If a third character were considered, the propor tions of the combinations can be determined in exactly the same way. Each one of the above nine possible combinations would be again divided into three different unions in the same way as the three combinations of the one character-pair gave nine different combinations in the second character pair. In the consideration of the three character pairs there would thus be 27 different combinations of parental characters. And again in each ovary fecundated, when only one potentiality of each character-pair occurred, the opposing character potentiality being in each case eliminated, such a cell should give a plant that would reproduce its characters true to type. It is well known that almost any two different races or species that may be chosen for hybridization will ordinarily differ from each other in numerous characters. When there are a number of these opposing characters which form Mendelian character-pairs, the deter mination of the possible combinations by Mendel's formula; becomes very complex and difficult to understand. It is only by taking a few well marked character -pairs and carefully studying them that the segregation and new combinations according to Mendelian proportions can be followed and understood. Any character -pairs, following Mendel's law, would segregate as indicated above in the case of bald or bearded heads and smooth and hairy chaff of wheat. These characters with wheat have been investigated by Spillman, Hurst and others, and are known to follow very closely Mendelian proportions in their segregation. The same segregation takes place in the case of the bald and bearded barleys, smooth and fuzzy cottons, sweet and starchy kernels in corn, and many other opposed characters in plants.
It is by no means probable that all characters follow Mendel's law of segregation and recombina tion, and secondary characters in practical work need be given no attention. The knowledge of Mendel's principles may not change greatly the practical methods of breeding which have been followed for a number of years, but they give us a more thorough comprehension of what we are do ing, and also greater surety that certain combina tions of parental characters can be secured.
(2) The use and fixation of intermediate or blended types.
The principle of the purity of the germ-cell, if strictly applied, would not recognize as possible the fixation into a race reproducing true to type of an intermediate hybrid, that is, one in which two characters of a certain pair are blended. Yet practical work shows that such a fixation certainly can be secured. In very many hybrids of plants cultivated for their flowers, intermediate colors have been bred to stability, showing that the inheritance is blended. The writer has been at tempting to fix a hybrid of Black Mexican sweet corn having blue-black kernels, with Stowell's Evergreen, which has a nearly white kernel, into a race of light blue-violet color, and strictly inter mediate in this respect between the two parental varieties. Ordinarily, the color of these hybrids breaks up in Mendelian proportions, but neither color can be considered to be dominant in the true sense of the word. In practically all cases when
the potentialities of the two characters are mixed in the same egg-cell, the coloration is intermediate rather than like one or the other of the parent vari eties. The writer has uniformly selected the seed of such intermediate light blue-violet kernels for planting, and has kept the patch completely iso lated. After four years of such selection, a type that produces nearly uniformly light blue colored kernels has been produced. There are still many reversions to the coloration of either parent, but these are growing fewer and the type is becoming fixed into a stable race, reproducing itself true to seed. Halsted, of the New Jersey Experiment Station, has produced such an intermediate colored race by the hybridization of Black Mexican with the Egyptian, and has already secured a new race which is practically fixed in its intermediate color. The writer thinks that in this and in a great many other cases it is possible by careful selection of plants showing the intermediate type to breed new races that exhibit a blend of characters, and such blends are frequently of great value.
The work that has been carried out by the writer in the Department of Agriculture in the breeding of citrons fruits very clearly indicates that valuable intermediates may sometimes be secured. The writer, in conjunction with Mr. Walter T. Swingle, hybridized the hardy, cold resistant trifoliate orange (Citrus trifoliata) with several varieties of the tender sweet orange, and as a result at least five different varieties of hardy oranges or citranges have been produced (Fig. 86). These hybrids are nearly intermediate between the two parents, having the characters in the first generation nearly blended. The leaves are trifolioliate, but are much larger than the leaves of the ordinary trifoliate orange tree, and show a tendency to drop off, the lateral leaflets producing an unifolioliate leaf. The trifoliate orange is decid uous, while the sweet orange is evergreen. The hy brids are semi-deciduous, holding a large share of their leaves through the winter. In hardiness they also seem to be intermediate, being much more cold-resistant than the ordinary orange, but not so hardy as the trifoliate orange. They are suffi ciently hardy so that they doubtless may be grown with safety as far north as South Carolina, or 300 to 400 miles north of the present orange region. Some of the fruits produced are as large as the ordinary orange, but the majority are very nearly intermediate in size. They are very variable, how ever, in the first generation. At least five of the fruits that have been produced are juicy and valuable. It is not probable that they would be reproduced true to seed, but orange varieties are dons, and the different types will, of course, be normally repro duced by buds or grafts, so that from a practical standpoint it does not matter whether or not they would repro duce true through the seed. In the second generation it is probable that these different characters would split up, possibly according to Men del's law, and it is likely that still more valuable va rieties will be secured when a second generation has been grown.