The Or'ujin of Chrysanthciuuin Scgciiuii Plcmiui. 173 



vestigations. In order to bring this phenomenon into 

 bolder relief let us imagine that the isolation has already 

 taken place and the new race isolated. In other words 

 let us examine Fig. 31 (p. 169) and the data from which 

 this is derived (pp. 170 and 176). Let us first fix our 

 attention on the ordinate at 21. It contains only indi- 

 viduals of the 21 -rayed race. But in 1894 a single ex- 

 treme variant appeared, which, although it belonged to 

 the 13-rayed race, nevertheless had as many as 21 rays 

 (p. 170). If the cultures of 1893 and 1894 had been 

 more extensive the number of these extreme variants 

 would obviously have been greater. For the ordinates 

 20, 19, 18, etc., it is still more evident that individuals 

 of both races can occur. 



If we choose plants which have 21 or more ligulate 

 florets in their terminal inflorescences we cannot at all 

 be certain that they all belong to the race which is be- 

 ing sought. And if they are left to pollinate one an- 

 other, or if their seeds are mixed in the harvest, there 

 is small likelihood of the strain being pure. Amongst 

 the majority of pure seed-parents a number of individ- 

 uals of inferior value may exist and it is necessary to 

 remove these as soon as possible, at any rate before the 

 harvest. 



The possibility of doing this is afforded by the 

 later flowers. By means of them a curve can be de- 

 termined for each plant, and in this way values can 

 be obtained which are independent of the chances in- 

 separably connected with small numbers. The curves 

 describing the separate parts of one individual are called 

 its ^'part-curves." I have therefore plotted such curves of 

 all the individuals selected at the beginning of the flow- 

 ering period as having 21 and more ravs in their ter- 



