American Breeders Association 6: 5-11 (1909)
University of Illinois.


The purpose in presenting at this time a matter which may not be altogether new to most of our members is to bring before this Association the latest results of what are probably the oldest corn-breeding experiments in existence.

Some of the records now cover thirteen generations, and to those who have been watching the progress of the work, the interest naturally grows more intense as the years go by. The records become more and more valuable as they accumulate as throwing light, not alone upon those questions connected specifically with corn improvement, but also upon some of the broader general problems of heredity.

In 1896 the Illinois Experiment Station took up the proposition to influence the chemical composition of the corn kernel by selection of the seed. It will scarcely be necessary to give here more than a mere outline of the general plan of the work. The history and results of these experiments for the first ten generations are given in detail in Bulletin 128 of the Illinois Experiment Station.

The plan proposed was to breed for four different purposes, namely, (1) increase of protein; (2) decrease of protein; (3) increase of oil; (4) decrease of oil; the selection being based upon the analysis of individual ears of a single variety. The ears thus selected for the several purposes were planted together in isolated breeding plots by the well known ear-to-row method. From each of these plots selection has always been kept up in the same general manner by analysis of individual ears.


An outline of the results obtained in the breeding for increase and decrease of protein content may be seen in Table 1.

Table 1 gives the average protein content of the crop produced on each plot each generation, thus giving a good general view of the progress of the work.

A glance at the figures shows that there has been great response to the selection in both directions, so that the general effect has been such that we have been able to produce, out of a single variety, two strains of corn, one of which contains more than half again as much protein as the other.

TABLE 1.—Increase and decrease of protein.

Year. High protein plot average in crop harvested Low protein plot average in crop harvested Difference between crops.
  Per cent. Per cent. Per cent.
1896 10.92 10.02 0.00
1897 11.10 10.55 0.55
1898 11.05 10.55 0.50
1899 11.46 9.86 1.60
1900 12.32 9.34 2.98
1901 14.12 10.04 4.08
1902 12.34 8.22 4.12
1903 13.04 8.62 4.42
1904 15.03 9.27 5.76
1905 14.72 8.57 6.15
1906 14.26 8.64 5.62
1907 13.89 7.32 6.67
1908 13.94 8.96 4.98

It is important to notice in passing that there has been at times a very pronounced seasonal influence upon the protein content. For example, the high protein tendencies are brought out very distinctly in the results for the years 1901 and 1904. On the other hand, 1900 and 1907 were years favorable to low protein.


Even more striking are the results obtained in the breeding to influence the oil content as shown in the more regular and uniform response to the selection and in the greater proportionate changes produced.

Table 2 gives the record of this work. Here it is shown that the general effect has been to produce out of this same variety two other strains of corn, one of which is now practically three times as rich in oil as the other.

It is of especial interest to observe that both in the oil and in the protein breeding the limits appear to have been reached. In the case of the high protein the high percentage in the crop of 1904 has never since been attained. In the low-protein plot the minimum percentage thus far obtained was in the crop of 1907. In the high-oil strain, there was a drop in the percentage of the last year's result, but in the case of the low oil the extreme point is represented in the last year. On the whole, however, if we consider the last four or five generations, the results appear to be fluctuating around certain points, varying back and forth with the season and soil conditions.

TABLE 2.—Increase and decrease of oil.

Year. High oil plot average in crop harvested. Low oil plot average in crop harvested. Difference Between crops.
  Per cent. Per cent. Per cent.
1896 4.70 4.70 0.00
1897 4.73 4.06 0.67
1898 5.13 3.99 1.16
1899 5.61 3.82 1.82
1900 6.12 3.57 2.55
1901 6.09 3.43 2.66
1902 6.41 3.02 3.39
1903 6.50 2.97 3.53
1904 6.07 2.89 4.08
1903 7.20 2.58 4.71
1906 7.37 2.66 4.71
1907 7.43 2.59 4.84
1908 7.19 2.39 4.80

On account of these environmental influences, it cannot be decided yet whether there may not still be further advance possible in some of these directions, and there is still as much interest as ever, if not more, in the continuation of these experiments along the same lines.


The question that this work most frequently calls out is in regard to the permanency of these characters. Will it require constant selection to retain these qualities or will they revert to the original condition if left without special selection? Some interesting data bearing upon this point have been obtained in connection with the high-oil corn.

In 1903, after the high-oil strain had been selected for six generations and the oil content had been increased up to 6.50 per cent, two selections of ears from this plot were made for other purposes quite independent of oil content, namely, for erect ears and for declining ears. These two new strains have been bred in separate plots since that time, with no attention whatever given to percentage of oil in the selection of seed. Incidentally we have, however, analyzed the crops harvested and the results of these analyses are given in Table 3.

TABLE 3.—Composition of erect ear and declining ear plots.

