66 NATURAL niSTOBY OF PLANTS. 



especiiilly re mark able for its imbricated rostivation — a character met 

 with in the rest of the order ; but in the third tribe {BcumnculacecB) 

 the seed is erect, not pendulous, and the petals are bilabiate, or 

 provided with a small l)asilar scale ; the Tlcllchore^, which constitute 

 the fourth tribe, have polyspermous carpels ; and the fifth {P(eonie<2) 

 especially characterized by the introrse anthers, is considered as, 

 perhaps more properly, a distinct order. Endlicher,' and Bentham 

 & HooKEK- accept De Candolle's tribes without alteration. 

 LiNDLEY^ slightly modihed them by uniting the Pseonies to the 

 Helkhorcce and putting Xanthorhza into a special section with the 

 ActecB. These different classifications are of more or less service 

 practically ; but we have not retained them, as they rest on the 

 absolute value of characters which are not constant. The opposition 

 of the leaves in Clematis is a character easily observed, but of no 

 great absolute worth, as many other genera exist of which some 

 species may have alternate and others opposite leaves. 



The aestivation would appear a completely satisfactory character 

 if it were not that at a certain stage in the life of Clematis the 

 perianth may become imbricated like that of a Ranunculus.^ The 

 absolute number of the ovules would have some value if we were not 

 now aware that Clematis, Anemone, Adonis, have all really five ovules 

 instead of one ;' while in Isopyrum we may have some carpels with 

 several, and others with only one.^ The direction of the ovule, 

 whether ascending or descending, is not more absolute as a distinc- 

 tion, for in Adonis alone we may observe instances of both.'' As to 

 the introrse or extrorse aspect of the anther, it has long lost much 

 of its value ; and if the Actece, which we put near the Paeonies, like 

 them have their anthers generally introrse, in some they are ex- 

 trorse ;" and so they are undoubtedly in many Banunculacece with 

 multiovulate carpels, as the Larkspurs, Aconites, NigellcB, &c. Hence 

 in our essay to group the Bannnculacece, we have been unable to 

 recognise the absolute loorth or the subordination of characters. We 

 have been compelled to admit and to combine the greatest possible 



' Genera Plantarum, sec. ord. nat. dispos. 5 ggg p^ 41^ flg_ 7(5^ p^ 45 & p, 51, 



(1836-W), 8'«, Ordo clxxviii. 6 See p. 48, note 3. 



- Genera Plantarum, ad Exempl. impr. in 1 See Adamonia, ii. 209. 



nerh. Keiren-s. Def., i. (]862_), l-lo. « In some flowers of Cimicifnqa frigida 



' Vegetabh Kingdom (18t6), 425, Ortl. cliv. Wall., tlic anthers are clearly introrse. 



■* Sec J). .'O, note 1, and Adansonia, iv. 5.5. 



