20 



DR. E. E. GATES— CONTEIBUTION TO A 



HOW at hand from numerous collections of wild seeds from all over the continent, as well 

 as from other facts, it seems that though the three species O. grandijlora, 0. biennis, 

 and O. muricata very largely overlap in their distribution, yet 0. grandiflora is on the 

 whole the most southern and O. muricata the most northern. I have collected identical 

 races of O. muricata from Nova Scotia, New Brunswick, and Manitoha, so that the 

 decrease in flower-size in the more northern latitudes may not be devoid of significance. 



IV. Variations and Mutations of O. Lamaeckiana^ %^ii. 



Since the important researches of DeVries, covering many years' experiments, in 



which he showed that a group of divergent types arise from 0. LamarcJciana when it 



is cultivated in large nnmhers, the need of a more detailed analysis of this hehaviour has 



become evident. The idea of recurring "mutation periods" as accounting for this 



behaviour has not found universal favour with biologists, and there is no detailed 



evidence in its support. It is probable that the premutation hypothesis of DeVries will 



also be found untenable- Indeed, the author's own views on some of these subjects 



have probably changed since the publication of ' Die Mutationstheorie ' in 1901. But a 



satisfactory hypothesis as to why the aberrant types appear has not been given. From 



the facts presented in previous papers and briefly summarized in the last section, it will 



appear that the mutation phenomena in O. Lamarckiana are closely connected with the 



crossing which has taken place in its ancestry, but, so far from depriving these 



phenomena of significance, they show rather that the species-unit in open-pollinated 



plants is much more complex than was previously realized. 0. LavnarcMana, 



O. grandiflora, and 0. Sooheri are all undoubtedly " hybrids," if by that is meant forms 



whose ancestry has undergone crossing, and has therefore been contributed to by 



several rather closely related races. But in this connection several things must be kept 



in mind : (1) That the mutants which appear in each generation probably are not 



the same as the races which contributed to the ancestry of the species ; (2) that 



0. Lamarckiana itself shows much more variation in different cultures (races) than has 



previously been supposed ; (3) that (as I shall show) the mutants themselves frequently 



exhibit a wide range of variation, in some cases much wider than would be supposed 



from the previous accounts of this miitation process. 



In the following account I shall endeavour to exhibit the range of fluctuating 

 variation of 0. Lamarckiana itself and of its various mutants. It is important to 

 remember, on the other hand, that, although some of the races frequently overlap in 

 their variability (so that an apparently continuous series can be formed from them), yet 

 they are really discontinuous in origin, as evidenced by their bebaviour when inbred or 

 crossed. Another important point, which has been neglected in previous studies of these 

 forms, is the series of ontogenetic stages passed through by the rosettes in their 

 development. Successive leaf types appear, which often differ from each other to 

 a striking degree, so that different stages of the same rosette may be much more unlike 

 than corresponding stages of rosettes belonging to different races. In the study of 



