82 MUTATIONS, VARIATIONS, AND REL,ATlONSIIIPS uK THIi OIJNOTHERfVS. 



generation which conformed to the biennis type. The same result is true of 

 the reciprocal cross. It remains to be seen whether the mutant characters are 

 merely recessive in this combination, or whether a condition of latency has 

 been assumed in which the mutant api)ears only in its customary frequency. 



The evening-primroses, then, afford the following exact evidence as to the 

 effects of intercrossing: I\Iany of the mutants when crossed with the parental 

 form give progeny composed of the parental form and the mutant in the first 

 generation. In the case of gigas and rubrinervis the proportion of the mutant 

 in the progeny is generally much higher than the parental type. Rubrinervis 

 when grown in close contiguity to the parental form showed only a small pro- 

 portion of crosses, and the newly discovered mutant of 0. biennis also gave 

 but a small proportion of crosses when exposed to danger of hybridization 

 with the parent. Lastly it is to be pointed out that O. brcvistylis, which pro- 

 duces but few seeds, and which is recessive w^hen crossed with 0./a»?arc^tana, 

 has been in continuous existence for 20 years, perhaps much longer, at Hil- 

 versum, where it is in direct competition and danger of being hybridized with 

 the parental form. Experiences in the guarded cultures seem therefore to be 

 confirmed by an analysis of natural conditions. Swamping of new forms, by 

 intercrossing, is a specter which looms vast and shadowy across the visions of 

 writers of a speculative habit, but at the present moment it is believed that no 

 evidence carefully tested by modern methods is available to justify the hypoth- 

 esis, so far as plants are concerned. 



The demonstration of the greater variability of the mutants as compared 

 with the parent form seems to be complete, and this relation is one of consid- 

 erable practical and theoretical importance. The greater variability of phylo- 

 genetically new characters as compared with older ones is not wholly new and 

 unique, though we believe our studies in 1904 gave the first statistical support 

 to the thesis when stated in this form. Darwin's (1859, Ch. II) view that 

 characters which distinguish species are more variable than those that sepa- 

 rate genera, and that varietal are more variable than specific characteristics, 

 gives the first recognition of this principle, if we take into account that accord- 

 ing to his conception of the relation of these several form-groups the variety is 

 an incipient species and the species an incipient genus. Field (1898) worked 

 out the relation between a varietal and a specific characteristic in the wings 

 of a lepidopterous insect, giving a confirmation of Darwin's rule, but he did 

 not relate his facts in any way to the relative age of the characters studied. 

 It is probable of course that the varietal characteristic was a more recent 

 acquisition than the specific, and this would make his results accord well with 

 the observations of the authors on Oenothera. 



To the student of evolution who wishes to institute pedigree-cultures, the 

 relation here demonstrated should prove of much practical value, for it matters 

 not whether he desires to study the possibility of fixing extreme fluctuations, 



