sive characters, any pair of individuals manifesting that combination will breed true. But if the desired combination contains one or more dominant characters, then each animal selected must be tested for the presence of undesirable recessives before one can be sure that the new race will breed true. In practise it is found best by the breeder not to work with too many characters at a time, but to eliminate the undesirable recessives one by one. Otherwise the search for the one individual in a large number which will breed true may prove a long and tedious process. If we deal with one character at a time, the chances are that one in four of the second generation of animals reared will meet our ideal; if we deal with two characters at a time, the chances are one in sixteen; while if we deal with three characters at a time the chances are only one in sixty-four; and so on, with the chances of success diminishing in a geometrical series.
From what has thus far been said it would appear that in alternative inheritance characters behave as units, and, more than that, as wholly independent units, so that to forecast the outcome of matings is merely a matter of mathematics. While this is in a measure true, it is, fortunately or unfortunately, not the whole truth. In alternative inheritance characters do behave as units independent of one another, but the union of dominant character with recessive in a cross-bred animal is not so simple a process as putting together two pieces of glass, nor is their segregation at the formation of gametes so complete in many cases ae: the separation of the two glass plates. The union of maternal and paternal substance in the germ-cells of the cross-bred animal is evidently a fairly intimate one, and the segregation which they undergo when the sexual elements are formed is more like cutting apart two kinds of differently colored wax fused in adjacent layers of a common lump. Work carefully as we will, traces of one layer are almost certain to be included in the other, so that while the two strata retain their identity, each is slightly modified b} r their previous union in a common lump.
Thus, when we cross short-haired with long-haired guinea-pigs, we get among the second-generation offspring a certain number of long-haired animals with hair less long than that of the long-haired grandparent, or with long hair on part of the body only (Fig. 13). Further, certain of the short-haired animals have hair a little longer and a little softer than that of the short-haired grandparent. Again, rough-coated guinea-pigs produced by cross-breds often have coats less fully rough than that of their rough ancestor, lacking certain of the typical rosettes (Fig. 14). Finally, when an albino is crossed with a fully pigmented animal, the result may be not a wholly pigmented animal, but one spotted with white. While such a cross-bred animal forms a full quota (one-half) of albino gametes, the pigment-bearing gametes formed by it frequently bear this spotted or modified pigmented condition.
