But again--if the case be Mendelian--the gametes borne by _AB_ will be either _A_’s or _B_’s[18], and the cross-bred _AB_’s breeding together will form _AA_’s, _AB_’s and _ BB_’s. Moreover, if as in the normal Mendelian case, _AB_’s bear on an average equal numbers of _A_ gametes and _B_ gametes, the numerical ratio of these resulting zygotes to each other will be 1 _AA_ : 2 _AB_ : 1 _BB_. [18] This conception was clearly formed by Naudin simultaneously with Mendel, but it was not worked out by him and remained a mere suggestion. In one place also Focke came very near to the same idea (see Bibliography). We have seen that Mendel makes no prediction as to the outward and visible characters of _AB_, but only as to the essential constitution and statistical condition of its gametes in regard to the characters _A_ and _B_. Nevertheless in a large number of cases the character of _AB_ is known to fall into one of three categories (omitting mosaics). (1) The cross-bred may almost always resemble one of its pure parents so closely as to be practically indistinguishable from that pure form, as in the case of the yellow cotyledon-colour of certain varieties of peas when crossed with green-cotyledoned varieties; in which case the parental character, yellow, thus manifested by the cross-bred is called “dominant” and the parental character, green, not manifested, is called recessive. (2) The cross-bred may present some condition intermediate between the two parental forms, in which case we may still retain the term “blend” as applied to the zygote. Such an “intermediate” may be the apparent mean between the two parental forms or be nearer to one or other in any degree. Such a case is that of a cross between a rich crimson Magenta Chinese Primrose and a clear White, giving a flower of a colour appropriately described as a “washy” magenta. (3) The cross-bred may present some form quite different from that of either pure parent. Though, as has been stated, nothing can be predicted of an unknown case, we already know a considerable number of examples of this nature in which the mule-form _approaches sometimes with great accuracy to that of a putative ancestor, near or remote_. It is scarcely possible to doubt that several--though perhaps not all--of Darwin’s “reversions on crossing” were of this nature. Such a case is that of the “wild grey mouse” produced by the union of an albino tame mouse and a piebald Japanese mouse[19]. These “reversionary” mice bred together produce the parental tame types, some other types, and “reversionary” mice again. [19] See von Guaita, _Ber. naturf. Ges. Freiburg_ X. 1898 and XI. 1899, quoted by Professor Weldon (see later).