[21] It will be understood from what follows, that the existence of mosaic zygotes is no _proof_ that either component gamete was mosaic. * * * * * We have now seen the essential nature of the Mendelian principles and are able to appreciate the exact relation in which they stand to the group of cases included in the Law of Ancestral Heredity. In seeking any general indication as to the common properties of the phenomena which are already known to obey Mendelian principles we can as yet point to none, and whether some such common features exist or not is unknown. * * * * * There is however one group of cases, definite though as yet not numerous, where we know that the Mendelian principles do not apply. These are the phenomena upon which Mendel touches in his brief paper on _Hieracium_. As he there states, the hybrids, if they are fertile at all, produce offspring like themselves, not like their parents. In further illustration of this phenomenon he cites Wichura’s _Salix_ hybrids. Perhaps some dozen other such illustrations could be given which rest on good evidence. To these cases the Mendelian principle will in nowise apply, nor is it easy to conceive any modification of the law of ancestral heredity which can express them. There the matter at present rests. Among these cases, however, we perceive several more or less common features. They are often, though not always, hybrids between forms differing in many characters. The first cross frequently is not the exact intermediate between the two parental types, but may as in the few _Hieracium_ cases be irregular in this respect. There is often some degree of sterility. In the absence of fuller and statistical knowledge of such cases further discussion is impossible. * * * * * Another class of cases, untouched by any hypothesis of heredity yet propounded, is that of the false hybrids of Millardet, where we have fertilisation without transmission of one or several parental characters. In these not only does the first cross show, in some respect, the character or characters of _one parent only_, but in its posterity _no reappearance of the lost character or characters is observed_. The nature of such cases is still quite obscure, but we have to suppose that the allelomorph of one gamete only developes after fertilisation to the exclusion of the corresponding allelomorph of the other gamete, much--if the crudity of the comparison may be pardoned--as occurs on the female side in parthenogenesis without fertilisation at all.