RECAPITULATION 63 



characters is so common that "we must conclude that a 

 tendency to this peculiar form of transmission is an integral 

 part of the general law of inheritance/' x The transmission 

 of ancestral characters in a latent condition through a long 

 series of generations is therefore of frequent and undoubted 

 occurrence. The only questions we have to consider are, 

 (a) Under what conditions do characters become latent ? 

 and (fr) Under what conditions do latent characters 

 reappear ? 



104. Morphological theories of inheritance usually explain 

 the latency and reappearance of ancestral characters by 

 supposing that discrete " ancestral units" pass in a dormant 

 condition through any number of ancestors till in the fulness 

 of time they become active in some descendant. Thus 

 Darwin speaks of "dormant gemmules derived . . . from 

 some remote progenitor." 2 Weismann's " ancestral ids " 

 struggle together. 3 Galton supposes that of its total heritage 

 the child derives a half from its two parents, a quarter from 

 its four grandparents, an eighth from its eight great-grand- 

 parents, and so forth. 4 In all these theories each of the 

 ancestors or many of them are supposed to be represented 

 by discrete active or dormant "units" which are present 

 during every stage of development. All these theories are 

 therefore, as regards reversion at least, in sharp opposition to 

 the theory of recapitulation, which supposes that the ances- 

 tors (i. e. mid-ancestors) are represented, not en masse, but in 

 orderly succession, beginning with the first and ending with the 

 last. If the one theory is true the other cannot be true. The 

 theory of recapitulation is so certainly true that the theory 

 of discrete ancestral units is quite certainly erroneous. 5 The 



1 Animals and Plants, vol. ii., p. 31. 2 Op. cit. ii., p. 395. 



3 The Germ-plasm, p. 260. 



4 Proceedings of the Royal Society, vol. Ixi., p. 402. 



6 When varieties are crossed there is sometimes exclusive inheritance 

 as regards some particular character in which they have become diver- 

 gent. But, as Mendel found, the suppressed ancestral trait tends to 

 reappear as an exclusive character in a definite proportion of grand- 

 children. Even the grandchildren who do not revert tend to have 

 offspring and descendants who do also in a definite proportion of cases. 

 The reversion to the second variety is however final. There is no 

 return to the first variety. The Mendelian law would appear to lend 

 colour at first sight to the hypothesis of discrete ancestral units, since 

 the two ancestral strains do not mingle. But ancestral strains which 

 are incompatible in some particular, and which ultimately separate as 

 regards that particular, are very ditierent from units representing 

 various ancestors belonging to the same strain, which keep together, 

 but remain discrete. (See 130, footnote.) When dealing with 



