INFLUENCE OF ISOLATION ON THE FORMATION OF SPECIES 287 



a number of pregnant bitches, and place each of them upon an island 

 not previously inhabited by dogs, and tlien a different breed of dog 

 would arise upon each of these islands, even if the conditions of life 

 were exactly similar. But if, instead of these variable bitclies, the 

 females of a Russian wolf were placed on the islands, the developing 

 wolf -colonies would differ as little from the ancestral species as tlie 

 various Russian wolves do from one another— similar climate and 

 similar conditions of life being presupposed. 



There is thus an evolution of varieties due to amixia alone, and 

 we shall not depreciate the significance of this if we consider that 

 individual variations are the outcome of the fluctuations in the equili- 

 brium of the determinant system of the germ-plasm, to wliich it is 

 always more or less subject, and that variations of the germ-plasm, 

 whether towards plus or minus, bear within themselves the tendency 

 to go on increasing in the direction in which they have begun, and to 

 become definite variational tendencies. In isolated regions such varia- 

 tional tendencies must continue undisturbed for a long period, because 

 they run less risk of being suppressed b}^ mingling with markedly 

 divergent germ-plasms. 



The probability that variational tendencies set up in some ids of 

 the germ-plasm by germinal selection will persist and increase is 

 obviously greater the more the germ-plasms combining in amphimixis 

 resemble each other. For instance, let us call the varying deter- 

 minants Bv, and assume as a favourable case that these are represented 

 in three-fourths of all the ids in the fertilized eggs of a buttei-fly- 

 female whicli has been driven astray on to an island, that is, that they 

 are present in twelve out of sixteen ids ; then of i oo offspring of the 

 first generation it is possible that seventy-five or more will contain 

 the determinants Bv, some of them in a smaller number of ids, some 

 in a great number than the mother, according as the reducing division 

 has turned out. If the pairing of the second generation be favourable 

 — and this again is purely a matter of chance — a third generation 

 must arise which would contain the variants Bv througliout. and thus 

 the fixation of this particular variation on this particular island would 

 be begun. In other words, the possibility would arise, that, if indi\i- 

 duals with a majority of Bv ids predominated, they would gradually 

 come to be the only ones, since by continual crossing with the minority 

 which possessed only the determinants B, they would mingle the 

 varied ids with those of the descendants of these last, till ultimately 

 germ -plasm with only the old ids would no longer occur. 



In following out this process it is not necessary to assume that 

 the first innnigrant possessed the variation visibly ; if determinants 



