VARIATION 427 



or at any rate in some cases, would thus exhibit more tail- 

 feathers than either of the parents. But this would not by any 

 means be an ' increase in the force ' of a ' character,' but simply 

 an increase in the nmnber of new feathers iti the indi^ndual. 

 Let us suppose that the male bird possessed two extra feathers 

 which were situated at a^ and d'^ between the normal feathers a 

 and b and c and d, and that the mother had two supernumerary 

 feathers in the positions f^ and h^ : a majority of the deter- 

 minants for the extra feathers of the two parents might then be 

 united in the germ-plasm of one of the offspring, so that the 

 latter would possess all the four new feathers. This • increase ' 

 in the character of additional feathers in the tail, therefore, 

 depends upon the constitution of the germ-plasm, and can 

 consequently be transmitted to the next generation. 



This example clearly shows that all really nciu structures are 

 not merely the result of transmission, but are due to the I'ariation 

 and frequefit multiplication of the determinants. The mere 

 extension of a • character ' over larger regions or the whole of 

 the body, even if we choose to speak of it as an ' increase, ' may 

 be produced by pairing individuals which possess the desired 

 'character* in different parts. But an increase which is con- 

 nected with the formation of new stnictures, and consequently 

 with the multiplication of the determinants in the germ-plasm, 

 can never be produced by .such means alone. When this 

 results, the cause of the modification must be the variation of 

 the determinants themselves. 



Thus in the case of the fantail, a new feather can never be 

 produced by transmission alone : the offspring can merely 

 possess new combinations of such feathers as were present in 

 the parents. All really new structures can only originate in a 

 previous modification of the germ-plasm. 



Let us now take an example from among sexually dimorphic 

 forms, in which we are quite certain of the phyletic modification, 

 apart from the complication arising from the existence of 

 sexual double-determinants. The long tail-feathers of male 

 hummijig birds have arisen by a gradual lengthening of the 

 ordinary tail-feathers, such as are possessed by the females 

 at the present day. As already remarked, this lengthening is 

 the result of a considerable multiplication of the determinants 

 which give rise to the feather : the process of lengthening implies 

 that variations in the ids which possessed a larger number of 



