Hairy red curved 91 47 42.2% Hairy red straight 15 7.5 14.1% Hairy white curved 23 12 14.1% Hairy white straight 17 8.5 4.7% Smooth red curved 23 12 14.1% Smooth red straight 9 4.5 4.7% Smooth white curved 5 2.5 4.7% Smooth white straight 12 6 1.6% The agreement is as comprehensive as might be expected from an experiment with about 200 plants, and there can be no doubt that a repetition on a larger scale would give still closer agreement. In the same way we might proceed to crosses with four or more differentiating characters. But each new character will double the number of the groups. Four characters will combine into 16 groups, five into 32, six into 64, seven into 128, etc. Hence it is easily seen that the size of the experiments must be made larger and larger in the same ratio, if we intend to expect numbers equally trustworthy. For [305] seven differentiating marks 16,384 individuals are required for a complete series. And in this set the group with the seven attributes all in a latent condition would contain only a single individual. Unfortunately the practical value of these calculations is not very great. They indicate the size of the cultures required to get all the possible combinations, and show that in ordinary cases many thousands of individuals have to be cultivated, in order to exhaust the whole range of possibilities. They further show that among all these thousands, only very few are constant in all their characters; in fact, it may easily be seen that with seven differentiating points among the 16,384 named above, only one individual will have all the seven qualities in pure active, and only one will have them all in a purely dormant condition. Then there will be some with some attributes active and others latent, but their numbers will also be very small. All others will split up in the succeeding generation in regard to one or more of their apparently active marks. And since only in very rare cases the stable hybrids can be distinguished by external characters from the unstable ones, the stability of each individual bearing a desired combination of characters would have to be established by experiment [306] after pure fertilization. Mendel's law teaches us to predict the difficulties, but hardly shows any way to avoid them. It lays great stress on the old prescript of isolation and pure fertilization, but it will have to be worked out and applied to a large number of practical cases before it will gain a preeminent influence in horticultural practice. Or, as Bailey states it, we are only beginning to find a pathway through the bewildering maze of hybridization.