140 ATKINSON— TWIN HYBRIDS. 



culture. Five or six different groups were recognized, but no 

 record was kept of the number of rosettes which could be assigned 

 to each group. The groups are as follows : 



1. Leaves broad, dark green, plain, white-veined. 



2. Leaves broad, dark green, crinkled, white-veined. 



3. Leaves narrow, dark green, red-veined. 



4. Leaves, broad, dark green, red-veined. 



5. Leaves medium broad, dark green, white-veined. 



These marked fluctuating variations in leaf form, representing 

 one of the features in the constitution of CEnothera franciscana 

 which is inherited in its hybrid progeny, marks this species as a 

 favorable one for stimulating great fluctuating variations in the 

 hybrids from crosses with other species, indicated not only by the 

 crosses of CEnothera franciscana^ and pycnocarpa described here 

 but also by the great fluctuating variations resulting from crosses 

 between Qinothera franciscana and CE. biennis as described by 

 Davis (1916). 



The marked fluctuating variations in the twin hybrids of the 

 reciprocal crosses between CE. franciscana and ffi. pycnocarpa might 

 be interpreted by some as indicating that two distinct hybrid types 

 were not present, but that the two forms represent merely a single, 

 wide range of fluctuating variation. That this interpretation is not 

 valid is shown by the fact that the franciscana twin type, although 

 variable, is fixed, it does not split in the second generation nor in its 

 fluctuations does it produce typical pycnocarpa twin forms ; while 

 the pycnocarpa twin type spHts in the second generation into the 

 two types. 



Further evidence that the interpretation given, in this paper, to 

 the results of these crosses, so far as the production of twin hybrids, 

 and one-sided splitting is concerned, is admissible, is found in the 

 very close genetic relation which CEnothera franciscana bears to 

 CEnothera hookeri (see Bartlett, 1914, p. 33). Reciprocal crosses 

 of CE. hookeri with CE. lamarckiana, or with certain of its mutants, 



3 The seed of CEnothera franciscana which I have used came from Dr. 

 H. H. Bartlett in the winter of 1914, from a series of cultures which he had 

 continued for a few years, and as I understand it has the same pedigree as 

 the seed employed by Dr. Davis in his interesting crosses with CE. biennis. 



