VARIATION 443 



they become apparent, and in stating this view he also specially 

 referred to bud-variations. I fully agree with him in this respect, 

 and in the course of this book have repeatedly shown how 

 these gradual modifications (' Umstimmungen ') of the germ- 

 plasm, or of individual parts of it, are to a certain extent the 

 natural result of its assumed structure. In the case of bud-varia- 

 tions these invisible modifications may occur in a much earlier 

 generation than that in which they appear ; and hence it is easy 

 to understand why this form of variation mostly occurs in those 

 plants which, like the rose and Azalea, have already varied in 

 reproduction by seed. For modified determinants are more 

 readily accumulated by amphimixis ; and a germ-plasm which 

 has inherited such determinants from its ancestors may, after 

 these have been still further modified, give rise to a bud in which, 

 by a fortuitous differential nuclear division, the modified deter- 

 minants are in the majority, and can thus become effective. 



Unfortunately it has not been observed whether complicated 

 modifications, like that of the moss-rose, owe their origin to bud- 

 variation. This is cjuite possible theoretically, for the invisible 

 preliminary to variation — the modification of certain determi- 

 nants — is just as likely to affect a single determinant as a whole 

 group ; even the formation of two or more new determinants, 

 by the multiplication of a primary one, is quite as possible here 

 as in the case of the ordinary transformation of species. The 

 sudden appearance of such modified groups of determinants is 

 due therefore to fortuitous differential nuclear division. Such 

 cases prove that the preparation for the modification is a slow 

 process, for it appears impossible to conceive of any cause pro- 

 ducing a sudden variation of an entire group of determinants by 

 any method. 



The extent of any spasmodic variation will depend on the 

 extent to which the various groups of determinants have been 

 permanently exposed to abnormal nutrition. It would doubt- 

 less be as yet premature to bring their qualitative characteristics 

 into any causal relation with definite influences producing 

 variation. We can only state the necessity of assuming, a 

 priori^ that the extraordinarily complex germ-plasm is provided 

 with special means for the transmission of nutrient fiuid, the 

 increase or diminution of which must produce purely local 

 differences of nutrition ; and that, on the other hand, the vital 

 units must, owing to slight changes in their structure, modify 



