HEREDITY 223 



" belong " to the same species often inhabit some restricted 

 region. They all resemble each other more than they resemble 

 the individuals that, we say, belong to other species. They are 

 all mutually and indefinitely fertile with each other if they repro- 

 duce sexually and they tend to be immediately or ultimately 

 infertile with the individuals of other species if they reproduce 

 sexually. This statement, also, is a first approximation. Well- 

 known species, such as those familiar to fishermen, sportsmen, 

 gamekeepers, breeders and naturalists have individuality, in a 

 way. There is no doubt at all as to their " specific identity," so 

 that a fisherman, for instance, recognizes '* at sight " the species, 

 or kinds of animals with which he deals. Without doubt there 

 is very much confusion in zoological literature as to many of the 

 species made by naturalists upon the evidence of only one or a 

 few badly known specimens and so almost every species has a 

 " synonymy." But here we have obviously to do with imperfect 

 inductions and as knowledge increases the status of the naturalists' 

 species becomes ever more clear. 



As natural history becomes more perfect the geographical 

 distribution of the species becomes well known. It is then 

 apparent that a systematic, or " Linnaean " species can be decom- 

 posed into local races. Thus the Atlantic cod (Gadus callarias) 

 is a perfectly definite kind of fish never to be mistaken for any 

 other kind by a fisherman. Nevertheless, the fisherman knows 

 different sub-categories of cod distinguished by geographical 

 prefixes and the naturalists know that there are about half a 

 dozen races of cod distinguished (among other things) by 

 differences in the numbers of vertebrae in the backbone. These 

 races of cod cannot, in general, interbreed with each other and 

 there cannot be much inter-migration between the various sub- 

 regions : otherwise the morphological distinctions between the 

 local races would become obliterated. Generally let there be 

 local races of a systematic species, «, h, c, d and e, and let these 

 races inhabit contiguous subregions, a being near b, h being near 

 c and so on, but a being far removed from e. We expect to find 

 that a will be fertile with 5, h with c, c with d and d with e, and 

 we also expect to find that a will tend to infertility with e and, 

 in any case, will not have the opportunity of freely interbreeding 

 with e. But not many good investigations of this kind have 

 been made. 



