Breeding experiments which show that hybridisation and mutation etc. 2H 



of the cross. (3) The changes in nuclear structure which occur in some 

 of these mutants. (4) The appearance of parallel series of mutations, 

 analogous to but not entirely agreeing with the de Vriesian forms, in 

 other races and species of Oenothera. These parallel mutations have 

 been independently shown to exist in a Swedish race of 0. Lamarckiana 

 by Heribert-Nilsson (1912), and in different races of 0. biennis by 

 Stomps (1912) and myself (1912 a). 



The Mendelian hypothesis totally fails to explain these four cate- 

 gories of facts, and the cytological behaviour in particular furnishes 

 the proof that the mutation process is directly contrary to any process 

 of Mendelian splitting. Yet we have been asked by many Mendelians 

 to suppose that all new characters find their origin merely through 

 the loss and recombination of unit characters. The fact that mutations 

 occur in pure lines of self-pollinated plants, as Johannsen (1908) has 

 shown, for example, in beans; and that similar phenomena involving 

 a more or less sudden germinal change occur even in bacteria (for the 

 literature see Dobell, 1913) in which sexuality is absent, in itself 

 shows that germinal changes do occur which are entirely different in 

 kind from the Mendelian shuffling of characters. But I shall show 

 further that, even in forms that have been cross-bred, some of the 

 mutations at any rate owe their origin to a cause which is independent 

 of the mere mingling of characters in hybrids. We must, therefore, 

 recognize that the combination and sorting of characters which already 

 exist is one problem, and the origin of those characters, whether it be 

 sudden or gradual, is another. These two problems must be kept in 

 separate categories, for Mendelians misinterpret the facts and con- 

 fuse the issue when they assert that the latter are in any way included 

 in the former. Any clear view of the facts now known must lead to 

 the conclusion that the origin of new characters is one question, and 

 their hereditary behaviour after they have originated, is another. 



The assertion has not infrequently been made, that all germinal 

 changes are due to the loss of characters, an assertion leading to the 

 grotesque view that all evolution, "from Amoeba to man", has taken 

 place through the successive "loss of unit-characters". But this view 

 seems now to be largely given up, and the various types of sudden 

 germinal change which are now known so obviously contradict such 

 a position that it is not necessary to labour the point here. 



Another matter on which this paper furnishes some definite evi- 

 dence is with regard to the nature of a unit-character. Are unit cha- 

 racters the indestructible, unmodifiable and undecomposable entities 



14* 



