MUTATION. 161 



The only explanation is, that in one or in both species, the 

 pollen and ovules are geno-typically distinct. Whatever the 

 nature of the difference, it must be such.as to persist through 

 several generations. For both kinds of hybrids are generally 

 stable for their characters, if self-fertilized. In further work with 

 these hybrids, de Vries discovered a very startling fact, which 

 up to a certain point gives an explanation of the peculiar mode 

 of inheritance. Namely, that in the Biennis-Muricata hybrids, 

 whose father was Muricata and whose mother was Biennis, all 

 the pollen had the properties of the pollen of Muricata, and all 

 the ovules had the properties of Biennis ovules. The correspon- 

 dent identity of the pollen of the hybrids with the pollen of the 

 father, and that of the ovules with the ovules of the mother was 

 observed in the reciprocal hybrid. This came to light when the 

 pollen of the hybrid was used in back-crosses with the parent 

 species. The pollen of the Muricata-Biennis hybrids, whose 

 father was Biennis, used in fertilizing Biennis flowers, gave 

 normal Biennis plants, just as the real Biennis pollen would 

 have done. The ovules of the same hybrid, fertilized with Muri- 

 cata pollen gave wholly normal Muricata off-spring. 



The result of this very curious phenomenon is that such hy- 

 brids breed true. In the Muricata-Biennis hybrids, the pollen 

 produced is Biennis pollen and the ovules are Muricata ovules. 

 Each self- fertilization therefore is a new hybridization, and 

 again produces the same Muricata-Biennis hybrids. We have 

 the following scheme: 



Initial cross A X B 



Hybrids A B 



Gametes: A ovules, B pollen 



Self-fertilization: A B 



Gametes A ovules, B pollen, and so on. One re- 



sult of this state of things is, that when the reciprocal hybrids. 

 A B and B A are crossed, one of the parent-forms results 

 A or B, according to which way the cross is performed. This is 

 certainly the most striking result, but the fundamental thing 

 in these experiments lies in the demonstration that in these 



11 



