20 A NATURALIST IN THE PACIFIC chap. 



in that group the genus is represented in each case by an inland 

 species, Erythrina monosperma, Canavalia galeata, and Sophora 

 chrysophylla, the last two species being peculiar to those islands. 

 The seeds of the three littoral species will float for a long time in 

 sea- water, whilst those of the three Hawaiian inland species have 

 no buoyancy. I may say that some very interesting questions 

 relating to the origin of these inland species are here raised. They 

 will be discussed in a later chapter (Chap. XV.). 



There are a number of plants belonging to the Convolvulaceae in 

 these islands that behave in an irregular way in flotation experi- 

 ments ; but their inconstant behaviour can in most cases be 

 explained in accordance with the principle that in the same genus 

 the shore species have buoyant seeds and the inland species non- 

 buoyant seeds. Thus, whilst the seeds of the littoral species, 

 Ipomea pes caprse, I. grandiflora (Lam.), and I. glaberrima (Boj.), 

 can float for long periods, and those of the inland species, I. penta- 

 phylla, I. tuberculata, and I. Batatas (Sweet Potato), have no 

 buoyancy, the seeds of other inland species, I. insularis (Steud.), 

 I. bona nox (L.), and I. turpethum (R. Br.), are inconstant in their 

 behaviour. The three last-named species are, however, to be found 

 also flourishing at times at and near the coast, and the varying 

 floating powers of their seeds may probably be connected with 

 their varying stations. This is indeed suggested by the case 

 of Argyreia tiliaefolia in Hawaii, in which in my experiments 

 the seeds of plants growing at the coast floated, sometimes for 

 months, whilst those from inland plants sank. 



This behaviour of the Convolvulacese becomes yet more intelli- 

 gible, and more in accordance with the principle, when we reflect 

 that the cause of buoyancy is not concerned with the seed-coats or 

 with the nucleus, neither of which are able to float, but with 

 the air-spaces left by the incomplete filling-up of the seed-cavity 

 by the crumpled embryo. The extent to which the seed-cavity is 

 filled up varies not only between different genera and between 

 different species of the same genus, but also amongst individuals of 

 the same species. Even the seeds of Ipomea pes caprse, amongst 

 the most typical of floating seeds, display this variation, and they 

 show it also in their floating power, since about a third of the seeds 

 usually sink during the first month or two of the flotation experi- 

 ments. We can thus explain also why in the case of Ipomea 

 insularis seeds from Fiji floated for months, whilst those from 

 Hawaii had no floating power. 



The seeds of the different species of Hibiscus also appear to 



