Varieties have to be freed by selection from all such impurities, since isolation is a means which is quite impracticable under ordinary circumstances. Isolation is a scientific requirement that should never be neglected in experiments, indeed it may be said to be the first and most important requisite for all exact research in questions of variability and inheritance. But in cultivating large fields of allied varieties for commercial purposes, it is impossible to grow them at such distances from each other [197] as to prevent cross-pollination by the visits of bees. This purification must be done in nearly every generation. The oldest varieties are to be subjected to it as well as the latest. There is no regular amelioration, no slow progression in the direction of becoming free from these admixtures. Continuous selection is indispensable to maintain the races in the degree of purity which is required in commerce, but it does not lead to any improvement. Nor does it go so far as to become unnecessary in the future. This shows that there must be a continuous source of impurities, which in itself is not neutralized by selection, but of which selection can only eliminate the deteriorating elements. The same selection is usually applied to new varieties, when they occasionally arise. In this case it is called "fixing," as gardeners generally believe that through selection the varieties are brought to the required degree of purity. This belief seems to rest mainly on observations made in practice, where, as we have seen, isolation is of very rare application. Most varieties would no doubt be absolutely pure from the first moment of their existence, if it were only possible to have them purely fertilized. But in practice this is seldom to be obtained. Ordinarily the breeder is content with such slow [198] improvement as may be obtained with a minimum of cost, and this mostly implies a culture in the same part of the nursery with older varieties of the same species. Three, four or five years are required to purify the novelty, and as this same length of time is also required to produce sufficient quantities of seed for commercial purposes, there is no strong desire to shorten the period of selection and fixation. I had occasion to see this process going on with sundry novelties at Erfurt in Germany. Among them a chamois-colored variety of the common stock, a bluish _Clarkia elegans_ and a curiously colored opium-poppy may be mentioned. In some cases the crossfertilization is so overwhelming, that in the next generation the novelty seems entirely to have disappeared.