INVESTIGATIONS ON THE PHYLOGENY OF THE ANGIOSPERMS 1 7 
from 85 per cent to 95 per cent (averaging 90 per cent or more) of 
the arborescent species have three or more gaps at the node. This 
type is also dominant in the temperate non-arid regions of the Southern 
Hemisphere, such as Chili, Patagonia and New Zealand; and, to a 
less extent, in upland areas in or near the tropics, as Nicaragua, 
Simla and the Bombay uplands. In the tropics, however, there is a 
strong contrast for here 50 per cent or more of the tree species are 
unilacunar. Almost all the unilacunar arborescent genera are typi- 
cally tropical in their distribution. Subtropical regions, such as the 
Florida Keys, the Gangetic Plain and Hongkong, show a percentage 
intermediate between the two extremes. In South Africa and extra- 
tropical Australia, the great arid areas of the South Temperate zone, 
the unilacunar type is only slightly less common than in the tropics. 
In connection with this concentration of the presumably primitive 
trilacunar woody type in generally temperate and mesophytic habitats, 
it should also be borne in mind that most of the genera and families 
regarded as being particularly ancient are best developed in temperate 
or warm-temperate regions. 
These facts, which are in agreement with a large body of evidence 
gathered by the writers from other sources, all point strongly to the 
conclusion that tropical plants are specialized in almost every par- 
ticular and that the Angiosperms therefore could not have arisen 
under a tropical environment, as has generally been taken for granted, 
but must have had their origin in a climate which, though doubtless 
very equable and devoid of extremes of temperature, was essentially a 
temperate one. Such a climate, as far as we are able to judge of condi- 
tions in the Mesozoic, could only obtain, as a general rule, in upland or 
mountainous regions. On such a supposition the Mesozoic Angiosperms 
which we know are to be regarded as plants which had migrated down 
into the tropical lowlands and swampy areas, where they could readily 
be preserved as fossils. If the earliest Dicotyledons were indeed 
largely confined to temperate upland regions (where fossilization 
could almost never take place), does not this explain why we 
know of so very few Mesozoic Angiosperms and why these are so 
diverse and often so highly specialized and far from primitive? And 
does it not strongly suggest the possibility that on these ancient 
mountains, about the vegetation of which we know so little, Angio- 
sperms may have been developing in extremely remote times, con- 
siderably earlier than the Lower Cretaceous? 
