498 MR. G. BENTHAM ON CLASSIFICATION AND 



union of its three normal segments, as may be seen in some spe- 

 cies of Eriocaulon and a few other genera. 



Commelynaceae have likewise been associated with Kestiaceae, 

 but with still less reason than Xyridese, and solely on account 

 of their anatropous ovules and, in some respects, the structure of 

 the seeds. The inflorescence is never glumaceous ; and the ovules, 

 even when solitary, are not pendulous. The order, though tech- 

 nically associated with Xyrideae and Philydraceae, having usually 

 a very delicate petal-like inner perianth or corolla, whilst the 

 outer one is strictly calycine, is a very distinct one, and shows 

 no immediate affinity to any other ; nor are there any connecting 

 genera, unless perhaps Cartonema may indicate some approach 

 either to Liliaceae (Chlorophyteae) or to Junceae (Xerotideae). 



In a third subdivision of Coronariae, those with the whole 

 perianth calycine, might be placed two large orders, not usually 

 found in juxtaposition, but which appear to me to be more conve- 

 niently associated under Coronariae than in any other series. 



Juncaceae are, without doubt, so closely connected with Liliacea?, 

 that some genera have been placed by some botanists in the 

 former, by others in the latter, according to which of the several 

 characters which usually separate them has been taken as abso- 

 lute. Of these characters, the most constant is the above-men- 

 tioned one derived from the nature of the perianth, corolline 

 or petal-like in the Liliaceae, calycine in the Juncaceae. In Xe- 

 rotes, however, for instance, the perianth is sometimes very little 

 different from that of some Asparageae, although in all other 

 respects it is rightly associated with the Juncaceae rather than 

 with Liliaceae. The prevalent habit of Juncaceae, approaching 

 that of the Grlumales, is very rare in Liliaceae ; but in Xerotes and 

 Xantliorrhcea it is often arborescent, as in Dracana and its allies. 

 Again, the black crustaceous testa and fleshy or hard albumen in 

 the seeds of Liliaceae have been given as absolute characters sepa- 

 rating that order from Juncaceae, where the albumen is usually 

 farinaceous, and the testa membranous or pale-coloured ; but in 

 Blandfordia, for instance, the testa is pale and soft ; in some of 

 the baccate genera of Liliaceae the hard albumen becomes some- 



Xanthorrh 



m 



albumen hard or fleshy, as in true Liliaceae. The embryo is m 

 most Juncaceae (except Xanthorrhcea) small and erect from the 

 base of the seed, whilst in Liliaceae it is very variously placed with 



