Page:The Botany of the Antarctic Voyage.djvu/305

Falklands, etc.] the tropics in the East Indies, and skirting the Arctic circle both in Europe, Asia, and America. In the south, besides inhabiting all the Antarctic Islands, it grows in New Zealand and Tasmania.

4. GUNNERA, L.

1. Gunjjeea Chilensis, Lamk., Enc. Meth. vol. iii. p. 61. IUi/st. t. 801. f. 1. Broivn et Bennett, Plant. Jav. Ear. pt. 1. p. 70. G. scabra, Ruiz et Pav. Fl. Peruv. vol. i. p. 29. t. 44. f. a. Kunth Nov. Gen. Am. vol. ii. p. 35. "Panke," Feuill. 06s. ii. p. 741. t, 30.

Hab. Clionos Archipelago ; C. Darwin, Esq.

Apparently the southern limit of a plant which is found along the whole eastern side of South America, from Caraccas, in lat. 10° N., whence we have specimens gathered by Mr. Purdie, as far south as the 46th degree.

After the elaborate and learned essay upon this genus by Mi -. Bennett, in the 'Plantse rariores Javse,' I have little to remark upon its history or structure, except that the embryo is very minute, heart-shaped, and placed at the opposite extremity of the seed from the hilum, towards which the cotyledons point. The albumen is surrounded with a delicate testa and attached by a very short funiculus to the osseous putamen, which (as Mr. Bennett rightly concludes) is derived from the inner coat of the ovarium, and not, as Blume supposes, from the outer coat of the seed.

Some years ago, after referring a Tasmanian genus to Haloragea, Mr. Brown had the kindness to direct my attention to Gunnera, a plant closely allied to the one I was then examining; this led to the remark contained under the description of Milligania in the ' Icones Plantarum ' (t. ccxcix.) and the latter, probably, to Endlicher's removal of Gunnera from Urticea. The correctness of this view of their affinity admits of no doubt, although the alternate leaves separate Gunnera from all the genera of this order known to me.

The more obvious points of affinity between Gunnera and the Haloragea proper, are the frequently unisexual flowers, the quaternary arrangement of their parts, the adherent tube of the calyx, the great similarity between the two petals of Gunnera and Meionectes, the form of the stamina and pollen-grains, the styles covered throughout their length with stigmatic papillae, and the solitary pendulous albuminous seed. I may add the rigid and more or less scabrid foliage, which is so conspicuous in Haloragis, the racemed and often pendidous flowers, and the frequently long recurved styles.

The Gunnera differ remarkably in having their leaves, as I mentioned above, alternate ; the ovaria, though furnished with two styles, are one-celled, with a solitary ovule; and the embryo, instead of being cylindrical and axile, is very minute and placed at the opposite extremity of the seed from the liilum, and it is also inverted, with the radicle turned away from the hilum. The stamens in Gunnera are opposite the petals, and so are two of those of Meionectes. There is a tendency to irregularity in the form of the ovarium and its investing calyx, observable in some of the plants of this group, and most evident in the following species and in Milligania, where four unequal teeth of the calyx are developed, this and the presence of two styles indicate that the ovarium is probably two-celled at a very early period, one of which cells is suppressed. Lastly, in Milligania, a more intimate affinity is observable between Gunnera and Haloragea, for there are frequently in that genus four evident styles united at the base into two, indicating a normally four-celled ovarium, or one formed of four carpellary leaves, placed Kke those of Callitriche, in pairs, but so intimately united as to appear more like the truly simple ovary of Hippuris.

Next to Haloragea, C/ilorantltea is the order with which tins genus has most in common, particularly through the Sandwich Island genus Ascarina, of Forster, where the flowers are spiked or racemed and unisexual, the male consisting of a solitary linear anther, sessile in the axil of a toothed bractea, and the female, when ripe, of a one-celled drupe, very like that of Gunnera, surmounted by a sessile obscurely 3-lobed stigma. The seed is compressed, pendulous from the apex of the ceD, covered with a delicate membranous testa ; the albumen copious and oily,