EVOLUTION 423 



structed from the characters or organs of the two parents 

 and termed a mid-parent; such mid-parents will be discussed 

 later (see p. 470), but the point to be noticed now is that 

 no change in the frequency polygon will take place if the 

 mid-parents are formed by pairs of individuals mating at 

 random. We can, however (see p. 485), establish a breed 

 or stock giving offspring of the same type (with of course 

 a definite amount of variability) from generation to 

 generation, if we select not merely an individual or mid- 

 parent a to breed from, but continue this selection for 

 several reproductive stages. 



Now, suppose owing to a change of environment, that 

 the modal type c is destroyed, and that owing to direct 

 selection of this character, or to the selection of one or 

 more correlated characters, a and b tend to become differ- 

 entiated types. Then any race in which a does not cross 

 with b will tend to become permanently differentiated ; 

 hence organisms with self-fertilisation, with endogamous 

 habits, or where like mates with like, will, with divergent 

 speed depending in part upon the stringency of the selec- 

 tion, be able to establish distinct stocks. Without such 

 selection, however, neither self-fertilisation nor mating of 

 like with like necessarily connotes a change of type. 

 Natural selection requires selective mating, sexual selec- 

 tion in its broadest sense, to produce that barrier to inter- 

 crossing on which the origin of species depends. Darwin 

 used sexual selection in the sense of the total or partial 

 rejection of one type of mate by one or other sex. This 

 I should prefer to term preferential mating. 



We may, therefore, classify the forms of selective mating 

 in the following manner : — 



{a) Autogamic mating,^ or self-fertilisation. 



1 The term autogamy is here used to cover all fertilisation within the 

 individual, — the individual being defined in a much narrower sense than that 

 of Huxley. Thus the cross-fertilisation of flowers on the same plant as well 

 as the self-fertilisation of cleistogamic flowers would be considered autogamic. 

 Endogamy embraces mutual fertilisation in the case of separate but related 

 individuals. I am aware that the mutual fertilisation of difi'erent flowers on 

 the same plant has been looked upon as a case of endogamy. Professor 

 Weldon suggests to me that the latter term should be used to cover the whole 

 ground, and subdivided as follows : (i.) Floral or Thalamic; (ii.) Somatic; 



