25 



which possesses a iiuiiilier of varieties. What Cyanra srabra is in Cyanca, 

 Deiissea laciniata in Ddksca. Clcrtnontia parvlflora CI. Wainieae are in Cler- 

 montia, Rollandia laiivcohila is in h'<illini<lia as far as polyniorpliisni of species 

 is concerned. These species are all in their prime, wliile otliers are on the point 

 of extinction {pahnac formes of Cyanea). The distribution of the genus Eol- 

 landia in these islands, (being confined to Oahu, except a single species "Rollan- 

 dia parvi folia" which occurs on Kauai), would indicate that its ancestor was a 

 different one from that of the other genera and that it did not arrive contemp- 

 oraneously with the other original immigrant or immigrants. 



The (>ecurrence of the large number of allied species of Lobelioideae would 

 indicate a very ancient occupation of this island group by their ancestor or an- 

 cestors, and we come to believe that these closely allied genera and numer- 

 ous species were evolved here in these islands, becoming modified to such an 

 extent that their ancestors cannot now be recognized, rather than that they 

 exi.sfed oi' originated elsewhere, v.iience they arrived in Hawaii, becoming extinct 

 in their ((luntry of origin. That their ancestor or ancestors was or were American 

 there remains no doubt whatever. The possibility exists however that the 

 baccate genera of Lobelioideae were once of wider distribution, that is to say, 

 that they existed on po.ssible intermediate i.slands, Ijetween this archipelago and 

 the coast of South America which islands became sulimerged at a later age. The 

 writer does not entertain the theory of a direct land connection with either Asia 

 or America; the theory of other volcani(.' islands having existed in the early age 

 of the Pacific, forming stepping stones between this island group and the con- 

 tinent of America and Asia seems to him more plausible. In fact such a chain 

 of islands is still in evidence between Japan and Hawaii, while the Galopagos 

 Islands are a remnant of the chain of islands which probably existed between 

 Hawaii and Central or South America. The chain extended probably down to 

 the island of Juan Fernandez, the llcn-a of which has much in common v.itli that 

 of Hawaii. Extraordinary however is the fact that neither Juan Fernandez nor 

 the Galopagos Islands, so close to the American continent, possess Lolielioideac. 

 Hallier* seems to think that Indonesia, Australia and Polynesia must have 

 once formed a mighty .\u.stralian peninsula which was bordered by concentric 

 mountain ranges and v.hose eastern boundary or margin was formed by the 

 Hawaiian Islands and the Pamnoto archipelago. This ]ieninsula he gradually 

 permits to sink, partly through periodic earthquakes, so that the lowlands 

 between these mountain ranges disappeared from the surface of the sea. 

 This still permitted an exchange of plants between the mountains of Tasmania, 

 New Caledonia, New Guinea, the iloluccas, Celebes, Philippines and Formosa. 

 He aLso says that in still earlier times this Australian peninsula was united with 

 America by means of a broad isthmus. 



The depths in the neighliorh'ood of Hawaii are so enormous that such a con- 

 nection does not seem plausible; it is probable that other volcanic islands existed 

 near the American Continent which brouglit the Hawaiian Islands closer to 

 America. 



The devehi]iment of the Hawaiian species of Leihelioidi a( and their great 



* Medejeeliugeu 's Rijiks Hcrbaiiuiii Leiden. No. 8-14, 1012. Hans Halliei', I'ber 

 friiliere Landbriicken, Pflanzen ami Vijlkerwanderungou zwischen Australasien und America. 



