24 ME. G. BENTHAM ON GBAMINE^. 



the bracts performing the functions of the deficient perianth 

 are always, I believe, on the maia axis, like the glumes of Gra- 

 minese. Then, again, the perfect union of the two lobes of the 

 palea or of the two lodicules, or even the occasional development 

 of a single central nerve or central lobe, is no absolute proof that 

 they are not in fact double organs ; for where the segments of a 

 perianth are united in a tube or cup, the lateral nerves of two 

 adjoining segments (sepals or petals) often coalesce into a single 

 one which may protrude at the top into an intermediate tooth or 

 lobe. Hackel has w^ell shown that the unity or duplicity is the 

 same in the case of the palea and of the two lodicules ; but it is 

 only conjecturally that he continues the parallel through the third 

 lodiciile, which, when present, never shows any tendency to divi- 

 sion, and whose insertion is not perceptibly higher up than that 

 of the two others. It is quite true that it is often much smaller 

 than the other two, sometimes very minute ; but in several spe- 

 cies of Stipa, in the majority of Bambuseae &c., I have seen the 

 three quite equal and perfectly similar. The only instances I know 

 of more than three lodicules are those of Ochlandra, where they 

 are exceedingly irregular, and of Reynaudia, where I find fonr in 

 two pairs, as described and figured by Kunth ; but then the outer 

 pair, although closely contiguous (on the opposite side of the 

 floral axis) to the upper ones, appear to me to represent the palea 

 which is otherwise deficient. The minute bodies above the lodi- 

 cules in the female flowers of Pariana, which Doell mistook for 

 additional lodicules, a,ppear to me to be rudimentary staminodia ; 

 they are very mipute and irregular, and not always to be found. 

 I have observed that the search for homologues to the palea and 

 lodicules in the Orders nearly allied to Graminese has met with 

 but little success. The only representation of the palea that I 

 can find is that mentioned in my above-quoted paper ( Journ. Linn. 

 Soc. (Bot.)xv. p. 516), where it is compared with the hypogynous 

 scales of Hypoelytrum pungens and Platylepis ; and I find that in 

 some species of Eriocaulon (Flora A.ustraliensis, vii. p. 190) the 

 perianth is composed of two outer segments inserted near the base 

 of the floral axis and two or three inner ones close under the 

 androecium, or these inner ones occasionally deficient, the arrange- 

 ment passing gradually through other species to the normal two 

 contiguous series of two or three each. It might therefore be 

 suggested that the palea and lodicules of Graminese represent 

 perianth-segments of an outer and inner series, although I by no 



