346 Willis. — Further Evidence for Age and Area; 
no less than 24 were Ferns or Lycopodiums, and 20 Orchids, while the 
remaining five were made up of two each from the Compositae and 
Asclepiadaceae, and one of the Apocynaceae. It is thus evident that 
even at a range of 40 miles most of the parachute mechanisms begin to break 
down. In any case the families in which long-distance transport by wind 
may occur are few. Or, to take the case of dispersal by birds, there 
is little or no good evidence for any beyond, say, at the most 1,000 miles. 
Or yet again, where there is a marine current, transport may occur by its 
means to great distances, provided the climates at the two ends are not 
hopelessly dissimilar. In the Maldives, for example, Prof. J. Stanley 
Gardiner and myself ( 16 ) showed that several of the water-borne species must 
probably have come from Malaya or from the Seychelles — a long distance 
in either case. 
But there are quite a number of cases in which none of these causes can 
be called in as aid, and yet a great discontinuity is accepted in the geo- 
graphical distribution. To take an extreme instance, the genus Cryptotaenia , 
DC. (Umbelliferae), is described as having one species in Canada, one in 
Calabria, and one in the Cameroons. If there were, say, twenty to thirty 
more with the same distribution, then it is obvious that some explanation 
would have to be found. But with this one case only, it is evident that the 
identification of these three species as belonging to the same genus must 
rest upon the convergence of evolution or upon a parallel development or 
polyphyly ( 14 , p. 446). Such, cases are very numerous, though there are 
not many with quite so marked a discontinuity as that quoted. 
Or, as another case, let us take the family Lardizabalaceae, sometimes 
split off from Berberidaceae, chiefly on the ground of climbing habit, uni- 
sexual flowers, and polycarpellary ovary. There are seven genera, of which 
five occupy the region from the Khasia Mts. (Assam) to Japan, whilst 
Boquila and Lardizabala are found in Chile. Now if the first five were 
found in Australia or New Zealand, this would be normal enough ; these 
countries have many groups in common with temperate S. America, and 
even species. But from eastern Asia to Chile is too great a gap. The 
fleshy fruit could not be carried by birds to such a distance in such a 
direction, and we have no longer any right to invoke the dyingout of linking 
forms in the intermediate region, without direct evidence. It may have 
occurred, but we require evidence to prove it. This being so, it seems to 
me almost impossible, on the present evidence , to regard the Lardizabalaceae 
as a monophyletic unit. The two isolated groups may have had a common 
ancestor not very far back, but on the existing evidence it will be safer to 
reunite them to Berberidaceae (or Menispermaceae, if preferred), as two 
separate sub-groups. This is a thesis which might easily be developed to 
any extent, but I have simply chosen these two examples to illustrate my 
main point, that geographical affinity is required as well as structural in 
settling relationship. 
