302 THE GERM-PLASM 



number of ids. The 'loosening' depends on the repeated 

 removal of half of the ids, which occurs every time the germ- 

 cells are formed. Half the germ-plasm of the hybrid consists of 

 maternal ids of the species A, and the other half of paternal ids 

 of the species B ; in the formation of the germ-cells, this per- 

 fectly uniform composition becomes extremely diversified, owing 

 to the fact that the ' reducing division ' halves the germ-plasm 

 in different ways. If we suppose that the ids, or even idants, 

 are arranged in a circle, the plane of division will sometimes 

 cut the circle across one diameter, and sometimes across 

 another, and the combination of the germ-plasm in the germ- 

 cells containing ids of A and B will thus be very diversified. 

 If the hybrid is then fertilised by its own pollen, so that amphi- 

 mixis occurs between two of the differentlv constituted germ- 

 cells, a still greater diversity in the idic combinations will result, 

 and a high degree of variability in the offspring must inevitably 

 ensue. 



The offspring of hybrids are also very variable, even in those 

 cases in which they are produced by a recrossi?ig with one of the 

 ancestral species. ' Fairly dissimilar offspring generally result 

 from the fertilisation of a hybrid with the pollen of an ancestral 

 form ; intermediate forms between the hybrid and the ancestral 

 species in question are as a rule the most numerous ; while a 

 smaller number of examples resemble either the original hybrid 

 or the ancestral species.' * This statement agrees perfectly with 

 our theory, for a consideration of the reducing division renders 

 it evident that in recrossing with one of the ancestral species, 

 a very dissimilar numerical proportion of idants of the two 

 ancestral species must prevail in the fertilised egg-cell of the 

 following generation. Such hybrids are generally spoken of as 

 ' I hybrids,' because in them the force of transmission of one of 

 the ancestral species is assumed to be one-quarter only, and that 

 of the other three-quarters. This term is probably indispen- 

 sable in practice, but is obviously totally inaccurate. ' Three- 

 quarter hybrids ' do not by any means contain all the hereditary 

 substances of the two ancestral species in the proportion of three 

 to one ; the proportion is in fact a very variable one. A ' f 

 hybrid ' is produced, for instance, from two species of pink, 

 Dianthus chinejisis and D. barbatus, by fertilising the hybrid 



* Focke, p. 485. 



