Permanence and Evolution. 85 



of descending to the ultimate unit of classifica- 

 tion the Family. 



But if with Agassiz we start from animal 

 " nations," it is easy to see how by (in any 

 of these nations) selecting certain families 

 and eliminating all impurities in them and 

 exterminating the rest, we change the whole 

 aspect of the nation, and this may be called, 

 rather loosely, modifying the breed, without 

 there being, in a scientific sense, any real 

 modification in the matter. For instance, we 

 know by the experiments of Hofacker, " Ueber 

 die Eigenschaften," etc., and others, the nearly 

 absolute permanence of colour in the horse, but 

 most breeds contain permanent strains of more 

 than one colour ; and if, for instance, the pre- 

 dominant white type of Normandy horse were 

 discouraged and roan strains encouraged, the 

 colour of the breed might, in common parlance, 

 be said to be modified. In the breed for 

 instance of Hereford cattle, we can still detect 



