ch. xvm THE LEGUMINOS^ OF THE PACIFIC ISLANDS 199 



the inland flora of a Pacific island. About half of the Leguminosae 

 of Fiji and Tahiti are coast plants ; and about 30 per cent, of the 

 littoral plants of the islands of the tropical Pacific belong to 

 this order. Since, therefore, Hawaii possesses much fewer shore- 

 plants (30) than does Tahiti (55) or Fiji (80), the paucity of its 

 Leguminous plants is readily accounted for. 



We have next to notice a principle, which is, in fact, deducible 

 from the first, namely, that buoyant seeds are much more charac- 

 teristic of the Pacific Leguminoscs than of any other order. Three- 

 fourths of the species have buoyant seeds, and, in fact, about a third 

 of the littoral Polynesian plants with buoyant seeds or fruits belong 

 to this order. 



It may, therefore, be inferred that the Leguminoscs owe their 

 presence in the islands of the tropical Pacific mainly to the currents. 



From Mr. Hemsley's conclusion that the Leguminosas are 

 wanting in a large number of islands where there is no truly 

 littoral flora, the presumptions arise that when inland species exist 

 that possess no capacity for dispersal by currents they are to be re- 

 garded as derivatives from the littoral flora, and that they owe their 

 origin to a strand-plant possessing buoyant seeds originally brought 

 by the currents. It has been shown in the case of Afzelia bijuga 

 and of Caesalpinia that when Leguminous shore-plants extend 

 inland the seeds often lose their buoyancy, and it is probable that 

 divergence in other characters may occur, leading, as in the moun- 

 tains of Fiji, to the development of a new species of Caesalpinia. It 

 is urged that by a continuation of the same process the inland 

 species, Erythrina monosperma, has been developed in Tahiti and 

 Hawaii, and the inland species, Canavalia galeata and Sophora 

 chrysophylla, have been produced in the last-named group. All 

 these species have non-buoyant seeds, and in all three cases there 

 is no littoral species in Hawaii, it being assumed that the parent 

 strand-plant has been driven inland from the beach. It is not 

 necessary that the littoral species should be now represented in the 

 flora. 



It is remarkable that in almost all cases the cause of buoyancy is 

 of the non-adaptive or mechanical kind, due either to cavities formed 

 by the shrinking of the seed-nucleus during the setting of the seed 

 or to the light specific weight of the kernel. There is but little 

 to show that the buoyancy of the seeds of Leguminosse is anything 

 but an adventitious character of the seed, as far as its relation to 

 dispersal by currents is concerned. Although this capacity has 

 been the great factor in the wide distribution of the species, yet it 



