492 MB. G. BXNTHAM ON CLASSIFICATION AND 



Alliance 3. Nudiflor^e. Ovary free, apocarpous monocarpellary or 

 rarely syncarpous. Perianth none, or reduced to a small scale under each 

 anther (except Alismacea). Seeds with or without albumen. 



* Anthers opening outwards or in terminal pores ; perianth none, or 

 scale-like : Pandanece, Aroidete (including Orontiacece), Typhacece, Lem- 

 nacece, Naiadece (including Zosteracece and most Juncaginece). 



** Anthers opening inwards ; perianth more or less petaloid : Alismacece 

 (including Butoiaece and Aponogeton). 



Alliance 4. Glumales. Ovary free, 1-ovulate or with 1-ovulate cells. 

 Flowers usually in heads or spikelets within imbricate bracts or glumes. 

 Perianth either none or scarious or glume-like, and usually concealed 

 within the bracts. Seeds albuminous. 



* Ovary often with more than one cell ; ovule pendulous : Eriocaulea, 

 Centrolepidece, Restiacea. 



** 



Ovary always 1 -celled; ovule erect : Cyperacea, Graminece. 



There are, besides, a few small American more or less anoma- 

 lous genera which have been proposed as distinct orders, but 

 which will, I believe, all be referable to some of the above orders 

 of Coronarise or Nudiflorae. 



The first of these alliances, the Epigynae, has been more or 

 less accepted by the majority of botanists as founded on a prin- 

 ciple established by Jussieu and never satisfactorily superseded. 

 It has been objected to by others as separating Hydrocharidese 

 from Alismacese, and Amaryllidese from Liliacese ; and the mam 

 character is ambiguous or fails entirely in some Bromeliacese ; yet 

 it is convenient and more definitely marked out in nature than 

 any other that has been hitherto proposed. 



The most striking objection to this arrangement is the position 

 of the Hydrocharidese, so far removed from the Alismaeeae, to 

 which they bear a considerable resemblance in their aquatic habit, 

 their embryo, and, to a certain degree, in the perianth and stamens, 

 and with which they have even been united, as a tribe only, by 

 Ascherson and others. The habit, however, is a resemblance only, 

 which they have in common with genera belonging to widely dis- 

 tant orders, such as Haloragese, for instance, among Dicotyledons, 

 and which, being in a great measure adaptive, affords no indication 

 of affinity. The want of albumen is important ; but neither here 

 nor any where else can it be taken as an absolute character ; and, 

 besides the technical difference in^the inferior ovary, the nume- 

 rous seeds or parietal placentas of several genera of Hydrocha- 



