16 



THE BALANOPHORAL DIVISION. 



in its parietes* alternate with the other 2. Brunonia however differs from Centaurea in the 2 larger and much more strongly marked lines being sometimes right and left of the axis instead of 

 always anterior and posterior, but this variation is also common in the Composite? and is therefore of no importance ; it is always accompanied by a corresponding change in the position of the 

 stigmas, and is consequently owing to a change in the position of the carpels. 



5, The indusium and stigmas. The only difference between the germen of Brunonia and that of the Composite apart from the non-adhesion of the calyx is in the indusiate stigma and in 

 the different relation of the lobes of the stigma to the 2 more obvious lines in the parietes of the ovary. I have described the 2 more obvious lines in the opposite sides of the parietes of the ovary 

 (which as in Composite are, I have no doubt, the midribs of the two carpels) as terminating each in the centres of the corresponding lobes of the indusium, whereas the 2 lobes of the stigma are 

 alternate with those of the indusium, and consequently alternate these lines instead of opposite as in Composite, which would be an anomaly with difficulty accounted for were it not that in Goodeniaceae 

 the stigmas are also alternate with the 2 lobes of the indusium. The 2 lobes of the indusium are in Goodeniacem anterior and posterior, corresponding with the 2 carpels which are as regularly 

 anterior and posterior, and the lobes of the stigma are consequently alternate with the carpels, i.e., in relation with the placentae instead of 4 with the dorsal ribs, and the inference therefore is that 

 such is also the structure of Brunonia. 



6. The ovule and the position of the fertile carpel. The ovule in its relation with the ovary agrees so constantly with the Composite that in connection with other characters it places beyond 

 donbt that the nearest affinity of Brunonia is with that family. I have noticed that the raphe in Composite (loc. cit.) is always next to the anterior rib of the ovary and in relation with the anterior 

 lobe of the stigma, i.e., when the carpels are anterior and posterior, and I find in Brunonia australis that the raphe is almost always in relation with the anterior of the 2 more strongly marked lines 

 in the parietes of the ovary, the exception being as far as calculation has extended only 1 in 28. The conclusion consequently is, that in Brunonia as in Composite, the raphe of the ovule is next 

 the dorsal rib of the anterior carpel. Sometimes however the carpels of Brunonia are right and left of the axis or oblique, and then I find the raphe removed so as to correspond as before in its 

 relation with 1 of the 2 larger lines. In fact Brunonia agrees so decidedly with the Composite in the structure of its ovary and in the position of the raphe, that I believe it has but a comparatively 

 distant affinity with Goodeniacece, as in the latter family the raphe is either lateral, or (when the ovule is single and erect,) next the placenta. 



In its appearance also the ovule closely resembles that of Composite in having a larger foramen than usual in monopetalous plants, and the raphe soon becomes lost in consequence of its 

 fibres spreading as they ascend as in some Composite, so that in the upper third of the ovule it is not discernible as a single cord. 



It now only remains to endeavour to show what would be the position of the carpel supposing the ovary to be reduced to only one, and this can be satisfactorily accomplished provided it 

 is taken for granted that the structure of the ovary and the position of the raphe are the same as in Composite, to which I believe no objection whatever exists. The position of the raphe then 

 being taken as a guide to that af the fertile carpel, it proves to be either anterior or lateral, occasionally having an oblique direction between these two positions. Of 28 ovaries of Brunonia 

 australis in one capitulum of flowers, the fertile carpel was anterior in 12, lateral or obliquely lateral in 15, and posterior in only l.f It may therefore in this character differ from Composite in the 

 carpel being occasionally though very rarely posterior, and agree with Dipsacece where in Morina % it is occasionally so, but as rarely as in Brunonia. 



As the ovary superior and the filament having a joint near its apex occur in Melastomacece, that family well represents the polypetalous form of Brunoniacece, having the same relation to 

 Myrtacea as the latter to Composite. The jointed filament of Melastomacem appears to me identical with that of Composite, and in the inner row of stamens in Heterocentron it is quite as near the 

 base of the anther. 



Lythraceje. 



The affinities of this well-known family have hitherto remained doubtful, but the newly discovered genus Axinandra, Thwaites, I believe, shows that it is very near Melastomacece, of which 

 it may possibly prove to be a section. Looking at the flower only, this genus might I believe be referred to the section Memecylea, as it has the peculiar anthers of Memecylon and so much so 



* In the smaller ovaries these 2 lines are not perceptible, and sometimes one is larger than the other so as to give the ovary at first sight a 3-ribbed appearance. 



f In the Ann. Nat. Hist. 2nd Ser. Vol. XL the position of the carpel in Brunonia was described as variable, being often posterior, but that inference was deduced supposing the raphe to be next the placenta, i.e. axial, the carpels being 

 two ;— the signs in the diagram therefore which is there inserted require only reversing to show the true position. The diagram gives a correct view of the dichotomous mode of the branching of the inflorescence. 



™ the ° nl7 genUS in Di P sac ™ ™ whi <* 1 h ave observed any carpels posterior, or rather lateral and internal, which I regard as equivalent to posterior, (V. Ann. Nat. Hist, 2nd Ser. Vol. XI.) and this occurs so rarely, perhaps 

 m ot 40 or 50 flowers, that it does not seem to require that Dipsacece should be on that account removed from their natural station, even if the carpel were not always anterior in other genera. 



