parva, imperfecta. Ovarium ovato-oblongum, obtuse trigonum, stigmate trilobo sessili terminatum, uniloeulare. Placentæ parietales, triseriales, nerviformes. Ovula plurima, biserialia, semi-anatropa, ascendentia. Bacca carnosa, oblongo-elongata, ⅓ unc. longa, prismatica, angulis obtusis, pallide rufa, perianthio duplo longior. Semina plurima, horizontalia, obovata, obtusa, latere unico compressa, subcarinata; funiculus crassiusculus, subelongatus, filamentosus; membrana exterior crustacea, subossea, aterrima, nitida; nucleus pendulus solutus, chalaza orbiculari apiceque fusca; membrana interior tenuis, pallide fusca. Albumen carnosum. Embryo parvus, ovatus, basi albuminis inclusus; radicula hilo proxima.
Since the first appearance of the 'Prodromus Floræ; Nov. Holl.,' in which Mr. Brown published the genus Astelia from the manuscripts of Banks and Solander, nothing seems to have been done by the many authors who have transcribed his characters and remarks towards determining its affinities. Mr. Brown himself views it as intermediate between Asphodeleæ and Junceæ, and retains it at the end of the former order; from this it has been removed with one consent by all future classifiers, some placing it after Junceæ and others with Melanthaceæ. Mr. Forster's name of Melanthium pumilum, given to the Fuegian species, shows that he considered it as being most nearly allied to the latter-named order; but I am not aware that any other author has stated his reasons for following Forster's views of its affinity, except perhaps Thunberg, whose dissertation de Melanthaceis I have never seen.
In 1819 Sir J. E. Smith (Appendix to Rees's Cyclopædia) added a new species to the genus, the A. Menziesiana of the Sandwich Islands; the form of the seeds is mentioned, but no particulars of their structure. This species was redescribed by Gaudichaud (Voy. Freyc. Bot. p. 420), who does not seem to have been aware of Sir James Smith's paper, and he named it A. veratroides, placing it in Melanthaceæ without any remark; the fruit seems unknown to Gaudichaud, but was described as three-celled by Smith and again by Hooker and Arnott (Bot. Beechey Voy. p. 97), who also retain it in the same order.
A. Richard published his 'Flora Novæ Zelandiæ' in 1830, wherein no notice is taken of the genus Astelia, but a species of it is figured and described as Hamelinia (nov. gen.) veratroides: the male flowers and ripe fruit appear to have been both unknown to that author, the ovary is described as trilocular, and the genus arranged in Colchicaceæ (Melanthaceæ).
In 1836 [[Author:Allan Cunningham (1791-1839)|Mr. Cunningham[[ described (in his Prodr. Flor. Nov. Zel.) two species of Astelia, under one of which {A. Banksii), Richard's Hamelinia is quoted as a synonym; it is placed in Junci. Shortly afterwards Endlicher (in his 'Genera Plantarum') removed it to the end of Junceæ. Lastly, Kunth takes up the genus Astelia (Enum. Plant, vol. iii. p. 364) and follows Endlicher's views of its affinity. Though however his work was published as late as 1841, all notice of Cunningham's species are omitted, and the A. Banksii receives the third name of A. Richardi.
I have had the opportunity of examining the ripe fruit of six species of the genus, and find the seeds of all to agree in structure and to partake of the peculiarities both of Liliaceæ and Junceæ with Melanthaceæ they have fewer characters in common.
Except in the more fleshy substance of the capsule in most of the species and its not bursting by valves, to which however there is a manifest tendency in the A. pumila, there is no material difference in that organ between Astelia and Juncus. The ovaria are the same in both, being one or more celled; when one-celled generally bearing the anatropous ovules in two series on three lines of parietal placentæ, and when three-celled they are pendent from the inner angles of the cells. The internal structure of the seeds of the two genera is identical; the outer coat alone, from becoming thick and even in Astelia, agrees only with Rostkovia amongst Junceæ, but in being crustaceous and black differs from all. The nucleus, raphe, chalaza, inner coat of the seed immediately surrounding the albumen, the albumen itself, and form and position of the embryo, are precisely alike in both. Together with these remarkable accordances in structure there are many points of discrepancy, especially in habit, as also in the chaffy covering of the leaves and other parts, the uniformly diœcious or poly-
