No. 517] A BIMODAL VARIATION POLYGON 29 



In the 1905 series all the lamina? found in the inflores- 

 cence were included in the countings. In the discussions 

 just given I have treated only the primary leaflets. The 

 data for the other lamina 1 are included in the notes on the 

 tables of data or in the notes on teratological cases. 

 Their addition would make little difference in our distri- 

 butions, but the data are available for any one who cares 



Further analysis of the data for leaf character- is not 

 justified by the quantity of material. Two conclusions 

 are seen to be amply justified by the facts: (a) In Syn- 

 desnion there is a strong- tendency to the production of 

 simple and trifoliate leaves, (b) The preceding fact, 

 taken in connect ion with the peculiarities of the inflores- 

 cence with regard to the number of leaves produced, is 

 quite sufficient to account for the bimodal condition of the 

 variation polygon for number of laminae per inflorescence. 



The bimodal polygon is therefore explicable on purely 

 morphological grounds without any assumption of the 

 mixture of two or more "races" or "minor species," pro- 

 vided the phmts producing two, three and four leaves 

 per inflorescence be not considered "small species" or 

 "biotypes." Personally I see no reason whatever to 

 think that they are, but in these days of minute segrega- 

 tion the possibility must not be left unmentioned. To me 

 it seems not unlikely that the three different classes of 



