240 GEOGRAPHICAL DISTRIBUTION [pt. ii 



always been successful, the assumption is therefore probably 

 correct. 



The endemics, then, of course with a good many exceptions, 

 are in general to be regarded as derived from the wides that 

 occur among them. In many cases, as we have seen, and those 

 most often cases in which there is reason to suspect greater age 

 than usual, a genus in any single country may have only endemic 

 species (cf. pp. 95, 155), sometimes only one, sometimes more, 

 and in these cases we may suppose some mutation, perhaps at 

 once, in the first wide to arrive, or perhaps subsequently and 

 en masse. 



But now, if, in general, the appearance of a new form does not 

 imply, as it did under the theory of natural selection, the dis- 

 appearance of its ancestral forms, there seems little reason why 

 both should not survive upon the earth, or, in other words, why 

 the whole, or great part, of the tree of a family or genus should 

 not survive. I have already worked out this question in regard 

 to the Dilleniaceae (120), suggesting that Tetracera, a wide- 

 spread and very simple genus, may have been the ancestral 

 form from which the family was derived. In the same waj' the 

 Polemoniaceae (p. 171) might have been derived from Pole- 

 monium, the INIenispermaceae from Cocculus and Cissampelos, 

 whilst in Cissampelos itself, C Pareira (p. 159) might have been 

 the parent of the other species, directlj- or indirectly. In Doona 

 (p. 152), D. zeylanica may in the same way be looked upon as the 

 probable parent, direct or indirect, of the other species, and 



It is clear that -when once the general principle of Age and 

 Area is established — and already the e^-idence in its favour is 

 \exy strong — it may be called into service in the study of 

 phylogeny. But if it be accepted, it is clear that Guppy's 

 Theory of Dijferentiation (p. 221) must almost necessarily be 

 accepted also. This subject will be dealt with in a later book 

 upon Evolution generally, and can only be mentioned here. 



Just as the endemics of small area are to be looked upon as 

 descended from species of larger area, so also we have seen that 

 the monotypic genera are to be looked upon in general as 

 descended from larger genera. The Avay in which the nimibers of 

 genera, not only in the total, but family by family, are arranged 

 (cf. p. 187) in hollow cur\-es, with a great preponderance of mono- 

 types and steady decrease to a few of large numbers, shows that 



