THE PHENOMENA OF REVERSION 307 



place with the pollen of the hybrid ; and even by continued 

 recrossing with one of the ancestral species, the combination of 

 the germ-plasms of the first generation would gradually become 

 weakened, and perfectly pure germ-plasm of the one ancestral 

 species could never be produced. But as soon as we assume 

 that the germ-plasm consists of units which remain separate in 

 the form of idants or ids, it is evident that the whole of the units 

 of one species may be removed from the germ-plasm of the 

 hybrid-offspring, either with or without the occurrence of re- 

 crossing ; and even if recrossing does take place, reversion must 

 occur sooner or latfer in individual descendants. 



Our theory also enables us to make certain predictions, which, 

 as far as my knowledge extends, have not up to the present been 

 confirmed by facts. If sometimes, though rarely, all the idants 

 of A pass into one germ-cell, and all those of B into the other, 

 in the process of halving of the germ-plasm of the primary 

 hybrid A x B a reversion to both the ancestral forms will occur 

 when the second generation is fertilised with the pollen of the 

 hybrid, as has just been shown : that is to say, such a reversion 

 will take place when a pollen-grain containing idants of A only 

 comes in contact with an egg-cell which also contains idants of 

 A only, or when idants of B meet with others of similar origin. 

 The same must also be possible in the hybrids of the third gen- 

 eration, even when all the complete reversions of the second 

 generation are left out of consideration, simply for the reason 

 that certain individuals of the second generation are produced 

 from the combination of ;/A with ;/B idants, and are therefore 

 exactly intermediate forms like all the hybrids of the first gen- 

 eration. The occurrence, even if only in individual cases, of 

 such intermediate forms, renders it possible that germ-cells may 

 again arise which contain idants of A or B only, so that ultimately 

 reversion to one or other of the two ancestral forms becomes 

 possible. These reversions will certainly be of rare occurrence; 

 and as it is entirely a matter of chance that germ-cells which 

 contain the most infrequent combinations of idioplasm should 

 come together in the process of fertilisation, it would be neces- 

 sary to examine a very large number of seedlings before such 

 cases would be found. 



Before passing on to the consideration of reversion to mdi- 

 vidual characters, it should be remarked that the same law of 

 reversion which can be recognised in the case of plant-hybrids, 



