ARTIFICIAL AND NATURAL SELECTION 39 



be relied upon, if used for the continuance of a single strain, but 

 each single beet of high percentage should be regarded with doubt. 

 Direct experiments of Laurent, of Kuhn at Naarden, and others have 

 proved the validity of this conclusion, showing that the progeny of 

 extreme variants does not necessarily give always high averages for 

 the amount of sugar. 



Regression has as yet chiefly been studied by means of statistical 

 methods. It seems probable that it is largely due to the combination 

 of embryonic and partial variability . If the latter has no influence, or 

 only a very small influence, upon the embryonic latent qualities of 

 the new seeds, there must be regression, even if embryonic variability 

 itself should not be liable to it. 



Turning now to the processes in nature, we may assert that the 

 result of the struggle for life depends on the qualities of the indi- 

 viduals, but not on the causes of this quality. Any advance gained 

 by partial variability will be of equal value as the same advance 

 gained by embryonic fluctuation. Taking the latter as heritable, and 

 the first as not, or nearly not, it is easily seen that the relation between 

 the struggle for life and the hereditary qualities is by far not so inti- 

 mate as has hitherto been assumed. 



It may readily be granted that the fittest survive. But it may 

 also be granted that, in broad figures, half of the fittest have the 

 power to transmit their fitness to their offspring, whilst the other 

 half have not. And if, perchance, the same proportion should hold 

 good for the unfit, it would be of no avail for the next generation, 

 whether it springs from the fitter or from the less fit parents. 



Probably no breeder and no physiologist would take such an ex- 

 treme view. Some effect of the struggle for life in nature must as 

 well be granted as for artificial selection. My discussion had only 

 the aim of convincing you that there is much exaggeration in the 

 current conceptions concerning the effects of natural selection through 

 the struggle for life between individuals of the same species. My 

 chief object was to show that a clear distinction between embryonic 

 and partial variability points to a prominent part of the latter and 

 to a lesser chance of fluctuation at large playing a notable role in 

 the evolution of the whole animal and vegetable kingdom. 



If the visible characters of an extreme variant are no reliable 

 base for the judgment of its hereditary excellence, the question of 

 course arises, what marks have to be put in its place. 



Obviously this is a most vital question. It is equally important 

 from a practical point of view as with a view to the whole problem 

 of the part of fluctuating variability in organic evolution. 



It has been answered in one and the same way by practical breeders 

 and by purely scientific experiments. Hays in this country and von 

 Lochow in Germany have propounded the idea that the fact of 



