It seems quite superfluous to give more details. In any case in which the selection is not done before the blooming period, some impurities must result. Even if it is done before that time, errors may occur, and among hundreds and thousands of individuals a single anomalous one may escape observation. The conclusion is, that flower seeds as they are offered in commerce, are seldom found absolutely pure. Every gardener knows that he will have to weed out aberrant plants in order to be sure of the purity of his beds. I tested a large number of samples of seeds for purity, bought directly from the best seed growers. Most of them were found to contain admixtures and wholly pure samples were very rare. I will now give some illustrative examples. From seeds of a yellow snapdragon, I got one red-flowered specimen among half a hundred [195] yellow ones, and from the variety "Delila" of the same species two red ones, a single white and two belonging to another variety called "Firefly." _Calliopsis tinctoria_ has three varieties, the ordinary type, a brown-flowered one and one with tubular rays. Seeds of each of these three sorts ordinarily contain a few belonging to the others. _Iberis umbellata rosea_ often gives some white and violet examples. The "Swan" variety of the opium-poppy, a dwarfish double-flowered form of a pure white, contained some single-flowered and some red-flowered plants, when sown from commercial seed are said to be pure. But these were only occasional admixtures, since after artificial fertilization of the typical specimens the strain at once became absolutely pure, and remained so for a series of generations, as long as the experiment was continued. Seeds of trees often contain large quantities of impurities, and the laciniated varieties of birch, elder and walnut have often been observed to come true only in a small number of seedlings. In the case of new or young varieties, seed merchants often warn their customers as to the probable degree of purity of the seeds offered, in order to avoid complaints. For example the snow-white variety of the double daisy, _Bellis perennis plena_, was offered at the start as containing [196] as much as 20% of red-flowered specimens. Many fine varieties are recorded to come true from seed, as in the case of the holly with yellow fruits, tested by Darwin. Others have been found untrue to a relatively high degree, as is notorious in the case of the purple beech. Seeds of the laciniated beech gave only 10% of laciniated plants in experiments made by Strasburger; seeds of the monophyllous acacia, _Robinia Pseud-Acacia monophylla_, were found to be true in only 30% of the seedlings. Weeping ashes often revert to the upright type, red May-thorns (_Crataegus_) sometimes revert nearly entirely to the white species and the yellow cornel berry is recorded to have reverted in the same way to the red berries of the _Cornus Mas_.