148 MTi. G. BENTUAM OX MrKTAC3:JE. 



the great genera we have adopted may yet be satisfactorily divided 

 into subordinate groups, but ^\e think that in this process he has 

 not met with more success than his predecessors. In attempting 

 to determine species by his work it has appeared to me that the 

 divisions he proposes are not natural enough to enable us to sort 

 the specimens approximatively without examination ; and at the 

 same time, if taken as artificial sections, their characters (beyond 

 the embryo, which is so rarely to be met with) are too vague and un- 

 defined to serve for practical purposes. AYe think therefore that 

 the classification of the South-American species of Campomanesia^ 

 Paidium^ Myrtus^ Myrcia^ Marlieria^ Galyptrantlies^ and Eugenia 



is a labour to be entirely recommenced when a botanist shall be 

 found courageous enough to undertake so tedious a task. 



The first genus we have adopted, Feijoa, Berg (first described 

 by him under the name of Ortlwstemon^ which proved to be pre- 

 occupied by a Grentianous genus of Brown's), is a single Brazilian 

 species, with something of the aspect of Psidmm^ but remarkable 

 for its thickish filaments, all, or at least the outer ones, erect in 

 the bud. That character alone would not, indeed, be necessarily 

 more than specific ; for we know that in Eucalyptus^ for instance, 

 although it distinguishes Cornutm as a section, it is specific only 

 in Tl. 7na7'gina fa and not even that in E, tereticornis and E. Old- 

 Jieldii. In Feijoa^ however, the seed, according to Berg, has the 

 embryo imbedded in albumen, which, as far as hitherto kno"v\Ti, is 

 absolutely exceptional in the Order. Berg has therefore pro- 

 posed it as a distinct subtribe of Myrtese. We know, however, 

 that in Leguminosse, Tiosacese, and other typically exalburainous 

 orders the occasional presence of albumen is no more than generic 

 and even then often separates species which are otherwise very 

 closely allied. Our specimens of Feijoa have no ripe fruit, and 

 Berg does not figure the seed, so that we do not know what is the 

 proportion of albumen present, nor can we conclude that it 53 

 always absent in the nearest allied genera until the seeds of more 

 of the species shall have been observed ; we have therefore retained 

 Fcijaa, but as a genus only. 



Campo-manesta, Euiz and Pav., was originally distinguished by 

 the authors from PsiJiiim by the few large seeds arranged m » 

 single scries round the central fleshy axis, to which Kunth added 

 that of the increased number of cells to the ovary. De CandoUe, 

 having no specimens, had no means of verifying these character?, 

 which have broken down when applied to the numerous species 

 now knowru Berg has supplied several others, which, althoug 



) 



