192 THE REV. С. HENSLOW ОМ THE ORIGIN OF FLORAL ASTIVATIONS. 
margins give rise to the posterior placentas as well as the stigma opposite to them (this 
latter being composed of two adjacent marginal and coherent stigmas). Similarly the 
two anterior margins of the 2nd and 3rd carpels will give rise to the anterior placentas, 
with their combined stigmas. Тһе four lateral margins I conceive as cohering down the 
middle of the valves; but their placentas and ovules are usually arrested. Тһе stigmas, 
however, of these lateral margins are normally present in Parolinia. 
That the stigmas are marginal in the Crucifere was noticed by Robert Brown (Misc. 
Bot. Works, i. p. 558); so that the anomaly of their position over or opposite the 
placentas disappears. Fig. 17, Tab. XXV. (copied from Duchartre /.c.), shows these 
marginal placentas and stigmas developed on a carpellary stamen. Тһе same stamen 
also shows (fig. 18) how the replum is formed, viz. by a prolongation of the outer side 
of the margins, the ovules being produced at a short distance within the edge. Lastly, 
fig. 19 (also from Duchartre) illustrates one of many combinations of carpellary stamens, 
in which the kind of arrest of the placentas and ovules I have spoken of, and which I 
conceive takes place along the median line of each valve, is seen at а. It may be ob- 
served, lastly, that the dehiscence of the siliqua, indicated at * in fig. 16, is, according to 
the above interpretation, strictly loculicidal, perhaps the commonest of all the methods, 
and is thus relieved of all the abnormal appearances. 
There are some fruits of the Crucifere in which the venation is conspicuous, and would 
seem to corroborate this view. Thus in Lunaria and Iberis, for example, the transverse 
and branching veins are given off from the margins of the valve, and terminate by anasto- 
mosing on the median line. In other cases, where the venation is more obscure, a trans- 
lucent line may be observed down the middle of the valve, marking, as I take it, the line 
of cohesion. 
Besides Parolinia, mentioned above, Tetrapoma has four carpels ; and Brassica as well 
as Cheiranthus not unfrequently develop two in addition to their normal number, 
under cultivation. 
With reference to the supposed quinary origin of the flower, it may be observed that 
Capparis t has sometimes five petals; and J/egacarpea, if the figure given by Le Maout 
and Decaisne be true, may have exactly 10 stamens. 
It may be asked on what grounds do I assume the right to suppress the fifth part of 
each eyele. Тһе reply is, that such symmetrical reduction from a normal and typically 
higher symmetry is extremely common. Thus Rue (including Haplophyllum), Ааоза, 
Hypopitys, Potentilla, Tormentilla, &c. have habitually both 5-merous and 4-merous 
flowers; while the petals of Lythrum Salicaria range from 5 to 7 in number; similarly 
4—8-merous flowers are found on Jessamine, Elder, &c. 
If now we attempt to account for this reduction or increase, it is only conceivable 
by its being done by some similar or identical arrest or development respectively 1 in each 
whorl. If, for instance, we draw alternate whorls of fives, and strike out one part, but 
any of each whorl, it will be seen how impossible it will be to form uniformly alternating 
whore of fours out of the remaining parts. If, on the other hand, each whorl be 
t 1 Лааг to о ей as whatever theory he advanced for Crucifere must apply to Capparidec as vell. 
