110 REDUCTION OF VARIABILITY. 



as a working method in breeding-work with these plants. In 

 corn, for instance, the "centgener" or "ear to row" method of 

 selection has become quite popular, and in tobacco a similar 

 system is now generally followed. Even in sugar-beets, in which 

 self-sterility makes even enforced self-fertilization impossible, 

 and where a special, modified system of mass-selection has 

 been consistently followed, we have begun to realize the handr 

 icap under which we are working. In sugar-beets all the mod- 

 ern seed-growers now use a method of comparing the progeny 

 of a great number of individual plants. We begin to under- 

 stand how the very great variability, and the "degeneration" 

 of sugar-beet-seed is to a great extent avoidable. The total po- 

 tential variability of one individual beet, and ol the descend- 

 ants of this beet is only a fraction of the variability of the 

 whole group. And if we select six families, the progeny of six 

 individual beets, out of all the lots compared, we know, that 

 the genetic variability of these six groups is not identical m all. 

 By mixing the seed, and growing the mixture without select- 

 ion for two generations in order to obtain the required quantity 

 of commercial seed, we allow these six groups to inter-breed. 

 The variability of the resulting seed and therefore of the resul- 

 ting crop is several times greater, than it need have been. By 

 limiting our choice to one single group, the progeny of one 

 single plant, and rejecting second choices, we reduce the po- 

 tential variability, and thereojre the "degeneration" to a 

 fraction of the usual variability and degeneration. Here, again, 

 it is not the selection itself which makes for purity, we select 

 the individual with the genotype which most nearly ap- 

 proaches our ideal, and therefore the characterso f the average 

 individual of the selected group will be as we want them to 

 be. But purity for those characters is caused by isolation. 



If we divide any population of plants or animals into two 

 parts, even if these are equally great, it is nearly certain that 

 the potential variability of these two groups will be different. 

 Genes which are carried only by very few individuals will be 

 all included into one group, and if in the original group only 



