524 



SCIENCE. 



[N. S. Vol. XXI. No. 5.36. 



.short-haired three-toed. From this array 

 of forms the breeder may now select the 

 particular combination of characters which 

 suits his purpose. 



Can we doubt that in nature a similar 

 choice is ot¥ered between every mi;tation 

 and its opposite, combined or uneombined 

 with every other mutation then present in 

 the race? 



It is true that cross-breeding may affect 

 to a greater or less extent the nature of the 

 characters involved in a cross, but this 

 sometimes facilitates the creation of de- 

 sirable breeds, for it serves to induce new 

 mutation, which in some cases is progres- 

 sive, in others regressive. For example, in 

 guinea-pigs, a cross between a coal-black 

 animal and an albino may restore in the 

 young the ancestral, or 'agouti,' coat con- 

 .sisting of black hairs ticked with reddish 

 j^ellow, or in other cases may result in the 

 production of a black-white spotted animal. 

 By selection either of these conditions may 

 be perpetuated in a distinct breed. The 

 one is a regressive or reversionary change, 

 the other progressive in that it leads to the 

 production of a new type of pigmented 

 coat. 



On the whole, it appears that the forma- 

 tion of new breeds begins with the discov- 

 ery of an exceptional individual, or with 

 the production of such an individual by 

 means of cross-breeding. Such exceptional 

 individuals are mutations. 



An examination of stock registers points 

 in the same direction. The beginnings of 

 new breeds are small. Pedigrees lead back 

 to a few remarkable individuals or to a 

 .single one, as in the Ancon sheep. But 

 given the exceptional individual, and a new 

 breed is as good as formed. The few gen- 

 erations which the breeder usually employs 

 in 'fixing' or establishing the breed and 

 during which he practises close breeding 

 serve principally to free the stock from 



undesirable alternative characters, not to 

 modify the characters retained. 



Modification of characters by selection, 

 when sharply alternative cenditions {i. e., 

 mutations) are not present in the stock, is 

 an exceedingly difficult and slow process, 

 and its results of questionable permanency. 

 Even in so-called 'improved' breeds, which 

 are supposed to have been produced by 

 this process, it is more probable that the 

 result obtained represents the summation 

 of a series of mutations rather than of a 

 series of ordinary fluctuating variations. 

 For mutations are permanent; variations 

 transitory. A moment's reflection will in- 

 dicate the probable reason. Variations 

 which are distributed symmetrically about 

 a modal condition, so as to produce when 

 graphically expressed a frequency of error 

 curve, represent the result of a number of 

 causes acting independently of each other. 

 These causes are principally external, con- 

 sisting in varying conditions of food-sup- 

 ply, temperature, density, moisture, light, 

 etc. These conditions alter from genera- 

 tion to generation, and so do effects de- 

 pendent upon them. Mutations, on the 

 other hand, have an internal origin, in the 

 hereditary substance itself. They are rela- 

 tively independent of the environment, be- 

 ing affected only by such causes as affect 

 the nature of the hereditary substance 

 itself, one of which apparently is cros.s- 

 breeding. 



There are, however, frequently found 

 in breeds of domesticated animals condi- 

 tions which are not sharply alternative in 

 heredity to the corresponding characters of 

 other breeds. It is an open question 

 whether such conditions could be main- 

 tained if cross-breeding were freely allowed 

 with animals of a different character. If 

 not, they could scarcely become racial char- 

 acters, under the action of natural selec- 

 tion. The race would then become, not 

 sharply dimorphic or polymorphic, as is 



