— 398 — 



the facts are so extraordinary, that I do not believe that the com- 

 mon rules are sufficient. 



For following reasons I maintain them to be two species: 



1. The chromosome numbers are very distinct: 13 and 17. 



2. Two well defined types: grand, lab, viol and parv, n.lab, 

 albid exist and are far more frequent than the transition 

 types. 



3. The two species, when crossed, cannot give segregation in 

 Mendelian proportions owing to the difference in chromosome 

 number and the irregular distribution of chromosomes in F^. 



VII. What is typical V. tricolor and typical V. arvensis? 



I know no better distinction between the two species than 

 Haller's above quoted diagnoses: grandiflora = tricolor and 

 parviflora = arvensis. But the dividing line between grandiflora 

 and parviflora is not quite the same as Haller's. According to 

 general use I determinate parviflora {arvensis in widest sense) 

 as all piants with the upper petals as long as or shorter 

 than the upper sepal (including both the homozygotic parvi- 

 flora and the intermediate heterozygotic parviflora) and grandi- 

 flora {tricolor in widest sense) as all piants with the upper 

 petals longer than the upper sepal. 



But when the question is what is typical arvensis and tricolor 

 i. e. the probable combination of the original species, 

 that crossed have produced the great multitude of types, we 

 have no other guide than the frequency of the combinations in 

 nature. 



Looking at table I, pag. 378 we will find, when only the three 

 first characters are considered, that one combination of parviflora 

 is far more frequent than the three other combinations. This 

 combination is parv, albid, pinn and must be considered as the 

 typical arvensis. The other combinations are of unfrequent occur- 

 rence: parv, albid, palm and especially the two combinations parv, 

 viol {n. albid) (only 46 individuals among 758). 



Among the grandiflora individuals two of the four combin- 

 ations are far more frequent than the remaining two. The com- 

 bination grand, albid is very rare (only 12 individuals among 

 1142). The remaining 1130 individuals are all grand, viol {n. albid) 

 but 449 have pinnate and 681 have palmate stipules. It is 



