GEOGRAPHICAL DISTRIBUTION 167 



of de Vries, we must of course recognize that, if a mutating 

 species were widely distributed, different individuals of the 

 species in widely separated localities and even with a dis- 

 continuous distribution, might throw the same mutants. 

 CEnothera Lamarckiana, for example, threw the same ele- 

 mentary species (mutants) in experimental pedigree 

 cultures in Holland and in various localities in the United 

 States. 1 Had 0. Lamarckiana (contrary to fact) been 

 widely distributed in nature, such mutants as O. gigas, 

 0. scintillans, O. Icevifolia, and others would possibly (or 

 even probably) have appeared in different and widely sepa- 

 rated stations, and these elementary species might con- 

 ceivably (and not improbably) have become established as 

 true species of the systematist. When, therefore, we 

 find a given species (or a larger group) in widely separated 

 localities, but not in the intervening regions, we must 

 (barring the phenomenon of mutation referred to above) 

 conclude, either that it has been able to migrate across 

 barriers where it could not become established (as when 

 seeds of land plants are carried by ocean currents across 

 barriers of salt water), or else it has formerly had a con- 

 tinuous distribution, but has subsequently died out in 

 regions between its present localities; in the latter case it is 

 referred to as a relict endemic. When these localities are 

 distant hundreds or, as is often the case, thousands of miles 

 from each other, one can readily understand that species 

 having such discontinuity of distribution must, other 

 things being equal, be older than species having continuity 

 of distribution; they must have existed long enough for the 

 changes above mentioned to have taken place. 



This principle is confirmed by the evidence of fossils. 

 A striking case is that (cited by Chodat) of Zelkowa, 



'See pp. 114-117. 



