on Haastia pulvinaris. In every habitat there is considerable variation in the dimensions of the leaves of neighbouring plants; yet the plants of any one area may be described as having longer, shorter, broader, or narrower leaves than those of some other area, and in consequence might be regarded as constituting distinct jordanons. Nevertheless, it is possible to select individuals from each quite indistinguishable as hand-specimens. This would seem to support the view that the species of the systematist develop from geographically segregated portions of a primary hybrid swarm and its derivatives.

Furthermore, careful search amongst the individual plants of C. Allanii var. canescens growing on the St. Arnaud Mountains enabled me to isolate individuals indistinguishable from selected individuals of C. incana var. nivalis in the fresh or growing condition. When dried, however, they were readily separable, for the tomentum of the former plant on the upper leaf-surface turns brown and adheres to the epidermis.

6. Celmisia Allanii var. canescens.

Celmisia Allanii is a new species created for the reception of the plant excluded by me from Hooker's conception of Celmisia Sinclairii, which he based on specimens of two distinct species obtained from two widely separated localities. I have selected as the type a plant from Mount Trovatore which Dr. H. H. Allan, who has examined the Kew material, considers identical with that gathered on the Dun Mountain and forming portion of the type material on which Hooker founded his C. Sinclairii. Var. canescens differs from the typical form in its whiter tomentum and more tufted habit as well as in its rather shorter leaves with very evident midrib. When the leaf is dried the upper surface turns brown as though the tomentum had mysteriously disappeared. On the lower side the tomentum forms a snow-white, woolly felt save on the rather conspicuous midrib. In this variety, as in C. incana var. nivalis, considerable diversity of form and dimensions is observable between the various units of the jordanon, comparable with what one might anticipate in the progeny of a cross between individuals of two closely related jordanons; in other words, if this be regarded as a valid jordanon, the range of divergence from the mean is particularly wide. If such be the correct interpretation of the evidence, it seems clear that the original parental forms have now been absorbed in subsequent crossing. Further, since no two members of a hybrid swarm are identical, it follows that the progeny of isolated portions of such a swarm cannot give rise to identical, but only to similar jordanons. The field evidence fully supports this interpretation of the origin not only of this variety, but of every other species and variety of Celmisia we have investigated. Var. canescens is known to the writer only from the St. Arnaud Mountains, the Red Hills, and the southern end of the Wairau Mountains.

7. Celmisia Cockayniana.

The only plants of this species gathered hitherto seem to have been two or three specimens secured by the late Dr. L. Cockayne in 1892 from a rocky ledge near the summit of Mount Fyffe, and

now preserved in the Dominion Museum; though there is in the Canterbury Museum a leaf bearing the label C. Sinclairii (?) which clearly belongs to this very distinct species, and which was gathered on the Camden Range by Prof. A. Wall. Strangely enough, the species is quite common and widely distributed in Marlborough, growing as it does on both the Seaward and Inland Kaikouras and parallel ranges to the west, as far as Mount B.J. on the ridge separating the Waihopai River from the Avon River. The species is endemic to Marlborough and, indeed, to the North-eastern Botanical District, but for a slight overlap into the North-western at Mount Schiza, and occurs always on rock at elevations of from 4000 to 5000ft. As this has hitherto been regarded as a very rare plant, I give the list of habitats where I have so far collected it, viz., Mounts B.J., Bounds, Pinnacle, Schiza, Camden, Malvern, Gladstone, Tapuaenuku, Kaitarau, Whakari, Fyffe, and Kahutara Saddle where it is abundant.

8. Celmisia hieracifolia.

Celmisia hieracifolia var. oblonga.

This abundant species is the analogue in the Sounds-Nelson portion of Marlborough of C. Cockayniana in the North-eastern District. On Mount Dobson it grows in soil pockets on rock faces as low as 2000ft., but in general it is commonest between 4000 and 5000ft. altitude. I have not seen it south of Mount Patriarch or east of the Wairau River. The smaller of the two forms on Mount Stokes is var. oblonga (T. Kirk, 1894, p. 329).

9. Celmisia Traversii.

This handsome species is met with in Marlborough only in the North-western District. It is quite common on the Bounds, the Raglans, and the St. Arnauds, though on the last-named mountains I have not seen it at the northern end. It is a plant of subalpine meadows and of soil-coated rocks at elevations exceeding 3500ft. in areas of high rainfall, and is seen to best advantage in stations not exposed to bright sunshine.

