The choice of the fullest or smallest crowns may be repeated during succeeding generations, and each time compared with a culture under average conditions. By this means we come to true selection-experiments, and these result in a notable and rapid change of the whole strain. By selecting the brightest crowns I have come up in three years from 40 to 90 and ultimately to 120 converted stamens in the best flower of my culture, and in selecting the smallest crowns I was able in three years to exclude nearly all good crowns, and to make cultures in which heads with less than half-filled crowns constituted the majority. But such selected strains always remain very sensitive to treatment, and by changing the conditions the effect may be wholly lost in a single year, or even turned in the contrary direction. In other words, the anomaly is more dependent on external conditions during the germinating period than on the choice of the seeds, providing these belong to the pistilloid variety and have not deteriorated by some crossing with other sorts. At the beginning of this lecture I stated that [399] no selection is adequate to produce either a pure strain of brightly crowned flower-heads without atavism, or to conduce to an absolute and permanent loss of the anomaly. During a series of years I have tested my plants in both directions, but without the least effect. Limits are soon reached on both sides, and to transgress these seems quite impossible. Taking these limits as the marks of the variety, and considering all fluctuations between them as responses to external influences working during the life of the individual or governing the ripening of the seeds, we get a clear picture of a permanent ever-sporting type. The limits are absolutely permanent during the whole existence of this already old variety. They never change. But they include so wide a range of variability, that the extremes may be said to sport into one another, so much the more so as one of the extremes is to be considered morphologically as the type of the variation, while the other extreme can hardly be distinguished from the normal form of the species. [400] LECTURE XIV MONSTROSITIES I have previously dealt with the question of the hereditary tendencies that cause monstrosities. These tendencies are not always identical for the same anomaly. Two different types may generally, be distinguished. One of them constitutes a poor variety, the other a rich one. But this latter is abundant and the first one is poor in instances of exactly the same conformation. Therefore the difference only lies in the frequency of the anomaly, and not in its visible features. In discovering an instance of any anomaly it is therefore impossible to tell whether it belongs to a poor or to a rich race. This important question can only be answered by direct sowing-experiments to determine the degree of heredity.