THE PRODUCTION OF DOMESTIC RACES. 19 



tion is due to unbalanced natural selection or to some other principle, 

 as, for example, the direct effects of use or disuse, or to indiscriminate 

 elimination. Just as indiscriminate isolation may produce unbal- 

 anced groups, and, therefore, segregation, so indiscriminate destruc- 

 tion in the isolated groups is liable to produce unbalanced propaga- 

 tion of diverse kinds, and so divergent transformation with intensive 

 segregation. If animosities arise between two sections of a tribe, 

 the domestic animals in the care of the two sections, though com- 

 pletely isolated, may present no apparent differences till famine or 

 some other calamity leads to the indiscriminate slaughter of all but a 

 pair or two of some species in one of those districts. This small 

 fragment will, in many cases, be unable to reproduce in all respects 

 the average character of the original race, and will become quite 

 perceptibly divergent. If heavy but indiscriminate elimination falls 

 upon the representatives of a given species in both sections of the 

 country, the divergence in the isolated groups will be likely to be 

 somewhat greater than if but one section suffers. This principle 

 differs from natural selection in that the exclusion is indiscriminate 

 instead of discriminate. Extreme elimination, leaving only a very 

 small remnant, is always unbalanced elimination as regards some of 

 the characters and, therefore, tends to produce transformation. 



Again, the effects of crossing between different strains and races of 

 the same species may occur in different degrees in the different dis- 

 tricts over which a species is distributed, and may, therefore, result 

 in divergences in different districts. This principle I have called 

 amalgamational transformation. 



4. The Stability 0} Races. 



For the preservation of a given race-type it is usually considered 

 necessary to exclude from propagation a certain proportion of the 

 variations that fall below the average which constitutes the type. The 

 degree of exclusive breeding that is needed to maintain the present 

 average depends upon the stability of the type, that is, the weakness 

 of the tendency to revert to forms possessed by more or less remote 

 ancestors. Whether a type may become so fully established as to 

 maintain a constant average without any tendency to reversion, is 

 perhaps an open question ; but in the case of the goose, which is one 

 of the most stable of domestic species, it is not certain that what has 

 'been called the birth average more closely resembles ancestral types 

 than does the average that conies to maturity and propagates the 

 species. The vitality of those that live and propagate is undoubtedly 

 higher than that of the whole generation; but this does not prove 



