438 THE GERM-PLASM 



abnormality became apparent as a variation in the flower or in 

 the leaf. 



The reason why 1 lay special stress on these cases is, that 

 natural selection plays no part in them, for we are here dealing 

 with artificial, and not with the natural conditions of life. The 

 imperceptible variations of the germ-plasm naturally did not 

 increase continuously, for the processes of ' reducing division ' 

 and amphimixis exerted their influence on every fresh generation, 

 and helped to bring about either a marked decrease of the modi- 

 fied determinants, or their sudden increase — a doubling, or a 

 still greater augmentation. In this way it is easy to understand 

 all the particulars connected with the occurrence of the abnormal 

 flowers and leaves. As a rule, the number of these abnormali- 

 ties increased in the course of generations with a fairly con- 

 stant regularity. The following results, for example, were 

 obtained from four generations of Nigella da/nascena, w^hen sown 

 close together : — 



1883. No double flowers. 



1884. " 



1885. Six double and twenty-three typical flowers, or 24 per 

 cent. 



1886. One double and ten typical flowers, or 10 per cent. 

 The number of double flowers was, however, not always 



constant, and in some cases they again disappeared entirely. 

 Papaver alpinmn, for instance, which Hoffmann had cultivated 

 unintermptedly since 1862, displayed "a trace of variability in 

 the form of the leaves, and a more decided variation in the 

 colour of the flowers in 1882.' These experiments were con- 

 tinued from 1882 to 1886, and yielded the following proportion of 

 double and normal flowers : — 



In 1 88 1 the proportion was 40-0 per cent. 



The complete c'isappearance of abnormal flowers in the two 

 last of these years seems inexplicable at first sight ; but it is 

 easily accounted for if we bear in mind the fact that natural, 

 and not artificial fertilisation occurred, — i.e., the plants were 

 crossed at random, — and that every reducing division and subse- 



