244 GEOGRAPHICAL DISTRIBUTION [pt. ii 



Indies, a region which contains colossal numbers of endemic 

 forms. Other regions of absurdly small size are the Galapagos, 

 Juan Fernandez, the Hawaiian Islands, Kerguelen, etc. 



It is difficult to understand why so much energy and labour 

 has been applied to the problem of differentiating floral regions, 

 for one fails to perceive any object which is gained by defining 

 them, for example, any progress in the study of geographical 

 distribution. The term floral region may, it seems to us, be added 

 to "constitution" in the extract from Huxley given on p. 231, 

 as one of the verbal anodynes by which the discomfort of igno- 

 rance is dulled. When we say that Lactoris fernandeziana (which 

 is now usually regarded as of family rank) is characteristic of 

 the Juan Fernandez floral region, it sounds as if we knew more 

 about it than if we simply stated the bald truth, that it occurs 

 upon the island of Juan Fernandez. In plain fact it is no more 

 specially characteristic of that island than Centaurodendron 

 dracaenoides, an endemic genus of Compositae, or Spergularia 

 rubra, which is of enormously wide distribution. 



What it reallj^ comes to is that as, on the whole, in recent periods 

 of the world's history, migration of plants has been largely 

 southwards (owing to the cooling of the north), and the subse- 

 quent northward migration has not yet had time to show very 

 obvious results, the southern regions contain greater proportions 

 of endemics. In the same way, the islands being at the edge of 

 the dispersal that has gone on, where the oldest and most 

 variable (p. 218) types occur, and being also isolated, show great 

 numbers of them. But, as pointed out on p. 170, it must not be 

 forgotten that the larger regions of the world have greater pro- 

 portions than the small. 



Ver}' little consideration is required to show that these divisions 

 or floral regions are very arbitrary, but very little trial of the 

 actual facts is needed to prove to an enquirer that it is a matter 

 of extraordinary difficulty to improve upon them. The islands, 

 by being clearly cut off from the rest of the world, are evident 

 divisions to make, but to divide the continents, except to cut 

 off a few such obvious regions as South Africa or ^Vest Australia, 

 is quite a different matter. 



The one thing that comes clearly out is that endemics are not 

 a good test to apply, and with the new light that is thrown above 

 upon the question of endemism, it would seem probable that 

 this test vAW no longer be used. It gives a much greater pro- 

 portionate value to small areas in the south or upon islands than 



