5^8 THE AMERICAN NATURALIST [Vol. L 



stable, species. We could hardly find better proof that 

 such mutations in (Enothera involve the a. gametes, and 

 are apparent in the zygotes without the need of subse- 

 quent segregation because the factors involved have no 

 counterparts in the (3 gametes. 



The same crosses, however, afford evidence that certain 

 characters are carried by both a and £ gametes, and may 

 therefore prove to show Mendelian segregation. The buds 

 of CE. numismatica have a short viscid pubescence which 

 is lacking in CE. pratincola. The matroclinic hybrid (E. 

 pratincola • X (E. numismatica can be distinguished from 

 the pistillate parent only by the presence of this hair-type, 

 inherited from the pollen parent. When the second hybrid 

 generation is grown, segregation with regard to this char- 

 acter takes place, and part of the progeny can not possibly 

 be distinguished from CE. pratincola. 



In these results we have a clue to the segregation shown 

 in certain hybrids, and the lack of it in others. Most of de 

 Vries's hybrids have involved CE. Lamar ckiana, a species, 

 according to my interpretation, with very dissimilar a. and 

 /3 gametes. He has therefore obtained and described 

 many measurably constant hybrids. Davis, however, 

 studying CE. grandiflora, which may conceivably have but 

 one type of gametes, has found segregation the rule 

 rather than the exception. In his later studies, involv- 

 ing CE. franciscana and CE. biennis, he has obtained twin 

 hybrids within each of which there was a considerable 

 degree of segregation. All of these varying results will 

 eventually become coordinated as we become more used to 

 distinguishing between non-Mendelian and Mendelian 

 characters. 



Another point which must be mentioned is the fre- 

 quence with which the various types of mutations give 

 rise to one another. For example, two mutations of CE. 

 pratincola, mut. nitida and mut. fallax, each give rise to 

 plants of mut. numularia, which are as typical as though 

 they had been derived directly from CE. pratincola. As 

 already brought out, some mutations appear to be re- 

 versible in that they revert to the parent species in part 



