MB. G. BENTHAM OK MYKTACEJE. 149 



none of them appear to be absolute, may yet, taken togetber, be 

 allowed to separate two groups wbicli we retain as the genera 

 Campomanesia and Pddiiim, each with about 100 species, accord- 

 ing to Berg's views (probably reducible by one half), distributed 

 by him, the one into five, the other into three genera. The first of 

 these characters brought forward by Berg, the embryo " spiralis, 

 2-3-cyclicus " in the Campomanesia group, only " uncinato-cur- 

 vatus, eubspiralis v. semiannularis " in Psidium and adjoining 

 genera, is, perhaps, the least to be relied on. Independently of the 

 numerous species where the embryo is unknown, if we look at 

 the seeds figured by Berg in the ' Flora Brasiliensis,' t. 53 {Britoa) 

 and t. 42 (^Psidium'), we should surely call the embryo in both 

 uncinato-cnrvatus, and certainly not 2~3-cyclicus in the Britoa as 

 it ought to be ; and other equally contradictory instances occur in 

 species not figured. It is probable, however, that in most species 

 the seeds have a thinner testa and a longer radicle in proportion 

 to the cotyledons in Campomanesia than in Psidium ; the calyx- 

 limb is less frequently produced below the lobes so as to split in 

 enlarging ; the ovary-cells are usually, but not always, 6 or more 

 in the former, only 4 or 5 in the latter ; and the most constant 

 distinction given by Berg, as far as I have had occasion to verify 

 it, is the arrangement of the ovules in 2 rows (or very rarely in 4) 

 in each cell, whilst in Psidium they are more irregularly crowded. 

 The species of Campomanesia are all American. 



"We propose reuniting with Campomanesia four of Berg's genera : 

 1. Abhevillea contains several species, correctly separated from 

 Psidium on account of the arrangement of the ovules and the 

 structure of the seeds (where known) agreeing with Campomanesia^ 

 from which he only appears to distinguish it bv the calyx-limb, 

 more or less developed below the lobes ; but in this respect I 

 cannot trace the difference between several species of Abhevillea 

 (e. g. A. Gmviroba, Berg) and others of Campomanesia (e. g. C. 

 lineatifolia^ Berg). 2. Acrandra^ which has the connectivnm of the 

 anthers produced into a very short point — a single character un- 

 accompanied by any other diff'crence in habit or structure. 3. 

 Pritoa, with an ample calyx-limb almost closed, with very small 

 lobes in the bud, but afterwards splitting as in the majority of 

 Psidia; but this character, though generally good, affords too 

 many gradations in Psidium itself to be taken aS generic when 

 unaccompanied by any other. 4. Lacerdcea^ at first distinguished 

 from Britoa and Campomanesia by the calyx-lobes free from the 

 base as in the latter genus but coriaceous as in the former. Tlie 



