28 PROCEEDINGS OF TUE 



strong flowers at the base of a spikelet have 3 stamens, the 

 weaker upper flowers 2 or 1. This happens also in the diclinous 

 species, as of Carex or Scleria; and I should thiuk I have seen 

 it in 800 Sedges. The same cause for reduction in number of 

 stamens was suggested by the elaborate numerical observations 

 on Stellaria media and Potentilla Tormentilla read last year in 

 this room. In the common case in Sedges I think it very pro- 

 bable that weakness has been the cause of reduction. Perhaps 

 some present think I am tedious and hesitating in assuming that 

 so probable a cause is inE cause of reduction in numbers. In 

 Oyperus, a genus of 300 species, we have one very natural section 

 of a dozen species, the Vegeti, which are large strong plants with 

 fine spikelets ; in this section the stamen is invariably 1 (aa 

 inner whorl anticous one). This fact may illustrate the danger 

 of assuming a cause, and the still greater danger of assuming 

 that a cause, true in a great number of cases, must always be 

 THE cause in a closely analogous case. 



I have already mentioned the few remarkable cases where an 

 increase of the styles from 3 to 4 is accompanied by an increase 

 of the stamens from 3 or 6 to 4 or 8. Leaving these aside, there 

 are only 2 species of Cyperacese out of 3000 that have more than 

 the normal number of 6 stamens. These two species constitute 

 one Australian genus, Evandra ; in this the stamens are 14 to 

 22, all equal in one symmetric whorl, so far as I can make out in 

 the dried material ; and the style has 8 long branches which show 

 no indication of being 4-bipartite. We cannot help asking our- 

 selves the questions, bow did this variation occur, what caused it, 

 and how long did it take to fix it, so that no trace of the evolu- 

 tionary route remains ? I may cite analogies : in the Umbelliferae, 

 say 1500 species, the stamens are never more than 5 ; in the 

 closely allied Araliaceae, 400 species more, the stamens are only 

 5 (or 6-7 when the petals are 6-7) ; but we have half a dozen 

 species of Araliacese where the stamens are very many, 50-70 in 

 Tupidantlius. Such analogies diminish our astonishment, but 

 explain nothing. 



Having imposed, on your attention so many minute facts, 

 excuse me if I recapitulate them shortly in a speculative form. 



"We may first conceive flowers in which the number of each 

 organ was indefinite ; and we may imagine the time occupied ia 

 getting the monocotyledonous ternary type fixed. We have then 

 in Cyperacese to suppose the 3 carpels, witb one ovule in each, 

 reduced to the stage now seen in the coconut, and thence to the 

 present Cyperaceous ovary. We have also to think how long the 

 present ovary must have been fixed that we never see by reversion 

 to arise any trace of the 2 ovules and 3 carpellary walls that have 

 disappeared. 



We have next to observe the anticous angle of the nut, with 

 the stigma belonging to it, disappearing gradually. We may then 

 begin to estimate how long since this stigma must have disappeared 



