286 W. D. MATTHEW— PHYLOGENY AXD CORRELATION 



will consist of a complex of groups of varying blood-mixture, inextricably 

 tangled up, each individual being a product of indirect hybridism of 

 many ancestral strains, some of which would be generally regarded as 

 distinct species. 



The old Linnsean species concept was founded on supposed impossi- 

 bility of blood relationship between mutually infertile forms ; the current 

 concept makes the specific group dependent rather on local or preferen- 

 tial segregation than on infertility for their distinctness. But if the mi- 

 gration and dispersal of a race involves this sort of indirect hybridism, 

 even the broader concept of the Linnaean species will not secure complete 

 racial distinctness to the subphyla of a •polyphyletic group. The horse and 

 the ass are today immiscible, yet the wild horses, if we assume that they 

 are a later invasion, occupying territory formerly peopled by asses, prob- 

 ably have ass blood in their veins, and individuals should present varying 

 proportions of characters due to that inheritance. It is not indeed the 

 characters of the modern ass that should appear, but the characters of 

 (1) the ass stage in the center of dispersal of the Equidae, more remote 

 and more modified by evolutionary change, but reinforced by (2) admix- 

 ture of the remnants of the former equine population of the region in- 

 vaded by the true horses. 



The net results of this process of admixture will be to produce a re- 

 actionary effect on the characteristics of an invading group as it reaches 

 regions remote from its center of dispersal, through continual and pro- 

 gressive absorption of its predecessors in the dispersal of the race. This 

 will be added to the effect of the local environment and of earlier en- 

 vironments which it has passed through ; and all these factors dependent 

 on shifting habitat must be added to those which operate in a fixed habi- 

 tat to produce or control evolutionary progress in a race of animals evolv- 

 ing and spreading out from a center of dispersal. 



Whether the characteristics be inherited according to Mendelian laws, 

 or be fluctuating variations tending to revert to type, will not affect the 

 results of this continuous admixture and readmixture, so far as I can see. 

 Nor will it make any particular practical difference from the present 

 point of view whether Mendelian characters arise suddenly or gradually. 

 Under the stated circumstances, it will not be possible to eliminate a 

 Mendelian character any more easily than a non-Mendelian character. 

 The problem is far too complex and the data are too scanty to afford con- 

 clusive evidence as to such niceties. We can obtain and be sure of ap- 

 proximate results and these only. Thus the three strains which Ewart 

 has shown to exist in the domestic horse present a convenient approxi- 

 mate statement of the facts ; but the exact sources are probably vastly 



