portion of this surface. Plants from Kekerangu growing on a heavy clay-soil in Mr J. Scott Thomson's garden at Dunedin in shady, moist conditions, when seen by me in May, 1933, had lost all trace of tomentum on the upper leaf-surface, and in consequence were so altered in appearance that I failed to recognise the Kekerangu plant with which I was so familiar in Marlborough. On the other hand, all plants grown by me in Blenheim developed pure white leaves in winter. In summer they displayed as much green as white. On the marl cliffs of Clifford Bay flowering specimens sometimes have leaves only 5–6 inches long, while in Woodside Gorge shade-epharmones may have leaves over 2 feet long and proportionately wide. It is evident that the development of tomentum on the top leaf-surface is proportional to the xerophytic conditions of the habitat.

4. Celmisia Sinclairii.

In a separate paper not yet published I have restricted this species to the glabrous plants described by Hooker from Tarndale, and have referred the tomentose Dun Mountain plants to C. Allanii.^* The lectotype at Kew appears to represent a shade-epharmone, which I have been able to match with plants gathered by me on Mount Schiza. These shade plants have flat, membraneous leaves which, unless the drying process is very rapid, dry to a black-brown. The normal form, however, has thicker leaves with recurved margins and a somewhat rugose lower surface, and on drying retains the green colour. I had long known each of these two forms before I realized they were epharmones, such as garden culture proved them to be. Subsequent search at Mount Schiza confirmed our conclusions, for the papery-leaved plants were found only in very damp, shaded stations, or growing up to the light through a scrub of Podocarpus nivalis, Phyllocladus alpina, Hebe rupicola, Aristotelia fruticosa, etc. Some of these plants had developed a semi-lianoid habit with stems up to 3 feet in length.

5. Celmisia graminifolia.

Near the top of a low saddle between Okaramio and the left branch of the Waikakaho Valley, at a height of 1000ft., this species occupies a small area, part of which is covered with marginal forest-scrub and part of which is open ground covered till recent years with “fern” which had been burnt off. The plants growing on each portion bore no superficial resemblance one to the other. In the open, each plant had wide, nearly oblong leaves 3–4 inches long and of a dark bronze colour when seen in the month of June. On the other hand, the plants growing in the open scrub had grass-like leaves up to 12 inches long, green in colour, and under ¼ inch in breadth. Such plants had a length-to-breadth ratio four times that of plants growing in the open. Both forms flower freely. In cultivation these plants show no convergence towards other members of the “longifolia” group. I have not grown them from seed, but all other available evidence points to the belief that C. graminifolia is a very distinct species.