LAMARCK1ANA SEMIGIGAS. 



563 



may be considered to offer a new proof for it. Moreover, in their 

 main features the groups are the same as those described for 0. 

 perennis. 



Table IV gives the number of chromosomes for the constituents 

 of these different groups. It is to be remarked that one of the 81 

 plants (one with 17 chromosomes) died in early youth, before develo- 

 ping its character. All the others reached the fall, mostly in flowering 

 condition, but partly as rosettes or young stems. 



In almost every group at least one of the chromosomes is doubled, 

 and the doubling of two or more such bodies, taken broadly, is 



Table IV 

 Mutants of O. semigigasXvelutina 



equally distributed among them. From this we may conclude that 

 the external features of the types as a rule are connected with the 

 doubling of one chromosome in each case, but that the doubling 

 of the others is of secondary importance. It produces, or is related 

 to, small deviations from the prototype, but does not essentially 

 change it. We may assume that in every case the factors of one chro- 

 mosome are dominant over those of the others, but why in one case 

 this chromosome should prevail, and others in other cases, we do 

 not know. This condition, however, obviously runs parallel with 

 the well known fact that the parent species produces in some instances 

 a lata, in others a scintillans or a cana, and this without the slightest 

 indication of the cause of such a choice. 



In every group two main causes of individual differences are to 

 be considered. One of them is connected with the doubling of more 



