494 mr. a. bentham: on classification and 



nous albuminous orders by the equitant leaves, regular or oblique 

 triandrous flowers with the anther-cells turned outwards, and 

 frequently by the centrifugal inflorescence. But none of these 

 characters are constant. Equitant leaves occur in various orders, 

 and are not in Crocus, for instance. The inflorescence cannot be 

 determined when reduced to a single flower, and in some genera 

 is certainly centripetal. Diplarrhena has the irregular suppres- 

 sion or imperfection of one stamen, as in some Scitamineae. 

 Campynerna, which, for reasons given in 6 Flora Australiensis/ I 

 have transferred from Amaryllideae to Irideae, has six stamens ; and 

 Hewardia, which, from its superior ovary, is placed in Liliaceae, 

 has the habit and all the other characters of Irideae. 



I have in the ' Flora Australiensis ' given my reasons for adopt- 

 ing the wide extent given by some botanists to Amaryllideae so 

 as to include Haemodoraceae, Hypoxideae, Vellozieae, and (with 

 somewhat less certainty) AJstroemerieae. Their anther-cells 

 turned inwards or lateral, a character apparently more constant 

 here than in the hypogynous or perigynous orders, is perhaps the 

 only certain character to separate this large order from Irideae ; 

 whilst the closely allied Liliaceae are removed only by the technical 

 character of the superior ovary. 



The small order Taccacese, closely allied to the Amaryllideae, is 

 distinguished by the parietal placentation with numerous ovules, 

 and Dioscorideae by the unisexual flowers and peculiar habit con- 

 necting them with some of the climbing genera of Liliaceae, and 

 especially with Roxburghiaceae — a specimen of&Boxburghia with- 

 out flowers having even been mistaken by Brown for a Dios- 

 corea. Bromeliaceae, including genera with the ovary partially, 

 and perhaps wholly, superior, connecting the Epigynae with the 

 Coronariae, are exclusively American ; and I have as yet exa- 

 mined in detail only a few of the genera. 



Coronariae is a name originally given by Linnaeus to a small 

 group of Liliaceae, but is here taken in the extended sense given 

 to it by Endlicher, with the still further addition of that author's 

 Principe* and of two orders taken from his Enantioblastce. It 

 conveniently expresses one of the chief characters of the alliance, 

 a normally biseriate perianth, of which the inner or both series 

 are spreading or prominent above the subtending bract. This 

 character is, I believe, constant in the alliance, though it is also 

 found in the Alismaceae classed, for other reasons, with the Nudi- 

 floraB and in two small genera of Eestiaceae. The superior ovary 



