on the Hortus Malaharicus, Tart IT. 173 



largely to the species of this genus, considers it as having no 

 considerable affinity to any other family of plants {Enc. Meth. 

 Sup. i. 575.). Mr. Brown in his valuable work on the plants of 

 New Holland (i. 340.) does not differ from M. du Petit-Thouars ; 

 for, although he has an order of Panda7iea, this order consists 

 of only two genera. He places this indeed next to the Aroidece, 

 and allows a certain similitude between these plants ; but the 

 genera of the Aroidece, which have most resemblance to the 

 Fandanus, that is those in the third section, Mr. Brown has not 

 placed nearest to the Pandanea, to which in my opinion they 

 bear a much closer affinity than they do to the Aroidece of Jus- 

 sieu. This third section of Mr. Brown constitutes the Typha: 

 of J ussieu, a natural order, which in my opinion should be pre- 

 served ; and to this, I have no doubt, should be added the genus 

 Fandanus, which bears an affinity to the Sparganium among the 

 Typhce, nearly as strong as the Bambusa does to the A vena among 

 the Graminea. Natural orders, in my opinion, must be founded 

 on general resemblances, taking into consideration the structure 

 of the whole vegetable ; and I regret, that the best botanists of 

 late have given too much importance to minute differences in 

 the structure of parts, which, however important in the propa- 

 gation of the species, are so minute and inconsiderable as to 

 deserve little attention, unless accompanied by a general resem- 

 blance. This is no doubt often the case ; but then this general 

 resemblance, of itself, is what should constitute the difference, 

 whether accompanied by these minutiae or not. Besides, in the 

 present state of science, it seems rash to exclude certain plants 

 from a natural order, because in some of these minutiae they 

 differ from some species that have been examined ; while in by 

 far the greater part of the order it has not been ascertained how 

 far the minute structure extends. Let it not be imagined, from 

 what I have now stated, that I undervalue the minute examina- 



2 A 2 tion 



