Figure 56.

112-7, 112-8) produced no moldy ears. Only one line 110-2 had all plants erect with ears free from mold. In the first generation this line had four per cent. of moldy ears and ten per cent. of fallen plants but no smut. In the second generation there were ten per cent. moldy ears, ten per cent. fallen plants and no plants showing smut infection. In the third, fourth and fifth generations there was no mold, smut or fallen plants on the three progenies grown each year. This strain is also productive for the variety, although surpassed in this respect by several other strains. The seeds are hard and bright but very pale yellow in color and almost white on top.

In contrast to this is line 105-20 with all the plants erect in the second, third and fourth generations but with 29, 17 and 44 per cent. of moldy ears in the same years. On the other hand, 40-8 had all of the plants fallen in two progenies of the fourth generation and no moldy ears. To complete the combinations 112-11 had 87 per cent. of the plants on the ground in 1922 and 67 per cent. of ears moldy.

CRITERIONS OF SELECTION.

At the beginning of the selection experiment the plan as previously stated was to self-pollinate five plants in each line and to select three of the best self-fertilized ears for planting the following year. Even when these ears differed greatly in appearance no consistent differences were noted in the progenies grown from them. The coefficient of association between the appearance of the ear and yield of the different progenies within several lines is-.18. This indicates that self-pollinating a large number of ears in order to make more extensive selection of desirable looking ears is of doubtful value. Of the three progenies grown only one was to be chosen to continue the line, the other two not being pollinated. It was soon noted, however, that there was very little relation between the appearance of the progenies at the time of bagging and their production of grain and the general appearance of their plants at harvest. The coefficient of association between the appearance of the plants at pollinating time and the yield of the different progenies within the several lines is 28. Seedlings were grown in the greenhouse and their weight and height after thirty days of growth were compared with the yield of the same progenies in the field. The third and fourth generations showed that those progenies that had the tallest seedlings yielded 1.6 bushels per acre more than the other progenies in the same lines. This difference is hardly enough to make a selection of the progenies on this basis worth while.

Since there is no appreciable correlation between the characters of the seed ear, weight of seed, size of the seedling, or the appearance of the plants at pollinating time and production of grain the only selection of progenies that can be made with any degree of

effectiveness is at maturity. Here also yield is highly influenced by the amount of heterozygosity remaining. In some lines there are more homozygous combinations than in others and they are correspondingly less vigorous and productive although they may be potentially more desirable. For this reason final judgment must be left until the plants are reduced to uniformity and constancy. Hence it is interesting to note what resemblance the resulting inbred strains, when finally reduced to uniformity and fixity of type, have to the same strains in the first generations of inbreeding.

CLASSIFICATION OF SELECTED LINES.

Taking into consideration all features of these selected lines as they grow in the field and after harvest in the fourth and fifth generations and giving most importance to the production of bright sound grain, the four outstanding good and poor strains in each variety are listed as follows, with their yields in bushels per acre in the fifth generation compared with that of the original variety grown the same year:

This is purely an arbitrary classification based upon the general appearance of the plants and ears. Some of the poor lines yielded more than the good lines but produced a very poor quality of grain. The original ears from which these lines descended (figures 35 to 38) show that there is no relation between the good and poor strains after uniformity was attained and the appearance of the seed ears from which they came. Low and high numbers are represented about equally in the good and poor strains.

CORRELATION Between the First and Last Generations.

In order to find out whether the elimination of the poor lines at the beginning of the inbreeding period is advisable, the correlation

TABLE VIII.

Coefficients of association between early and later generations of selffertilized corn.

Generations Compared

Variety

Burwell's Flint..
Gold Nugget.
Century Dent..
Beardsley's Leaming.

豪

Ave. Flints...
Ave. Dents.
Average.

1-4 Height

1-4 Mold

1-5 Tillers

2-5 Smut

.60

.89

.64

.38

.38

Yield 0 .14

.80

.28

.95

.50

.50

11111

between the behavior of the plants in the first generation and the last generation has been worked out for the most important characters. In Table VIII are shown the coefficients of association

Figure 58. Diagram representing the average of the upper and lower groups in the first generations and the average of the same lines in the last generations, based on the data in Table IX.

between the first or second inbred generation and the fourth or fifth for height of plant, per cent. moldy ears, number of tillers, per cent smutted plants, and yield of grain. The fifth generation, grown in 1923, was so variable on account of the extremely dry season affecting different parts of the field unevenly that the coefficients for height and yield are based on the first and fourth generations. There was very little smut infection in the first