Journal of Heredity, 3(4): 293-295 (Oct 1912)
J. ARTHUR HARRIS Cold Spring Harbor, Long Island, New York

In the practical growing of animals and plants both breeding and feeding are factors of great significance. In our enthusiasm over the possibilities of the newer methods in genetics, we are apt to forget that there must be limits to the improvements of the innate racial qualities which can be attained by hybridization or selection. These limits once reached, we are thrown back upon refinements in culture and feeding-upon the physiology of the individual as contrasted with the germinal constitution of the race—for further margins of improvement.

a See "A First Study of the Influence of the Starvation of the Ascendants upon the Characteristics of the Descendants. I-II." American Naturalist, vol. 46, pp. 313-343, 656-674, 1912.

For these reasons, I have always planned my breeding experiments to obtain incidentally as much information as possible on the strictly physiological factors influencing yield. Such factors are, for example, the influence of the environment of the parent plants furnishing the seed planteda upon the characteristics of the offspring, the influence of the size of the seed planted, etc.

The purpose of this note is merely to explain a diagram illustrating the importance of one of these physiological factors, namely, the weight of the seed planted, in determining yield.

b That is, class 3=0.050 to 0.075, or a mean of 0.0625; class 4=0.075 to 0.100; class 5=0.100 to 0.125; class 24 = 0.575 to 0.600.

The data for the diagram are drawn from twenty experimental crops of garden beans, involving Navy, White Flageolet and Ne Plus Ultra, represented by many thousands of individuals. The scale at the bottom shows the range of variation, of the weight of the seed planted, in units of 0.025 gram.b The vertical scale on the left hand side shows the mean number of pods per plant. The height (on the latter scale at the left of the diagram) at which the twenty sloping lines cut the ordinates (vertical lines) erected on the weight classes, gives the smoothed mean number of pods per plant for that weight of seed.

c Full details are given in a paper In vol. ix, part I of Biometrika: "On the Relationship Between the Weight of the Seed Planted and the Characters of the Plant Produced. I."

The actual means are of course very irregular, since the bean plant is very sensitive to its environment, for in ordinary field cultures, uniform conditions cannot be given. Moreover, exact agreements of the empirical and theoretical means are never secured because of the errors of sampling common to all statistics.c Of course, the lines as given here are to be looked upon merely as a conventionalized representation of the increase in mean number of pods per plant associated with an increase in the weight of the seed planted. But, considering the difficulties inherent in the materials, they are very accurate conventions.

The twenty series not only represent three distinct varieties but were grown under widely varying conditions and show in consequence great differences in the slope of the lines which express in concrete terms the mean number of pods per plant. Yet, in every case there is a conspicuous gain by the planting of heavier seeds.


Considerable attention has already been given by experiment station workers to the question of light and heavy seed, with the general result that the heavier seed gives the heavier yield. But generally, the lighter seeds have been separated by fanning and in many cases included blighted or shrivelled seeds. Here all seeds were perfect, as far as could be determined by individual examination.

The practicability of seed grading depends entirely upon the ratio of the cost to the returns from the increase in yield thus secured. These are problems which practical men must figure out. The purpose of this note will have been fulfilled if it suggests to the breeder the importance of planning his work so as to take more fully into account than is generally done, the purely physiological factors.