gamous flowers, the texture of the leaflets of the perianth and their being often united above the base, the subulate and generally terete, not linear or compressed filaments, the short anthers and different nature of the pollen, which is minutely granular and furrowed on one side, and the constantly nearly sessile and three-lobed stigma.
With Asphodeleœ, Astelia has many points in common: though, as stated above, the internal structure of the seed is more manifestly that of Junceæ, it is not opposed to the description of that organ in Asphodeleæ, and the brittle black shining testa is almost typical of that order. The stamens are the same, in some Asteliæ being much elongated and bearing versatile anthers; and the form of the pollen likewise, though I have always observed it to be granulated. The thickened and subincurved apices of the leaflets of the perianth are analogous to the cucullate tips of these organs in Drimia and Albuca. The form of the perianth is variable in the New Zealand species, nor is its texture always peculiarly semi-glumaceous, as in one species the laciniæ, which are linear, are also thin, membranous and white; in another it expands at the base into a broad, flattened or cup-shaped disc with six equal lobes; in a third the base is subcampanulate, with erect linear laciniæ, subpetaloid in texture; and in a fourth species it completely surrounds the ripe berries, which are very large. Nor is the berry itself very constant in internal structure, as in Dianella, Lam., amongst Asphodeleæ, it is baccate, and as in Allium, L., it varies in the number of cells from one to three. In one of the above New Zealand species it is, as far as I can judge from dried specimens, one-celled, with many seeds pendulous from the summit, attached by short funiculi to what probably was a fleshy pendent column, but of which I see only the membranous remains; if any dissepiments existed they must have been very imperfect. Another species has the berry constantly three-celled, with several seeds pendulous from the upper inner angle of the cell; and in a third the fruit is membranous and subcapsular with three to six cells, each containing two or more pendulous seeds, which are convex at the back and with the sides much compressed like those of many Asphodeleæ. To Melanthaceæ the genus is allied inhabit, in the polygamous flowers, in the perianth sometimes (exactly similar to that of Wurmbia, Thunb.) forming a tube round the fruit, in the baccate fruit, and in the form and surface of the pollen, which in some Melanthaceæ is minutely granular: from this order however they essentially differ, in the æstivation of the perianth not being induplicate, in the anthers being at no period extrorse, in the single style, sessile, lobed stigma, and in the crustaceous integument of the seed. Whilst alluding to this order I may mention another plant whose affinity has been considered dubious, the Campynema linearis, Lab. (Flor. Nov. Holl. vol. i. p. 93. t. 121). This I have lately examined, and find it to be, as Mr. Brown rightly conjectured (Prodr. p. 290), truly Melanthaceous, with the tube of the perianth united to the ovarium. The fruits I have only seen in an immature state; but in them the adhesion of the perianth to the capsule is evident, and in a forwarder state the line of separation would doubtless be more clear. The plant is diœcious or probably polygamous, the anthers extrorse and caducous, the filaments after their falling away becoming recurved and projecting between the segments of the perianth; the pollen is yellow and granulate. The immature seeds are very numerous, imbricated in two series in each cell, and are attached to the middle of the dissepiments.
I have not ventured to subdivide the genus Astelia, as I doubt if characters of sufficient importance will be found to render it necessary, especially until good specimens in all states of the New Zealand species shall have been examined. The A. pumila, Br., is the most abnormal species in habit and in the subcapsular fruit; it is allied to the Tasmanian A. alpina, Br. in the form of that organ and shape of the leaves, and, on the other hand, to A. linearis in the short two-flowered scape. The A. alpina again, having a racemose inflorescence, connects these with the New Zealand species, in one of which the ovarium is one-celled. There are probably two species in the Sandwich group, one of which has two seeds in each of the cells of the berry and the seed-coat is very thick and osseous.
The A. linearis is the most inconspicuous of any of the species, owing to the grassy appearance of the leaves and its small size. Both the A. pumila and A. alpina are very striking plants; the former constituting singular hard flat green beds on the bogs, often several yards across; while the latter, with its beautifully silky and copious foliage, is a great ornament to the top of Mount Wellington and other mountains of Tasmania.
