. 86 
which enter the cavity of the carpel through the abovemen- 
tioned aperture (Spermopyle miAz), and hang freely in them 
without ever adhering to the parietes of the cell at any period 
of their growth. It is probable that Don (Prod. Fl. Nep. 
p- 186) took the carpels for the outer envelop of the seed 
when he asserts, as quoted by DeCandolle, in his Prodro- 
mus, that the seeds are erect. But in the genera Hedera 
and Adoxa it is evident that the seeds are attached to the 
axis withoutside the corneo-membranous carpels. So in 
Panax in the words of DeCandolle, “ Fructus carnosus, 
compressus, orbiculatus aut didymus, bilocularis, loculis co- 
riaceo-chartaceis monospermis,” and in Aralia itself he de- 
scribes the fruit ‘‘ Bacca 5-locularis sepe torosa; Pyrene 
chartacee.” Hence I maintain that my conclusions are not er- 
roneous, and that consequently the extracarpellary attachment 
of the seeds in these cases is proved. 
Lastly, in the family of the Viburnee Bartl. or Sambu- 
cee Kunth, which by the structure of the fruit as well as by 
the habitus is so different, as remarked by Bartling, from the 
Caprifoliaceze, with which it has hitherto been united even by 
DeCandolle,—in the young ovary three cells may be plainly 
observed, and outside of them a continuation of the axis or 
floral peduncle, emitting into the cavity of each cell a sus- 
pended ovule. In the first stage of the growth of the seed it 
1s easy to perceive that that hard shell which has been erro- 
neously taken for the outer integument of the seed is the 
carpel, for its cavity is then very considerable, and_ the 
ovule scarcely occupies a tenth part of it. So I have found 
it in the Sambucus racemosus; and the close analogy be- 
tween the carpels of the genus Viburnum with the mo- 
nospermous cells of the Hamamelider, Br. and Cornee 
DC. (in which no one has taken the stony carpels for the 
testa of the seed) proves that it is by abortion that here the 
external symmetry is destroyed of three carpels, to which 
correspond the three styles in the flower, placed by Linnzeus 
in the order trigynia of his fifth class. 
The object of my present observations does not extend to 
the critical examination of the structure of the fruit in all the 
genera belonging to the natural families I have quoted, but I 
venture to assert, that the greater the number of observations 
the more they tend to confirm the truth of Agardh’s rule, that 
the seeds are attached not to the margin of the leaf, but to a 
special supporting organ. 
