trigonous-ovate broadly acuminate and as wide as tall and not very stiff,, 

 the stamen not being appressed to the labellum. 



These constructions indicate a certain alliance between these two species 

 and the genus Qeanthus, which by its inflorescence approaches to Amomum. 



Stamen. The stamen is always rather broad linear, with adnate anthers, 

 with a short lorate or without a filament, it is always, at least just before the 

 anthesis applied to the labellum and covered by the dorsal petalum. The 

 anther shows two very different types both represented by the original types 

 of the genus, Homstedtia Retz. H. scyphifera, H. Pininga, H. paludosa, 

 H. villosa, H. rubra, the other is H. Leonurus, H. minor, H. lycostoma, H. 

 mollis, H. clongata, H. Rumphii, H. alliacea, H. cyathifera 



In the first group the filament passing without a demarcation into the 

 connective to which the thecae are adnate is rather stiff leathery, linear 

 oblong, projecting beyond the anther with a flat oblong rounded crest, also 

 broader than itself and with thick fleshy margins. The thecae are linear ) 

 very hairy and produced at the base each into a sterile spur as long as 

 itself, adnate, confluent with the thickened margins of the filament, The pic- 

 ture of GRIFFITH'S (PI. 358 f. 7) is not very exact, the short suture of the 

 thecae and the sterile appendices having been neglected although they are 

 clearly visible to the nude eye. 



In H. villosa these spurs are wanting, here the filament with the con- 

 nective is very broad ovate petaloid with much thickened fleshy margins. 



in H. rubra, which in many ways is somewhat diverging from the other 

 species the spurs are wanting also and here the crest is rather short and 

 truncate, the filament is so short that the anther may be called subsessile. 



In the other group the anther is sessile or subsessile, (with exception 

 of H . cyathifera where the filament is long and broad », the connective 

 does not project beyond the anther either laterally or at the end, so there 

 is no ovate rounded crest, only the back wall of each theca is in some 

 species a little produced, and forms a small rounded mostly recurved lobe. 

 Sometimes {H. lycostoma, H. leonurus, H. deliana) the lobes are visible to 

 the nude eye and the anther is said to be instructed with a bilobed or 

 emarginate crest, in others (H. minor, H. alliacea) the appendix is invisible, 

 and the anther is called emarginate and nude. The limit between the nude 

 and crested anther is here very vague. The rather long thecae split along 

 their full length. Comparing the structure of the anther in this group with 

 that of Alpinia (Catimbium), Elettaria, Qeanthus, the resemblance between 

 these groups is extremely great, and the likeness surpasses by far that be- 

 tween the species of the two groups of Homstedtia. BLUME taking the 

 structure of the anther as a criterion, applied his generic name Donacodes 

 to the crested group only, and he put H. minor where the crest is wanting 

 into Elettaria with radical spikes, notwithstanding the close resemblance of 

 the inflorescence with that of Donacodes Pininga. In the long peduncled 

 species the anther is very long, and in the adult flowers only the upper part 

 is free and opposed to the lip, the lower fourth being adnate to the basal 



