Returning to the floral diagram, we see that the two lodicules, the three stamens and the ovary still remain to be explained. Much discussion has been held regarding the lodicules. Functionally they are said to aid in the divarication of the paleæ when the period of anthesis arrives, and the stamens and stigmatic lobes are to be exposed, by swelling and driving the valve-like paleæ apart. Morphologically they have been explained as representing the rudimentary perianth, here reduced to two minute scales, but in some exotic grasses (_Bambusa_, _Stipa_, &c.) three lodicules, or even more, are present. (Figs. 35, 36.) On the other hand they may be, and probably are, scales of the nature of minute bracteoles and of no significance to the flower itself. [Illustration: Fig. 35. Floral diagram of ordinary grass. Each pair of paleæ--_i.p._ inner and _o.p._ outer palea--encloses three stamens (_s_), two lodicules (_l_) and the ovary. _st_, stigmatic plumes. _a_, axis.] [Illustration: Fig. 36. Floral diagram of a Bamboo, showing six stamens, three inner (_i.s._) and three outer (_a.s._), and three lodicules (_l_) in addition to the ovary. _i.p._ inner and _o.p._ outer palea. _a_, axis.] If this is so the flower of the grass is perfectly naked, and consists in the typical case of three stamens and one carpel. The development of the ovary lends no support to the view that there are two carpels: the stigmatic plumes are not separate styles. Nor does the fact that some grasses have six stamens lend support to the idea that the flower is derived from the trimerous type so common in Monocotyledons: other numbers may occur--e.g. as few as two (_Anthoxanthum_, Fig. 32) or even one only (_Uniola_, _Cinna_, &c.) or as many as 20 or 40 in certain other exotic grasses. Even when three stigmatic plumes are developed, as in some Bamboos, close investigation does not confirm the view that the ovary consists of more than one carpel. We must therefore regard the flower of the grass as typically composed of one carpel and three stamens, with no perianth whatever. It is subtended by one or more bracteoles (the lodicules), and enclosed in a pair of bracteoles one higher than the other (the paleæ). The glumes are bracts of the partial inflorescence--the spikelet. That there are some departures from this type in detail does not invalidate the importance of the fact that most grasses conform to it. I now pass to the consideration of a grouping of our ordinary grasses according to their floral arrangements.