84 PHYLOGENY 



whether by artificial intraspecific selection it is possible to trans- 

 gress those limits. If this were possible, we could pass from one 

 species into another, and, by slow and gradual changes, convert 

 a constant type into another equally constant form. 



This question of the constancy is the chief side of the problem. 

 One point is the production of a new character, the other the loss of 

 the old one, which is assumed to have been changed into the new. 

 Such losses, however, are exceedingly difficult to obtain. One of the 

 most instructive examples is that of the lifetime of the sugar-beets. 

 These races consist of biennial plants, and the whole process of their 

 culture relies upon the heaping-up of sugar in the roots of the first 

 year, leaving the production of the stems to the second summer. 

 Now, this quality is far from being complete. Each year some annuals 

 are seen in the fields. And not as rare exceptions, as accidental 

 cases of atavism. Quite on the contrary, one per cent, or even more, 

 is the rule, and often they go up to ten or more per cent. Selection, 

 of course, is on this point always as absolute as may be; it has lasted 

 half a century for the sugar-beets, and many centuries for the forage- 

 crops. It has not been adequate to root the annuals out, and to 

 render the biennial character pure. 



So it is also with striped flowers and striped radishes, which 

 yearly produce some unicolored samples, notwithstanding continu- 

 ous and most severe selection. So it is ever in numerous other 

 cases; everywhere the intraspecific selection is capable of pro- 

 ducing ameliorated races, but incapable of making them as constant 

 as wild species use to be. 



Steady and regular advance of cultivated races no doubt occurs, 

 although it is not at all so general as is often assumed. But when- 

 ever it occurs, the advance is due to a corresponding continuous 

 improvement of the selecting methods, and not simply to repeated 

 selection after the same method. 



The truth of this assertion is most clearly seen in the case of the 

 beets. They are usually adduced as the best proofs of what can be 

 obtained by continuous selection. But the methods of judging the 

 beets are steadily being improved, and they have been so, even 

 since the time of Vilmorin. 



Vilmorin's own method was a very simple one. Polarization had 

 then not yet been discovered. He determined the specific weight of 

 his beets, either by weighing them as a whole, or by using a piece 

 cut from the base of the roots and deprived of its bark. Solutions 

 of salts were made in which the beets swam, and diluted until they 

 began to sink. In this way the heaviest beets could be selected, 

 and it was assumed that they were the richest in sugar, too. This 

 method has afterwards been improved in two ways. The first was 

 to make large quantities of the salt-solution, choosing a medium 



