KAri'LKSlA AKNOi.Dl. 107
when tlicy again become active and increase in size until they once more com[)letely cover it.
AVhile the development, as I have liere described it, of the nucleus and its integument in llaj/lcsia is going on, another change is at the same time gradually taking place, namely, at first a slight bending, which at last ends in a coniplete inversion, in the direction of the nucleus and its integument in regard to the placenta, with which, in this advanced stage, the perforated apex of the latter is nearly or absolutely in contact.
In this change of direction, the ovulum of Ihijllcmi resend)les that of the far greater part of Phienogamous plants: the change, however, is effected in a w^ay which is much less connnoii, the curvature in Rajllesia taking })lace solely in the npj)er part of the funiculus, the direction of the inverted ovulum being parallel with, but distinct from, the portion below the curvature; whereas in Phsenogamous ])lants generally, the curvature is produced in that part of the funiculus which is connate with the testa or outer integument. For this difference, a reason, perhaps, may be assigned; the integument which generally forms the testa or outer coat being in Ilafflesia entirely wanting, or only indicated ])y the remarkable dilatation of the apex of the funiculus.^
In the more essential points of structure, the ovula of llf/diiora and Cijtunis agree with that of HaJ/lesia. They ditl'er, however, in both these genera in retaining their original direction.
In Hydnora I have ascertained the perforation of the single hitegument and the position of the included nucleus,:^-ri but the very earliest stages I have not yet distinctly seen; while in Ci/tinus, in addition to the coat analogous to that of IxajjlcHia and Jl)/dnora, a two-lobed or bipartite mem- brane is observable.
Of these three genera, I have hitiierto observed the pollen
> The. earlier production ol' the inner of llie two coats generally present in the ovula of ^lueno^'an)()Us |)lants, and vhe absence of the outer in this and several other cases, will probably be considered a valid objection to the tcrminolo<:y ol M.Mirbcl.
�� �
