1818.] M. De Candolle on Double F/oicers. 331 



The staminal petals, those which are produced by the trans- 

 formation of the stamens, form the most frequent kind of double 

 flowers, at least among the ranunculaceae. The stamens maybe 

 transformed into petals in two ways ; either the filament is 

 enlarged, and the anther totally or partly disappears, or the 

 filament remains in its natural state, and the anther is increased 

 in size. The first of these modes is the most frequent occurrence ; 

 and with respect to this variety, we may make the following- 

 observations. 1. The facility with which the filament of a 

 stamen is transformed is as much greater as the number of 

 stamens is more considerable ; this is exemplified in the rosacea?, 

 the malvaceae, the roagnoliacese, and the ranunculaceae. 2. The 

 facility with which the filament is transformed is greater in 

 proportion as the thread is flatter ; this is observed with respect 

 to the rosacea? and the lihacese, which have filiform threads. 

 3. When a flower has several rows of stamens, it is the exterior 

 rows which have the most tendency to become transformed, 

 either in their natural state or in consequence of cultivation. 

 Only one exception to this rule has been observed, which is 

 mentioned by Mr. Brown, as existing in a plant, from New Hol- 

 land, to which he has given the name of eupomantia. The 

 transformation of the anthers into petals is a much more rare 

 case than the preceding ; and, indeed, there is scarcely any well 

 marked example of it except among the ranunculaceee ; in some 

 of the genera of this family, the filaments remain in their natural 

 state, and the anther is developed into a horn, almost always 

 nectariferous at its base, divided into two lips, producing a very 

 singular form ; an example of this is to be observed in the 

 twisted columbine of the florists, the aquilegia resupinata. 



These remarks upon the conversion of the stamens into petals 

 of different forms apply to the origin of true petals. These may 

 be regarded as merely exterior stamens, which, in the natural 

 and ordinary course of things, are transformed into plates or 

 horns, as the stamens of the interior rows are in certain acci- 

 dental cases.* All the ranunculacese which become double by 

 the development of the threads into flat petals have their 

 common petals generally flat ; all those which become double by 

 the development of the anthers into horns that are more or less 

 labiated, have their natural petals in the form of labiated horns ; 

 all those which when they become double have petals of both 

 kinds, have both kinds in the natural state ; and it is remarkable 

 that these differences, deduced from the structure of double 

 flowers, perfectly agree with the classification of the ranuncu- 

 laceae deduced from the structure of the fruit. 



The proper rammculacea?, that is, those which have the 

 anthers open on the external side, are divided into three sections : 



* This doctrine is maintained and illustrated by M. De Candolle in his " Theorie 

 Klrmenlairc de la Butanique." 



