Dr Dickson on Diplosteinonous Flowers, dec. 249 



If my explanation of the apparent diplostemony in tLe 

 Geraniacese be admitted, the analogy between such an ar- 

 rangement and that of polyadelphous stamens will be at 

 once allowed. In this regard, it is not unimportant to 

 inquire whether these two forms may not sometimes pass 

 into one another ; and I believe that instances of a passage 

 of this kind do really occur. 



the Dombeyese. In Melhania incana there are five staminodes superposed to 

 the petals, and five stamens which apparently alternate exactly with the 

 staminodes. At first sight, the arrangement of parts in this plant seems very 

 incomprehensible. Here are apparently two staminal whorls, and yet the 

 carpels are superposed to the sepals, as in the isostemonous Hermannia. On 

 further examination and reflection, however, I have come to the conclusion 

 that the diplostemony here is only apparent, and that we have merely to do 

 with a much reduced form of the staminal groups which are found in Astrapcea. 

 In Melhania incana, the stamens and staminodes are connected below into a 

 short tube or ring, which adheres to the petals at points corresponding to the 

 bases of the staminodes. When, however, we detach the corolla from the 

 flower, the staminal ring becomes broken into five parts, a stamen and a 

 staminode coming away with each petal. In those flowers which I have 

 examined, the stamen is always to the left side of the staminode, which, as I 

 have already stated, is superposed to the petal, and adherent to it by its base. 

 From the regularity with which the rupture of the staminal ring takes place, 

 it seems reasonable to infer that the fertile stamens do not exactly occupy the 

 indilferent or neutral position between the staminodes which we should expect , 

 were this a case of two alternating whorls. 



In M. decanthera, where, instead of one stamen, there are two in each inter- 

 val between the staminodes, I find that of these two stamens there is always 

 a longer and a shorter one, whose position to right or left as regards each other 

 is constant in the same fiower, although differing in different flowers. These 

 facts seem to indicate that the pairs of stamens here have not an indifferent 

 relation to the staminodes between which they are placed. 



When we consider how easy the transition is from 3Ielhania (through M. 

 decanthera) to Doriibeya, which again is closely allied to Astrapcea where 

 Baillon has distinctly traced the origin of the stamens and staminodes to five 

 groups superposed to the petals, we can scarcely doubt that the androecium of 

 M.'lhania is referable to a polyadelphous type, and thus the difficulty as to 

 tlie position of its carpels disappears. 



It will probably be very difiicult, in the Dombeyese, without organogenic exa- 

 mination, to apportion the fertile stamens to their proper groups, as they appear 

 to vary very widely in their ultimate relation to the staminodes : thus, in 

 Dombeya vihurniflora (Bot. Mag. tah. 4568), the fifteen fertile stamens are 

 collected into five bundles, which apparently alternate with the staminodes ; 

 while, in an opposite direction, an example may be found in Trochetia grandi- 

 flora (Bot. Reg. tab. 21), where the stamens and staminodes unite to form five 

 phalanges, each phalanx consisting of a staminode from which four fertile 

 filaments spring, two on either side. 



