POLYGENBSIS 75 



what is true of a small area will hold equally well of a large 

 region, and the recurrence of the same habitat form may be 

 accepted as conclusive proof of polygenesis. The most 

 convincing evidences of multiple origin, however, are to bei 

 found in what DeVries has called "mutations". It makes 

 little difference whether we accept mutations in the exact 

 sense of this author, or regard them as forms characterised 

 by latent variability. The evidence is conclusive that the 

 same form may arise in nature or in cultivation, in Holland 

 or in America, not merely once, but several or many times. 

 In the presence of such confirmation, it is unnecessary to 

 accumulate proofs. Polygenesis throws a new light upon 

 many difficult problems of invasion and distribution, and, as 

 a working principle, admits of repeated tests in the fiield. 

 It obviates, moreover, the almost insuperable difficulties in 

 the way of explaining the 'distribution of many polygenetic 

 species on the basis of migration alone. 



In 1898, the author first advanced a tentative hypothesis 

 to the effect that a species homogeneous morphologically 

 may arise from two distinct though related species : during 

 subsequent years of formational study, the conviction has 

 grown in regard to the probability of siach a method of 

 origin. Since the appearance of Engler's work, a polyphy- 

 letic origin for certain genera has been very generally ac- 

 cepted by botanists, but all have ignored the fact that the 

 polyphylesis of genera carries with it the admission of such 

 origin for species, since the former are merely groups of the 

 latter. I cannot, however,agree with Engler, that polyphylet- 

 ic genera, and hence species also, are necessarily unnatural. 

 If the convergence of the lines of polyphylesis has been 

 great, resulting in essential morphological harmony, the 

 genus is a natural one, even though the ancestral phyla may 

 be recognizable. If, on the other hand, the convergence is 

 more or less imperfect, resulting in sub-groups of species 

 more nearly related within the groups than between them, 

 the genus can hardly be termed natural. This condition 

 may, however, prevail in a monophyletic genus with mani- 



