XXVI.] THE RATE OF VARIATION. 347 



Thus there ■will be less variation in the wild state than 

 in the domestic ; and such variations as occur will be much 

 less likely to be kept separate so as to produce modified 

 races. For, as we have seen, the improvement of domestic 

 breeds depends altogether on separation, and would be 

 impossible if the animals were permitted freely to obey 

 their instincts as to breeding. It is true that instinct 

 tends to keep wcll-estahlished breeds distinct, but this is 

 not the case when they are only beginning to diverge. 



To recapitulate the foregoing argument : 



Geological time is only about a million, or a few million. Summary, 

 times as long as the period needed to form what is really 

 a new species by the accumulation of small variations 

 under domestication. 



Variation goes on much less rapidly in the wild than in 

 the domestic state, and favourable variations, when they 

 occur, are much less likely to be so preserved as to 

 produce a modified race. We shall probably be greatly 

 within the mark if we assume that variation is ten times 

 less rapid in the wild state than in the domestic, and that 

 the chance of any favourable variation being so preserved 

 as permanently to modify the race is ten times less. If 

 these numbers are correct, then the efficiency of selection 

 among small spontaneous variations in producing new 

 races is one hundred times as great under domestication as 

 in the state of nature. If this is so, then five hundred 

 millions of years of variation and selection under nature 

 are equivalent to only five millions under domestication. 

 And if it takes five hundred years, or anything approach- 

 ing to that period, to form a race like the greyhound out of 

 a wild dog, it can scarcely be maintained that five millions 

 of years of change at the same rate would suffice to form 

 an elephant out of a fish. Five milKons are only ten 

 thousand times five hundred, and the proportion between 

 those two changes is certainly measurable by no such 

 number as ten thousand. 



For these reasons, I believe there must have been sudden 

 variations, amounting to the origin of new species all at 

 once ; not widely different in any case, perhaps, from the 



