1908.] IMPROVEMENT OF THE POTATO. 401 



ters 2.950; for the array of daughters due to a mother of given 

 character,, an average of 2.610. Even after considering that moth- 

 ers are a selection and not a race, the following conclusions are 

 drawn : 



,(i) "In asexual reproduction an individual does not produce a facsimile 

 of itself, and the variability of its offspring is not immensely reduced below the 

 variability of the race. 



(2) The asexual individual has offspring exhibiting regression, just like 

 the sexually reproductive individual. Its offspring tends to regress from the 

 individual to the race type. 



(3) With high probability but not definitely, the asexual individual repre- 

 sents the mid-parent (i), i. e. .466 and .619 are well within the probable errros 

 of the values .424 and .600, which we have found (Grammar of science p. 471) 

 for the correlation and regression of the mid-parent in the case of the bi- 

 parental inheritance." 



Summing up the whole case, he says : 



(1) "Whatever be the physiological function of the sex in evolution, it is 

 not the production of greater variability." 



(Note) This does not mean that sex is not a great factor in 

 producing a great variety of forms by combinations of existing 

 diverse characters, but these differences must exist in the parents 

 or previous ancestry of the individuals producing this combination. 



(2) "The variability of the individual makes itself felt not only in the bi- 

 parental reproduction but in autogamic and parthenogenetic reproduction, and 

 further in the undifferentiated like parts of the same individual. 



(3) Whatever amount of selection has taken place, there seems no possi- 

 bility of -reducing variability beyond some 10 percent or II percent." 



From the evidence given above, it seems reasonable for us to 

 make the following conclusions as a working basis for our own 

 problems. First, that there is fluctuating variation in asexually 

 propagated vegetation which when reduced to a basis comparable 

 to sexually propagated plants, is not to a great degree inferior. Sec- 

 ond, that decreasing variability in a potato variety need not be due 

 to a fixation of type which presupposes an inheritance of fluctua- 

 tio*ns. It may be explained by the decreased reaction to environ- 

 ment of an aging individual. Third, that, if partial fluctuations 

 obey the same laws as other fluctuations, types may be changed by 

 their selection ; and this selection reduces the variation of the char- 

 acter in the neighborhood of 10 percent. 



Our whole problem is reduced to the question whether types 

 may or may not be changed by the selection of fluctuations. If 

 types may be changed by the selection of individual fluctuations, 

 they may likewise be changed by the selection of partial fluctuations. 

 Until recently an affirmative answer to this theorem would not have 