10. Celmisia cordatifolia.

Celmisia cordatifolia var. similis.

Celmisia cordatifolia var. Brockettii.

Celmisia cordatifolia was established by J. Buchanan (1878, pp. 427–8) for a cordate-leaved plant from Mount Starveall which was clothed on the scape and lower leaf-surface with rich ferruginous tomentum. In Marlborough this same plant is to be found on Mount Fishtail, Mount Rintoul, and Mount Patriarch, and it formerly grew on Mount Duppa, though I failed to locate it there two years ago. Two other very distinct jordanons of this species have also been discovered, one on Mount Richmond and the other on Mount Bounds. The former differs from the type mainly in having a thin pellicle of silvery tomentum in place of the thick, red-brown, felty tomentum which is its most conspicuous character. The leaf is also thinner and more pointed, and when dry has strongly recurved margins. This variety has been distinguished as var. similis. Var. Brockettii is a much smaller plant discovered by me on Mount Bounds, and

combines the cordate leaf-form of C. cordatifolia with the characteristic type of tomentum seen on C. spectabilis. Several dozen plants, all of similar aspect, were noted in a small area, but no one seems to have met with this variety elsewhere. The location is at an altitude of 5000ft. on a ridge leading from the top of Mount Bounds to the left bank of the R. Gosling, a tributary of the Waihopai. I have named it in compliment to Mr A. E. Brockett, M.A., B.Com., who has been a constant companion on all my field excursions. All three varieties grow in places kept perpetually moist with percolating water, and all affect stations not directly exposed to the sun.

11. Celmisia Rutlandii.

This plant has been recorded only from Mount Stokes (T. Kirk, 1894, p. 329), but it is quite a common plant on all the higher peaks of the Wairau Mountains (e.g., Mounts Patriarch, Richmond, Fishtail, Royal Knob, Sunday, and Riley). Till recently I had regarded the Wairau Mountains form as being distinct from that on Mount Stokes, but garden culture and further field examination have convinced me of their identity. The provisional name of C. umbrosa used by me (1932, p. 10) in my “Vegetation of Marlborough” should therefore be dropped. This is characteristically a plant of shaded rocks and rock-chimneys at an elevation of between 4000 and 5000ft.

12. Celmisia spectabilis.

Celmisia spectabilis var. angustifolia.

Celmisia spectabilis var. albomarginata.

Celmisia spectabilis is a common Marlborough plant at elevations of 1000ft. and over, but on the foothills of the Wairau Mountains it is sometimes common as low as 200ft. Until two years ago my field notes took no cognizance of the existence of distinct jordanons, but at high levels (4000 to 6000ft) in the “North-western” part of Marlborough the var. albomarginata almost or wholly replaces the typical form, and the same thing probably holds for the Sounds-Nelson District also. This differs in having somewhat larger and commonly more pointed leaves with a whitish pellicle on the upper surface margined with a rim of denser furry tomentum forming a conspicuous white border, whereas the typical form is glabrous above. Var. angustifolia is likewise glabrous on the upper leaf-surface, but differs from both these other forms in the smaller and relatively much narrower leaves. This variety is practically restricted to the North-western District, though it overlaps on to Mount Patriarch, the southernmost peak of the Wairau Mountains, in the Sounds-Nelson area. It is a plant of herbfield on the Raglan, St. Arnaud, and Bounds Ranges. In the North-eastern District, only the type form is present unless a stunted plant growing in the Ure Basin proves to be a distinct jordanon; I think, however, that it is more likely to prove an epharmone.

13. Celmisia viscosa.

Nowhere in the North-eastern District have I observed this species, but on the wetter Wairau Mountains (e.g., Mount Fishtail or Mount Richmond) it forms extensive colonies at 5000ft. altitude.

On the Raglan Range near the gorge of the Wairau River and on the mountains near the Rainbow River in the North-west District this species is again met with.^* That it should be apparently missing from the Seaward Kaikoura Mountains seems surprising. C. viscosa is wholly confined to fellfield or to the veneer of soil overlying flat, alpine rock-surfaces.

14. Celmisia coriacea.

Celmisia coriacea var. lancifolia.

