These figures are very striking. From the same quantity of seed, in equal spaces, by similar exposure and treatment I got 10 fully developed instances in one and 32 in the other case. The weeding out of supernumerary individuals had not only increased the percentage of bright crowns, but also their absolute number per square meter. So the greatest number of anomalies upon a given space may be obtained by taking care that not too many plants are grown upon it: any increase of the number beyond a certain limit will diminish the probability of obtaining these structures. The most successful cultures may be made after the maximum number of individuals per unit of area has been determined. A control-experiment was made under the same conditions and with the same seed, but allowing much less for the same space. I sowed only 1 cu. cm. on my bed of 2 square meters, and thereby avoided [395] nearly all weeding out. I got 120 plants, and among them 30 with full crowns of converted stamens, practically the same number as after the weeding out in the first experiment. This shows that smaller quantities of seed give an equal chance for a greater number of large crowns, and should therefore always be preferred, as it saves both seed and labor. Weeding out is a somewhat dangerous operation in a comparative trial. Any one who has done it often, knows that there is a strong propensity to root out the weaker plants and to spare the stronger ones. Obviously this is the best way for ordinary purposes, but for comparisons evidently one should not discriminate. This rule is very difficult in practice, and for this reason one should never sow more than is absolutely required to meet all requirements. Our second point is the manuring of the soil. This is always of the highest importance, both for normal and for anomalous attributes. The conversion of the stamens into pistils is in a large measure dependent upon the conditions of the soil. I made a trial with some 800 flowering plants, using one sample of seed, but sowing one-third on richly manured soil, one-third on an unprepared bed of my garden, and one-third on nearly pure sand. In all other respects the three groups were treated in the same way. Of [396] the manured plants one-half gave full crowns, of the non-manured only one-fifth, and on the sandy soil a still smaller proportion. Other trials led to the same results. I have often made use of steamed and ground horn, which is a manure very rich in nitrogenous substances. One-eighth of a kilo per square meter is an ample amount. And its effect was to increase the number of full crowns to an exceptional degree. In the controlling trial and under ordinary circumstances this figure reached some 50%, but with ground horn it came up as high as 90%. We may state this result by the very striking assertion that the number of large crowns in a given culture may be nearly doubled by rich manure.