value of the individual plant. The time spent on selecting the

Figure 35. Self-pollinated ears grown on selected plants of Burwell's Yellow Flint, No. 40. Each ear is the starting point of a selected line. These are numbered 1 to 9, top row, and 10 to 18, bottom row, left to right.

parents and on pollinating each generation would not be repaid by the possible gains. Furthermore, with corn, selection is greatly handicapped due to the fact that the principal objective, production of grain, is not visible until after pollination.

A new method of attack, which will make possible a control of the heredity transmitted thru the pollen as well as thru the egg, is needed for all naturally cross-fertilized plants. Since inbreeding is a sorting-out process, selection carried on during the time the plants are being reduced to uniformity and constancy makes

it possible to look for desirable qualities with a certainty of being able to hold them, when once secured, that has never before been possible. From this viewpoint inbreeding is not so important as a method of gaining the maximum effect of hybrid vigor when the inbred strains are crossed as it is of separating out and making visible the very best hereditary qualities that may exist in a heterozygous stock. Strains when once reduced to fixity remain the same indefinitely, barring mutations. With due regard to seasonal variation, crosses between inbred strains give the same result whenever the same combination is made. The uniform production of the first generation hybrids between homozygous strains is an important feature. In this respect cross-fertilized plants are equal to self-fertilized plants in uniformity and fixity of type and have the added advantage of crossing to bring together and use in the first generation the desirable qualities within the species, which in a self-fertilized organism can be used only when recombined and fixed in a homozygous condition. It should therefore be clearly understood that the crossing of inbred strains as such is without particular value and that the opportunity afforded to find and to fix the very best hereditary qualities possessed by a cross-bred race is the more important function of inbreeding. Crossing is merely a means of utilizing this good heredity by giving it maximum vigor. It is to be expected that many inbred strains will have only medium value and give no improvement over the original variety when crossed. The bulk of the germplasm in every population is mediocre. Of necessity only the exceptionally few will give outstanding results. For these reasons the outcome of selection in self-fertilized lines depends upon how extensively and skillfully it is applied.

POSSIBILITY OF OBTAINING VIGOROUS INBRED STRAINS.

Most of the inbred strains of corn so far produced have been reduced to about fifty per cent. or less of the production of the original cross-bred varieties. Some strains have failed to reproduce after one generation of self-fertilization. Others have persisted in a weakened condition for several generations and then perished. Still other strains are able to survive, but are continued only with the greatest difficulty. The majority of the selffertilized lines, when uniformity and fixity of type are reached, are about one-third as productive as at the start. A few are exceptionally good. They grow more vigorously and yield more than the rest and are equally uniform and fixed in their type. But even the best of these are still below the original variety in amount or quality of grain produced. On the basis of hybrid vigor being due to dominance of the more favorable factors it is theoretically possible to secure inbred strains that will show little or no reduction in vigor, and a few may sometime be obtained that are even

more vigorous and productive than the cross-bred variety. This is deduced from the fact that most heterozygous combinations of factors are less effective than the homozygous combinations of the same factors. Thus the cross of yellow and white corn gives a lighter color than pure yellow. The cross between a determinate growth type of tobacco with an indeterminate growth type (Jones, 1921) which involves a single factor, differs from either parent in size of plant and number of leaves. Dominance is seldom perfect and while there is little direct evidence in this respect for characters

Figure 36. Self-pollinated ears grown on selected plants of Gold Nugget, No. 105. Each ear is the starting point of a selected line. These are numbered 2 to 10, top row, and 11 to 20 bottom row, left to right. (Ear 1 was shelled before photographing. It was similar to No. 2.)

which directly affect vigor there is every reason to expect that a homozygous combination of all the more favorable dominant growth factors will make possible a greater development than the heterozygous combinations of the same factors with weaker allelomorphs. However, as just noted, certain results are obtained from heterozygous combinations that can not be obtained from either factor alone. If there are many of these that play a part in growth vigor, then heterozygosity may be indispensable to maximum development. Moreover, recombinations of large number of

factors are extremely difficult to obtain and since favorable and unfavorable growth factors are distributed indiscriminately throughout the hereditary mechanism the chances of securing self-fertilized strains of corn which equal the cross-bred varieties are so exceedingly small that there is little hope of obtaining them. The most that can reasonable be expected are inbred strains which are appreciably better than any that have so far been produced. The results that have already been obtained from self-fertilizing corn, and the theoretical possibilities, some of which may be attained in the future, are shown diagrammatically in figure 34.

Figure 37. Self-pollinated ears grown on selected plants of Century Dent, No. 110. Each ear is the starting point of a selected line. These are numbered 1 to 9 top row and 10 to 18, bottom row, left to right.

SELECTION IN SELF-FERTILIZED LINES.

To demonstrate the value of inbreeding as a means of isolating good heredity a system of selection in self-fertilized lines was begun in 1918. Four varieties of corn were chosen as material with which to work. These varieties have been grown in Connecticut for many years and are well adapted. In a variety test of long duration they have proven to be among the best in production of grain and in other qualities. The four varieties are as follows:

Burwell's Yellow Flint, No. 30 and No. 40. An eight rowed yellow corn of the Canada Flint type. The ears are medium in size, one or two on the stalk. The plants are medium in maturity.