geographic distribution. Indeed, if the differentiation of 
cotton species is to be explained in terms of man’s move- 
ments, then there are other genera which are not culti- 
vated, in which speciation ought likewise to be so ex- 
plained ; a procedure which would soon reduce the thesis 
to an absurdity. 
Secondly, their classification of Gossypium, based upon 
the assumption of a recent origin of the New World tet- 
raploids, is not in harmony with some of the sound tax- 
onomic conclusions of earlier students. For example, the 
endemic cotton of the Galapagos Islands, formerly re- 
garded as a good species, G. Darwinii, is now treated as 
a variety of the mainland cotton G. barbadense. 
Finally, the endemic wild tetraploid cotton of Hawaii, 
G. tomentosum, presumably derived from the American 
tetraploid, presents an almost insuperable difficulty to 
the entire hypothesis. How could the Hawaiian cotton, 
in a few thousand years or less, have become so differ- 
entiated from the mainland allotetraploids that it is now 
generally regarded as a distinct species, since it differs in 
many characteristics, and since there is a high incidence 
of seedling mortality in the F, when G. tomentosum is 
crossed with the American species G. hirsutum. The 
genetic gap between the Hawaiian tetraploid and the 
American tetraploids is perhaps a fourth to a half as 
great as the gap between the American and Asiatic dip- 
loids, yet the differentiation in the one case is supposed 
to have required only a few thousand years, in the other, 
since it is assumed to have begun in the Cretaceous, some 
120 million years. Differentiation of species does not, 
of course, proceed uniformly in time and space and the 
degree of differentiation is not a reliable measure of time. 
Yet it is difficult to believe that the rate of speciation 
within the same genus, and involving in part the same 
chromosomes, could have been roughly ten thousand 
[ 287 ] 
