Journal of Heredity p. 95 (1918)
Washington, D. C.

FINAL selection of improved races by mere inspection has become, in most cases, more or less discredited among plant-breeders. The progeny test is now the main reliance of the breeder, and actual measurements are found to be usually more trustworthy than ocular estimates. Hence selection often includes some kind of testing before the act of choice. Without trial of some kind, selection consists in picking out the best looking individuals, in breeding for "fine feathers" and not for performance. This leads to pretty exhibits at the agricultural shows and county fairs, but does not cheapen the production of food. It is usually simple enough to pick out individuals which "look" better than the bulk of the crop, but to prove whether they are or are not better is often a longer process. In some cases the results of the progeny tests mechanically determine the best lines, with little or no personal selection on the part of the operator. Thus in improving Swedish wheat, Nilsson-Ehle takes at random plants of the old native wheat (which had already been shown to be a mixture of lines) and tests their progenies alongside for maximum crop, etc. Even in the progeny of crosses (e. g., potato) where the plants are multiplied as clons (by scions, cuttings, tubers, etc.), the testing of the likeliest looking plants is often the hardest part of the work. Selection of seed-plants is carried on in two directions: (1) to obtain superior lines and (2) to breed them to constancy. The more genetic variation we have, the more material there is for selection. Bud selection, so far, has only scored notable successes in a very few cases. Even in potatoes most, if not all, of the recognized market varieties originated in crosses or seedlings. Crossing is the recognized means of gaining material for selection. It seems probable, for example, that most, or all, of the many strains of the common bean (Phaseolus) and of the cowpea were selected from natural or artificial crosses. The nurseryman's class of "sports" usually includes the products of natural crossing by insects. In plants which are naturally much crossed, as maize, a modified mass selection must be practiced, because of the loss in productivity which usually follows self-pollination or inbreeding. But in predominantly selfed crops, as beans and wheat, mass selection is no longer used, but strict individual pedigrees are kept. In such cases, progress depends on never mixing the seeds of two or more plants until relative constancy is obtained.

In all cases characters are selected directly and plants indirectly, and all knowledge gained in the progress of the work, or previously, as to the mode of inheritance of these characters, shortens and cheapens the process. Thus, with the increase of our knowledge which is in large part due to American workers, plant-breeding will be a more certain and a less costly operation than in the past. Even in the present, probably many plant-breeders who have created assured values for the community, have lost money in the work. This has been stated to be the case even with Luther Burbank, whose work is primarily commercial, and not research.