Tlic Origin of CJirysiUithcuiiim Scgetiuii Flciiuni. 163 



tained by saving seed from plants such as those figured 

 in Fig. 34 A and B, p. 184. 



These remarks also apply to my new Chrysanthcinuiii 

 scgctuin plenum. Many specimens set absolutely no seed 

 because the doubling has gone too far. For the same 

 reason others afford only a meagre harvest. Too drastic 

 a selection at the beginning of the flowering period would 

 destroy any prospect of a harvest and might even result 

 in the extinction of the variety. 



Moreover plants with a high degree of ''doubling" 

 produce no pollen for the fertilization of the others, 

 because they are almost exclusively female ; so that they 

 can take no part in the perpetuation of the race in this 

 way either. 



My novelty is probably the first horticultural variety 

 which has arisen in an experimental culture. By this 

 I mean that pure fertilization has been insured since the 

 beginning of the culture and that exact and detailed 

 records of the course of the experiment have been kept 

 every year. Moreover the selection of the seed-parents 

 has constantly been carried out from the verv beijinnini?-, 

 with a view to the same ideal. Selection began in 1897, 

 the "double" race was obtained in 1900. The selection 

 occupied, therefore, a period of three years. 



The corn marigold, being a composite, is admirablv 

 adapted to form the material for a statistical investi- 

 gation of its variability. The number of ray florets 

 fluctuate in accordance with the well-known law of Lun- 

 wiG based on Braun and Sciiimper's series. By this 

 means the exact composition of a culture can be ex- 

 pressed in figures and plotted graphically by recording 

 a sufficient number of inflorescences. The course of the 

 selective process can in this way be displayed in all its 