Year and strain Oil.
    Per cent.
1896 Original corn 4.70
1903 High oil 6.50
1905 Erect ear 6.90
  Declining ear 6.60
1907 Erect ear 6.65
  Declining ear 6.23
1908 Erect ear 6.37
  Declining ear 6.02

From these results it appears that in 1905 after two generations of non-selection there was an actual increase of oil; in 1907 after four generations the percentage was about the same as at the beginning of the new plots. In 1908 there was a decrease apparent which may quite possibly be the effect of a "low-oil season."

On the whole, we may say that after five generations of non-selection, there is certainly no great decrease of oil content, but in view of the fluctuations due to environment, it is too early to draw final conclusions in regard to this point.

A theoretical discussion of great interest is involved in these data, but it is the present purpose simply to put on record here these facts as they have been found, hoping to have the opportunity at some later date to present for more thorough discussion this phase of the experiments.


Some other interesting examples of what may be accomplished by continuous selection for special purposes are furnished by some experiments designed to influence certain physical characters of the plant, namely, the height of the ear on the stalk and the position of the ear at maturity, with reference to the angle at which it hangs.


Seven years ago two lots of ears were selected from an ordinary corn field, one of these lots representing ears growing high on the stalk and the other those borne low down on the stalk. These two sets of ears were planted in separate breeding plots, and selection for high ears and for low ears from the respective plots has been made each year since.

The general results of this work are shown in Table 4.

TABLE 4.—Breeding for high cars and low cars.
[Average height in inches.)

Year. High ear plot. Low ear plot. Difference.
1903 56.4 42.8 13.6
1904 50.3 38.3 12.0
1905 63.3 41.0 21.7
1906 56.6 25.5 31.1
1907 72.4 33.2 39.2
1908 57.3 23.1 34.2
1909 64.3 25.3 30.0

Here again as in the case of the selection for composition of grain, there has been a gradual response so that by breeding this variety in opposite directions two strains of corn have been produced in one of which the ears are now borne about three feet higher on the stalk than in the other strain.

Incidentally it is of interest to notice in this connection the correlation existing between the height of ear and the total length of plants, the total number of internodes, and the average length of internodes. Selection for high ears has produced a taller, later maturing plant than that resulting from low-ear selection; and interesting enough from the practical standpoint, the yields from these two strains are thus far about equal.


Another character that has likewise responded in a striking manner to seed selection is the declination of the ear at the time of maturity. The details of the plans and the early results of this work, together with those of the preceding experiments will be found in Bulletin 132 of the Illinois Agricultural Experiment Station.

The results now cover six generations, and they appear in outline in Table 5.

TABLE 5.—Breeding for erect cars and declining ears.
[Average angle of declination from staIk.]

Year. Erect ear plot. Declining ear plot. Difference.
  Degrees. Degrees. Degrees.
1904 42.0 45.0 3.0
1905 62.2 117.1 54.9
1906 40 5 76.2 26.7
1907 42.3 91.6 39.3
1908 46.0 88.5 12.5
1909 31.2 110.7 79.5

We observe from these results that, with the exception of 1905, which appears in this respect to be an abnormal season, there has been a steady progressive response to the selection, until finally after six generations the average difference in the angle amounts to almost 80 degrees.


Aside from their practical significance, there is a peculiar interest attached to these results as representing in every instance what has been accomplished by the method of breeding, sometimes designated as "continuous selection," a method in which some investigators would appear to have little faith. It should of course be recognized in this connection that we are dealing here with an open-fertilized plant. It would be interesting to have on actual record the results of parallel experiments with some self-fertilized plant such as wheat.

[Professor Smith's article splendidly represents those well-conceived experiments in heredity and breeding which are placing these subjects on a scientific basis and are showing that large economic results may be secured. Though this particular experiment was not begun with such a thought in mind, it may be that strains of corn thus made highly efficient and pure-bred in specific lines may be found useful as the basis of hybrid breeding. In other words, the characters thus highly intensified and made pure-bred may be recombined in first-year hybrids, or in fixed hybrids, which have the desired very large value per acre. This also is an illustration of the fact that scientific work which is prosecuted primarily for science's sake often leads to economic results of large value. Here not only is the scientific proof secured that profound changes in the heredity of corn may be made, but that varieties which were not bred for all round per acre value may prove useful as the constituent units of new varieties.—W. M. H.]

"An experiment with plants that is often held up as being particularly convincing was launched in 1903 at the Illinois Agricultural Experiment Station and continued until 1927. A population of corn was grown, the parent for the next generation always being the plants with ears closest to the ground. At the start, the average height of ears above the ground ranged from 43 to 56 inches. At the end of twenty-four years, the average had become a mere eight inches, and this trait bred true. As a check, plants with the highest ears were also selected and bred; in this case the average height rose to 120 inches —10 feet — by the end of the experiment."

Taylor, Gordon Rattray. Great Evolution Mystery (London: Sacker and Warburg, 1983), p. 34.

BulbnRose note: Compare these results with Cooper's work on 'Wimmera' rye grass.