152 MK. G. BEXTHAM OX MTKTACEJi;. 



at that time disregarded), it was, after the reformation of the 

 tribe founded on embryonic characters, limited by DeCandoUe 

 to the species with few seeds, a bony testa, and horseslioe embryo, 

 always supposed to be accompanied by 5-merous flowers ; and 

 numerous 4-merous, especially Chilian, species, of which the seed 

 was unknown, were transferred to Eugenia. The greater number 

 of these, however, have since proved to have the seed and embryo 

 of MyrtuSj or nearly so, and have either been restored as sections 

 of Myrtus, or raised into distinct but closely allied genera, thus 

 forming a group, distinguished from Campomanesia and Fsidivm 

 by the ovary never more than 3-celled, and from the latter by 

 the form of the embryo-limb and generally by the habit, from 

 Myrcia^ Marlieria^ and Calyptranthes by the ovules alw^ays more 

 thaU two in each cell, from Eugenia by the embryo only as a 

 positive character, with occasional collateral aids from inflores- 

 cence and habit, and from various smaller genera by the absence 

 of the exceptional characters which have severally induced their 

 separation. It is the group thus (perhaps still somewhat vaguely) 

 limited that we have adopted as the genus Myrtus^ reducing to 

 sections some tolerably distinct subordinate groups established 

 by A. Grray and others. If I observe that the generic limits are 

 still somewhat uncertain, it is because the number of cells of the 

 ovary, although perhaps never more than three in Myrtus^ except 

 in a few abnormally exceptional flowers, is nevertheless sometmies 

 although rarely, reduced to two or three in Fsidium ; and the em- 

 bryo in a very few species has so thick and little curved a radicle^ 

 and the cotyledons so very small, that it may be mistaken for the 

 apparently homogeneous embryo of Eugenia, not to speak oi the 

 numerous species of which the embryo is as yet unknown. 



Eight or nine genera have been proposed to be dismembered 

 from the American Myrti^ some of which form excellent sectlon^ 

 which we might even have adopted as genera, had it not been 

 for some Australian, and even a few American, species, whiea 

 tend to invalidate their artificial characters, whilst there is nttie 

 or nothing to render them really natural divisions. These are : 



1. TTgni^ proposed by Turczaninow for those Chilian and Andme 

 species which, on account of their 4-merous flowers (the emhry 

 being then unknown), had been referred by DeCandoUe to Eugenu^^ 

 They have, moreover, the calyx-lobes spreading in the bud, an 

 the erect anthers of Calycolptcs and other Psidioid genera. -*- ^ 

 is thus the most distinct of all the subordinate groups, and we hesi- 

 tated much wliether we sliould not admit it as a substantive genus , 



J 



J 



