310 OBSERVATIONS ON THE NATURAL FAMILY
It will be attended with similar advantage to form a separate family of
Erunonja,
as a link of equal importance, connecting Composita? with Goodenovice, but from both of which it is in many respects very distinct. As I have formerly described this genus, and made several observations on its principal affinities, 1 T shall here only state the more important relations and distinctions between it and those families to which it appears to me most nearly to approach.
Brunonia agrees with Goodenovia in the remarkable indusium of the stigma ; in the structure and connexion of 133] the antherae; in the seed being erect; and essentially in the aestivation of corolla. ' It differs from them in having both calyx and corolla distinct from the ovarium ; in the disposition of vessels in the corolla ; in the filaments being jointed at top ; in the seed being without albumen ; and in its remarkable inflorescence, compatible, indeed, with the nature of the irregularity in the corolla of Goodenovia, but which can hardly coexist with that characterizing Lobel'iacece?
With Composites it agrees essentially in inflorescence ; in the aestivation of corolla ; in the remarkable joint or change of texture in the apex of its filaments ; and in the struc- ture of the ovarium and seed. It differs from them in having ovarium liber am or super urn ; in the want of a glan- dular disk ; in the immediately hypogynous insertion of the filaments ; in the indusium of the stigma ; and in the vascular structure of the corolla, whose tube has five nerves only, and these continued through the axes of the laciniae, either terminating simply (as is at least frequently the case in Brunonia sericea), or (as in B. australis) dividing at top into two recurrent branches forming lateral nerves, at first sight resembling those of Compositae, but which hardly reach to the base of the laciniae.
It is a curious circumstance that Brunonia should so
» Prod. Flor. Nov. Hull. p. 589.
2 See Flinders's Vojage to Terra Aubiralis, ii. p. 55'J [col. i, p. 32].
�� �
