36 MUTANTS AND HYBRIDS OF THE OENOTHERAS. 



STATISTICAL COMPARISONS OF ONAGRA (OENOTHERA) LAMARCK- 

 IANA WITH TWO OF ITS MUTANTS.* 



So general is the experience in garden practice that a variety 

 which has been improved by selection rapidly loses its improved 

 character upon the cessation of the selective process that the sta- 

 bility of any modification which is discovered either in nature or 

 under cultivation demands the fullest possible proof. The mutation- 

 theory is so diametrically opposed to Galton's law of ancestral 

 heredity that it needs especial investigation from the same standpoint 

 and by the same methods by which this law was established. Accord- 

 ing to Galton's law the offspring shows a certain definite degree of 

 inheritance from each generation of its ancestors, one-half from its 

 parents, one-quarter from its grandparents, one-eighth from its great- 

 grandparents, and so on. (Galton, 1889.) As a consequence of this 

 law the children of extreme parents are on the average less extreme 

 than their parents, because their preparental ancestry is on the average 

 more mediocre. The departure of the offspring from the mean con- 

 dition of the race to which it belongs toward the extreme condition of 

 its parents has been designated ' ' regression . ' ' One of the most serious 

 criticisms which has been made upon De Vries's conclusions has been 

 that of Weldon , who points out that no satisfactory evidence has been 

 presented to prove the completeness of regression, in the Galtonian 

 sense, in the Onagra mutants. For, unless such regression is com- 

 plete, these mutants could not maintain themselves distinct from the 

 parental type except through the agency of man in guarding pollina- 

 tion and in selection, a fact which would deprive them of all signifi- 

 cance in the explanation of evolution. (Weldon, 1902.) 



To test quantitatively the continuity or discontinuity of a few of 

 the differential characters of the Onagra mutants, and to begin the 

 work which, when continued for several years, will forever set at rest 

 the question of the completeness of Galtonian regression, the investi- 

 gation the results of which are reported in this section were undertaken . 

 The number of specimens available for study was not sufficient for the 

 most satisfactory statistical work, but the results offer a number of 

 suggestive lines for future investigation. 



As all the characters chosen for this investigation are notably 

 affected by the physical conditions to which the plants are subjected, 

 it should be pointed out that all these specimens were grown near 

 each other in an experimental garden which presents nearly uniform 

 conditions throughout, and that they were planted at the uniform 



*Prepared by G. H. Shull. 



