likely to be insistent. There can be no doubt that the practical breeders have made advances by selecting from the best individuals. No genetist or scientific breeder will deny this. It is simply the question of the interpretation of how the results were obtained that is in doubt and whether these results can be considered as permanent, new unit characters. Before we can thoroughly understand this subject it is probable that each individual case will require to be carefully analyzed, to determine the nature of the advance made and the interpretation of the process or processes concerned. At present we can only partially understand the phenomena presented.
It appears to me that we are dealing in breeding with two markedly distinct types of selection, based on different principles and arriving at different results, both right in principle and productive of equally valuable practical results, but of very different value, when considered from a strictly evolutionary standpoint.
It would seem that such cases of improvement as are illustrated by the sugar beet indicate that the continuous selection, generation after generation, of maximum fluctuations shown by a character, will result in maintaining a strain at nearly the maximum of efficiency; and that within a pure race the progeny of a maximum variate which would probably be classed as a fluctuation, does not regress entirely to the mean of the race in the first generation succeeding the selection, but that we only have a certain percentage of regression similar to the regression determined by Galton. It would further seem to be indicated by the evidence now available that in some cases we may even expect the continuously selected strain to exceed the ordinary maximum of the unselected population. In the Illinois corn experiments the maximum oil and protein content seems clearly to have exceeded the ordinary maximum, and is certainly maintained at a very high degree with a new mode and range of variation. If a new mutant of high protein content has been secured in the course of the experiments with a change of type it is probable that this high protein content will behave as a unit character in inheritance. Upon the other hand, if the results are interpreted as simply the maintenance by isolation of a strain produced by selecting fluctuations, there would probably be a rapid return to the normal range of variation of this character if the selection was discontinued.
De Vries has pointed out that natural selection can produce races and maintain them, but its power to develop races beyond the natural range of variability remains to be demonstrated.
With reference to his experiments with the potato beetle Tower states:
It is demonstrated that among the fluctuating variations there are individuals which are able to transmit their particular variation and give rise by selection to a race, while the majority are not able to hand on their particular conditions to their progeny. Races developed by selection from such variations have not been carried beyond the normal limit of variability of the species.
These races or selected strains maintain themselves as long as the selection is continued, and when the selection is discontinued rapidly regress to the mean of the species.
The above examples from the sugar beet, corn and potato beetle will illustrate the type of improvement usually secured by practical breeders. By their selection they maintain a strain of high efficiency without having in general exceeded the limits of variation of the species or race and without having produced new unit characters which would be maintained without selection and
