98 SECTIONAL ADDRESSES. 



closure, and in other ways ; but it remains unique, the only attempt made 

 with mammals to test this vital point. 



Schmidt has, however, conducted a similar experiment with the 

 viviparous blenny, Zoarces viviparus. This fish, which is a bottom-living 

 animal supposed not to migrate extensively, forms a series of local races 

 in the North Sea and the Danish waters. These are distinguished from 

 one another by statistical differences of the curves representing the varia- 

 tion in the number of vertebrae, of fin rays in the pectoral fin, and of 

 similar characters. These races appear to be stable. Their distribution in 

 some areas such as the Roskilde Fiord shows a gradation along a line 

 over which the salinity also changes, but the correlation so suggested 

 between this environmental condition and structure breaks down when 

 other regions are taken into account. 



There is evidence derived statistically from the nature of the mothers 

 that the variations are inherited, and an indication that, as in Peromyscus, 

 the differences are not obviously Mendelian. Schmidt carried out trans- 

 plantation experiments exactly parallel to those conducted by Sumner, 

 and found, just as he did, that no direct environmental efiect of the 

 kind required was produced during the few generations he could in- 

 vestigate. 



Thus here again we are faced with the fact that an apparent correlation 

 of structure with the surrounding physical and chemical conditions exists, 

 and that such evidence as there is does not confirm the view that this 

 correlation has arisen directly. There remains as the only other alter- 

 native the view that the apparent correlation is illusory. 



It may be accepted as a working hypothesis that the variable characters 

 which separate one geographical race from another are produced under the 

 influence of a number of genes, all independent, and all producing similar 

 efiects. As Prof. Karl Pearson pointed out in 1904, the effect of such 

 multiple factors will be to produce an apparent blending inheritance ; a 

 view now very generally accepted. It follows that, in certain cases at any 

 rate, if a small group of individuals phenotypically similar, though geno- 

 typically different, differing from the norme of a population, be isolated 

 and left to breed freely, they will, when considered as a population, tend 

 to vary still more from the original mode in the population from which they 

 sprang and that they will do so in the direction in which the original isolated 

 group differed. Prof. Pearson has reached the same conclusion from his 

 own very different standpoint and has evidence that the expected result 

 does actually occur. 



If then we can conceive of circumstances which will bring about such 

 isolation in such a way that the individuals so separated are determined 

 by an environmental condition, we shall have an explanation of the 

 divergence of local races which will account for the appearance in them 

 of individuals which lie outside the range of variation actually observed 

 in the small samples of the parent races which have been investigated. 



An explanation of this type accounts for some of the peculiarities which 

 Sumner has noticed in Peromyscus. For example, the existence side by 

 side of very light and very dark individuals in the same spot will present 

 no difficulties, and the fact that there is no or very little correlation between 

 such characters as colour, hind-foot length, tail length and width of tail 