There are no plants attributable to C. coriacea in the Sounds-Nelson District, while in the North-eastern District it has been observed by me only on the Seaward Kaikoura Range. On the other hand, it occurs commonly throughout the greater part of the North-western District. The plant listed by me as C. Monroi var. conspicua might well have been considered a variety of C. coriacea. Var. lancifolia, though recorded by Cheeseman (1926, p. 950) from the mountains flanking the Wairau and Clarence Valleys, has so far eluded my notice in Marlborough.

15. Celmisia Monroi.

Celmisia Monroi var. robusta.

Celmisia Monroi var. conspicua.

Dr. L. Cockayne (1916, p. 194) has drawn attention to the fact that there is no true type for C. Monroi, as Hooker's original description was based on plants from two widely separated localities and belonging to more than one species. By the removal of C. Boweana Petrie (1911, p. 18) from the conception of this species, the Awatere plants from Upton Downs have been left as the virtual type of C. Monroi.

The narrow-leaved race referred to by Dr. L. Cockayne (1916, p. 194) is probably identical with that growing on the summit-rocks of Mounts B.J. and Ferny Gair. This I have had under observation for some time, and in my garden a measure of convergence has taken place between this and the typical form; the plants are, however, still amply distinct, and I anticipate that the high-level form may prove to be a good jordanon distinct from the three listed, but I refrain from according it varietal rank. Side by side with these two forms we have grown plants from 4500ft. on the St. Arnauds and Bounds which represent a much more robust and wider-leaved variety. Seedlings from these areas have also been raised and have developed the distinctive characters of the variety when grown side by side with plants from Upton Downs. I have designated this variety var. robusta.

C. Monroi var. conspicua comes very close to var. robusta, especially in the characters of the leaf, but may readily be distinguished by the large size of the bracts, especially of those at the base of the scape. It is a plant of the lowlands and is restricted to the limestone area between the coast and the mountains forming the northern extension of the Inland Kaikouras from the Clarence River north to Lake Grassmere.

16. Celmisia insignis.

This very distinct species has not previously been collected. It is closely allied to C. Monroi, of which it might be regarded as an exceptionally narrow-leaved variety. Even in large specimens where the leaves reach a length of a foot or more, the width rarely attains to a quarter of an inch, and commonly is only an eighth of an inch. The plant is usually silvery-grey in colour, though in winter it may often be quite white. It ranges from the Awatere on the east to the Wairau on the west, and from the Ned in the north to the Wye in the south. I have it from the Ned, Ward Peak, Omaka Basin, Ferny Gair, Mount Harkness, Mount Blairich, and from the valley of the Gosling in the Waihopai Basin. It is to be found in soil pockets on a rock-face, or growing in clay derived from rotten rock, usually on the bank of a stream. More rarely it occurs as a true chasmophyte. Its altitudinal range is from 1000–4500ft.

17. Celmisia gracilenta.

Celmisia graminifolia.

Celmisia alpina.

C. gracilenta was first described from specimens collected near Admiralty Bay in the Marlborough Sounds, where it still grows, but throughout the province it is undoubtedly less common than it was in the days before the introduction of stock, game-animals, and agricultural implements. It shows very little epharmonic variation in its response to the water-content of the soil, plants from wet swampy soils being very similar to those from nearby, open, wind-swept pasture. C. graminifolia, as already pointed out, shows considerable variation, but neither in nature nor in cultivation does it show any approach to C. gracilenta on the one hand or to C. alpina on the other. All our field evidence and all observation of these three plants in cultivation strongly supports the view that each is a valid species. Indeed, C. alpina, which is commonest in Marlborough as a plant of fellfield rather than of bog, is another species which has a very constant form in all its Marlborough habitats. Whether the bog plant described in the Manual, with its small leaves of 1–2 inches, and the Marlborough shingle plant of 3–5 inches are in reality identical has not been investigated, but a form on the Dun Mountain just beyond the confines of Marlborough appears to be distinct. C. alpina is a plant common on fellfield at an altitude of 5000–6000ft., but is as a rule confined to stations where the ground is nearly level and the shingle is very fine. C. graminifolia is common as a swamp plant near the head of the Kahutara Saddle. The form on the Okaramio Saddle may belong to a distinct jordanon.

18. Celmisia laricifolia.

This species is present on all the higher mountains of Marlborough at elevations of 5000ft. and over.

19. Celmisia Macmahoni.

Celmisia Macmahoni var. Hadfieldii.

The typical form of this species is apparently restricted to Mount Stokes, the highest point in the Marlborough Sounds, and even there it is far from common. It forms compact, silvery cushions