V. The phenomenon of the _reversion_ of the descendants of a hybrid to the paternal or maternal type, or _disordered variation_, has given rise to some interpretations which it will be well to rectify, and has also raised important questions. The attempt has been made to assimilate the latter to the _oscillations_ presented by mongrels for some generations. But daily experience should suffice to refute this opinion. Breeders are crossing _races_ every day for some purpose or other, and they would never do so if the crossing were to result in the production of a disorder which would exhibit the smallest resemblance to that displayed by the Linariæ of M. Naudin, and the silkmoths of Guérin Méneville. They expect, however, a few irregularities more or less marked, in the first generations, but they know that the race will soon _settle_ while the disorder would only increase if the crossing had taken place between _species_. Again, an attempt has been made to consider the facts of _atavism_ and _reversion_ as identical. There is, however, a fundamental difference between them, for the mongrel which by atavism reassumes the characters of one of its paternal ancestors, for example, still preserves its mixed nature. This is proved by the possibility of its offspring of the first or second generation reproducing, on the contrary, the essential traits of its own maternal ancestors. Darwin gives many examples of facts of this nature from the agricultural history of his country. One of the best to quote is that furnished by the genealogy of a family of dogs observed by Girou de Buzareingues. These animals were crosses between the setter and spaniel. Now one male, a setter to all appearances, united with a female of pure setter breed, produced spaniels, which makes it evident that the latter blood was by no means annihilated, and that the return to the setter type was only apparent. It is different in the cases of _reversion_ displayed by hybrids, for one of the two bloods is irrevocably expelled. We are justified in making this assertion in the case of mammalia, by experience extending as far back as the Roman period, or at least as far as the seventeenth century. Titires and musmons have never since those times had _offspring affected by atavism_. A ram and sheep have never been known to produce a kid, nor a male and female goat to produce a lamb. It is the same with plants, according to statements with which M. Naudin has kindly furnished me. Far from being similar, the phenomena of _atavism_ and _reversion_ are absolutely different and characteristic, the one of crossing between _races_, the other of crossing between _species_. The first proclaims the persistency of the physiological connections between all the representatives, more or less modified, of one species; the second proves the complete rupture of the same connections between the descendants of two species accidentally brought into contact by the promoter of the hybridism.