MUTATIONS, VARIATIONS, AND RELATIONSHIPS OF THE OENOTHERAS. 83 



or whether he is engaged in tracing the character, frequency, and causes of 

 mutation a preliminary statistical investigation will aid in the choice of mate- 

 rial best suited to his purpose. It is obvious that the great variability of any 

 elementary species would disfavor it as research material for the study of 

 mutations rather than indicate its desirability, as the wide range of its fluctua- 

 tions would suggest the likelihood of its being a recent mutant rather than a 

 mutating species. A more promising procedure with respect to any variable 

 form would seem to be to use the most nearly related or contiguous species 

 having a low degree of variability in the hope of thus hitting upon the parental 

 type from which the variable form originated. The student of pedigree- 

 cultures frequently receives letters from the kindly disposed, calling his atten- 

 tion to strikingly variable forms which are, on the ground of their variability, 

 conceived to be especially valuable. It is to be hoped that in the future more 

 attention will be given to constant forms having one or more variable near 

 relatives. 



Whether the relative variability of old and new structures could likewise be 

 used by the morphologist as a new criterion for working out lines of descent it 

 would be somewhat reckless to say at present, but an important field for 

 research is indicated. Are forms which are obviously more recent at the same 

 time more variable than the older? Are the structures characteristic of the 

 Orchidacese, for instance, more variable than those of more generalized mono- 

 cotyledons, and those of the Compositge more variable than those of other 

 dicotyledons? Are monocots on the whole less variable than the clicots? 

 While not wishing to press the application of the principle too far, it may be 

 pointed out that one feature of the investigation here reported seems to bear 

 directly upon the phylogenetic significance of the greater variability of new 

 characters. The characteristic feature of the Ocnothera bud is the hypan- 

 thium. It doubtless represents one of the most fundamental mutations in the 

 history of the genus. It accords well with the hypothesis here offered to find 

 that the length of the hypanthium is in every species the most variable part 

 of the bud. The great excess in its variability over that of other parts of the 

 buds is made strikingly apparent by comparing the average variability of the 

 several parts of the buds in all the species studied. The average coefficients of 

 variability are as follows, in order of their magnitude : Length of hypanthium, 

 14.97; length of cone, 9.56; length of ovary, 8.66; thickness of cone, 7.85; 

 thickness of ovary, 6.30 ; thickness of the hypanthium, 5.83. The consistency 

 of these results make it certain that the great variability of the hypanthium 

 originated with the origin of the hypanthium. 



The fact that a new species is more variable than an old one should also 

 give to the phytogeographer a method of investigating the exact status of an 

 endemic species. If it presents a case of initial endemism it should be more 

 variable than the nearly related, wide-ranging, and hence presumably older 



