the North Island, forwarded to me a single specimen of a plant found by him at Opepe, near Lake Taupo, which from its habit and general appearance I at once provisionally referred to the Burmanniaceæ. Positive identification, however, was quite impossible, as the whole of the interior of the flower, including the stamens, upper part of the ovary, and the style and stigma, had been removed through the attacks of some insect. The discovery induced me to make a special journey to Lake Taupo, in which I was accompanied by Mr. Hill; but, although a long search was made in the locality where the first specimen had been obtained, another damaged specimen was all that could be found. In January, 1905, I made another search for the plant at the southern end of Lake Taupo, but entirely without success. In January, 1907, Mr. Hill paid another visit to Opepe, accompanied by Mr. A. Hamilton, of Wellington. On this occasion he was fortunate enough to find a considerable number of specimens in full flower, most of which he very kindly forwarded to me. An examination of these proved that the plant had been correctly referred to the Burmanniaceæ, and that it must be included in the subfamily Thismieæ, which differs from the rest of the Burmanniaceæ in possessing both the inner and outer whorls of stamens, and in the unusually large development of the anther-connective. It further appeared that there was little to separate Mr. Hill's plant from the genus Bagnisia, as amended by Engler in “Die Naturlichen Pflanzenfamilien,” (vol. ii., part 6, p. 48), where it is made to include both Bagnisia and Geomitra of Beccari (“Malesia,” vol. i, pp. 249—50, tt. 10—12), the only difference of importance being that in Bagnisia and Geomitra three of the perianth-segments are wanting, or reduced to mere rudiments; whereas in the New Zealand plant they are well developed, being quite half the length of the inner segments. In this respect it agrees with the genus Thismia, but differs in the inner perianth-segments being dilated and connivent at the tips, exactly as in Bagnisia and Geomitra. In the position of the stamens and the structure of the anther it agrees with Geomitra; in fact, the anther-connective of Geomitra episcopalis, as figured by Beccari, very closely resembles that of the New Zealand plant. If, therefore, Engler is to be followed in merging Geomitra with Bagnisia, it appears to me that the New Zealand plant should be placed in the latter genus. In a recent number of the “Kew Bulletin” I have therefore applied the name of Bagnisia Hillii to the plant. It gives me great pleasure to associate Mr. Hill's name with the species, as some slight recognition of the long-continued interest he has taken in New Zealand botany, and of his unwearied kindness in supplying both Mr. Colenso and myself with specimens of many interesting plants collected during his journeys in the interior of the North Island. The following description will enable the species to be recognised:—

Bagnisia (Geomitra) Hillii, Cheesem. in “Kew Bulletin,” 1908, p. 420.

A minute colourless saprophyte, perfectly smooth in all its parts; leaves wanting or reduced to minute scales. Rhizome creeping amongst humus at the base of tall forest-trees, sparingly or copiously branched, 5–10 cm. long, 1—1.5 mm. thick, fleshy, naked. Peduncles springing from the axils of minute fleshy bracts, 1-flowered, erect or curved, 0.5—1.5 cm. long; bracteoles 3—7, alternate, lanceolate or ovate-lanceolate, acute or acuminate, the upper gradually larger, 1—5 mm. long. Flowers solitary, terminal, large for the size of the plant, 1—1.7 cm. long, about 0.7 cm. broad, bright rose-pink when fresh. Perianth campanulate-lanterniform; tube obovateoblong, distinctly 6—12 costate; outer perianth-segments smaller, free,

linear or oblong, at first erect but ultimately spreading, sometimes abruptly recurved. Inner segments almost twice the length of the outer, linearspathulate to obovate-oblong, connate or connivent at the tips, gaping in the middle, keeled on the back, with the keel produced into a subulate point at the tip. Stamens 6, affixed to the throat of the perianth-tube, shortly exserted, abruptly deflexed within the tube; filaments very short, free; anther-connectives much enlarged and expanded, connate into a membranous tube which is bilamellate at the apex; anther-cells small, distinct, towards the base of the tube formed by the connectives. Ovary inferior, broadly obovoid, 1-celled; placentas 3, free; ovules very numerous; style short, thick; stigma 3-lobed, the lobes broad, almost quadrate, concave, truncate at the tip. Fruit unknown.

Hab. — Primeval woods at Opepe, near Lake Taupo; alt., 2,000 ft.; H. Hill. Flowers in January.

Bagnisia Hillii is usually found on the mound of decaying leaves and humus which accumulates at the base of the trunk of the kahikatea pine (Podocarpus dacrydioides). On account of its small size it is easily overlooked, even in the flowering season, the flowers being often partially concealed by fallen leaves. Once noticed, however, the bright rose-pink colour of the flowers enables the observer to pick it with ease. Like many of the Burmanniaceæ, the flower has a most bizarre appearance. The three inner perianth-segments, widely separated in the middle, but closing together and connivent at the tips, give it somewhat the appearance of a bishop's mitre, or perhaps of a lantern with three elliptical windows or openings. In the bud, or in the newly expanded flower, the three outer segments partly close these openings, but they gradually spread outwards, and ultimately (judging from specimens in formalin) become sharply reflexed. The structure of the anthers is most peculiar, and deserves careful study. As stated in the description, the connectives are enormously enlarged, and are connate into a broad membranous tube, which, owing to the curious manner in which the stamens are deflexed, lies parallel with the inside of the perianth-tube, the tips of the connectives pointing to the base of the flower, and the minute anthers opening into the narrow space between the connective-tube and the wall of the perianth-tube. The connective-tube is split into two delicate lamellæ at the apex, and the outer lamella of each connective is 2-lobed, with a slender bristle arising from the bottom of the sinus. On the commissure of each connective, but on the inner face of the lamella, is a small oblong gland. The inner lamella is very delicate, and has an undulate margin fringed with delicate cilia.

It is difficult to see how pollination is effected, seeing that the minute anthers, which only produce a small quantity of pollen, are hidden away at the back of the connective-tube, in a narrow passage to which access can only be obtained through the small openings between the short filaments, just at the entrance of the flower, or by crawling up behind the connective-tube from the base of the flower. It is significant that many flowers are found partly eaten by insects. Can it be that the conspicuous coloration of the flower and its juicy texture attract minute insects, which, while destroying certain portions of it, pollinate the short stigma at the base of the flower? It is much to be desired that some local observer would work out the fertilisation of the plant in detail.

Considerable attention has been paid to the Burmanniaceæ of late years, the American species in particular having been worked up by Warming,