American Breeders Magazine 3(4): 311-312 (1912)
Gainesville, Florida.

One of the most pressing problems of today for the science of biology, as well as for the scientific plant‑breeder, is to discover whether genetic factors are constant or are variable. This can probably be determined by experiment, if executed with the precision of the physical laboratory. The practical question is, whether selection can have any measurable effect when carried out in two opposite directions within a homozygous strain. Leguminous plants that are functionally cleistogamic (self‑fertilized) seem to offer, in the dimensions of their dry seeds, suitable material for the experiment. Several sources of error, however, must be guarded against. Dry seeds measured in the spring, and again a month later, will sometimes be found to have shrunk perceptibly in the interim. The actual length of the dry seeds is also, in some strains, certainly determined by the amount of crowding in the pods.

The dimensions of a bean are maternal characteristics, and the only advantage of measuring more than one seed to each plant is to obtain an average which will be freer from the effects of "modifications." Now it seems that we should first reduce the modifications to as low a degree as possible. We can sometimes do this by taking our beans only from ripe pods of the average length and number of seeds, and by omitting the proximal and distal beans in strains of plants where these are often much smaller or larger respectively than the median beans. We should, I think, select in each generation those plants the average size of whose seeds is respectively highest or lowest. Any seeds from each of these selected plants can then be grown, and the selection of plants continued for further generations. Each of the variates used in our work will then always be the mean dimensions of the typical beans of one individual plant.

If we mix at first all the beans of the plants of the homozygous strain, and then select the largest beans for one line and the smallest beans for another line, we shall, in some leguminous plants at least, be testing the distal beans of a pod against the proximal beans, and we may perhap find no effects from such selection, even after ten years. The separate beans of a leguminous plant are certainly not individuals with regard to their dimensions, but are only limbs or members of the zygote on which they grow. The great modifications, in beans from different parts of the pod, and in pods with different numbers of seeds, will, in some leguminous plants at least, quite mask, for the purpose of selection, any small genetic differences which might or might not arise from variations of the individual genes.

It has been considered as proved, with Phaseolus vulgaris, that selection continued for several years produce no effect in different homozygous lines. But in these experiments, the separate beans, and not the individual plants, were selected.a Hence, I think that further search for a possible result of selection in pure lines (much less quantitatively than the immediate result of selection among the progeny of heterozygotes) will be useful.

  1. W. Johannsen, Elemente der exakten Erbilichkeitslehre, pp. 113-157, Jenna, 1909.