Art. XVII.—Notes on New Zealand Floristic Botany, including Descriptions of New-Species, & c. (No. 3). By L. Cockayne, Ph.D., F.L.S., F.R.S., Hutton and Hector Memorial Medallist. [Read before the Wellington Philosophical Society, 24th October, 1917; received by Editors, 31st December, 1917; issued separately, 30th May, 1918.] Plates IX, X. I. Introduction. In this series of papers, two of which have already appeared,* Trans. N.Z. Inst., vol. 48, pp. 193–202, 1916; and ibid., vol. 49, pp. 56–65, 1917. I am carrying out, as far as lies in my power, the views regarding species and taxonomic procedure expressed in my paper entitled “A Consideration of the Terms ‘Species’ and ‘Variety’ as used in Botany, with Special Reference to the Flora of New Zealand.”† Trans. N.Z. Inst., vol. 49, pp. 66–79. These views are by no means of my own formulating. On the contrary, they represent what I believe to be the consensus of opinion of those engaged in the only true way of studying specific distinctness—i.e., by means of experiments in genetics according to present-day methods. If my views possess any originality it lies in the method of stating the case and in the proposals suggested for meeting the practical difficulty of making a flora serve its primary purpose of enabling any plant to be readily recognized and accorded its proper name. As the length of the paper cited above and the method of presentation of its arguments may serve to somewhat becloud the practical application of the theories advocated therein, I now briefly state the principles which in this series of papers are the guide for the establishing of species or varieties:— (1.) The starting-point in the setting-up of species is the individual. (2.) Groups of individuals which resemble one another in every character and reproduce their like, subject, of course, to unfixed fluctuating variations, constitute specific units and may be designated “microspecies.”‡ Other names given to such groups are “biotypes,” “petites espéces,” and “elementary species.” (3.) One microspecies, if all its related microspecies have been obliterated or have never existed, constitutes an invariable or fixed systematic species. Examples: Agathis australis, Veronica cupressoides, Epilobium pallidiflorum.§ It might well be argued that there are no species which consist of only one microspecies, and that intense study and experiment will demonstrate their polymorphy. (4.) Two or more closely related microspecies may be united into a group for the sake of: (a) convenience in identification, (b) emphasizing the close relationship of minor groups (microspecies), (c) phytogeography. (5.) Such a major group as constituted in (4) forms an aggregate or collective species. (6.) Aggregate species are the “variable species” of floras. Examples: Poa anceps, Ranunculus lappaceus, Pimelea prostrata. (7.) An aggregate species has obviously no real existence; it is a convenient abstraction only.

(8.) Each microspecies of the combination forming an aggregate species should theoretically receive a varietal name. (9.) But in practice the procedure advocated in (8) would defeat its purpose if the microspecies were too much alike, so in this case groups of virtually identical microspecies can receive varietal rank. (10.) It follows then that, similarly with species, varieties are of two kinds, one reproducing itself true and the other an aggregate. (11.) Aggregate varieties, though abstractions only, so far as the eye goes approximate to true entities. (12.) The description of an aggregate species applies to no special individual, but includes the striking characters common to all its varieties; obviously, then, there is no “type.” (13.) If the opinions as given above are accepted, a trinomial nomenclature becomes necessary, the first name being that of the genus, the second that of the species, and the third that of the variety. (14.) If the opinion is held that every microspecies has been at one time related closely to other microspecies, it follows that even the invariable species mentioned in (3) should be given varietal names. But this procedure seems unnecessary, and perhaps mischievous, since a binomial for such species is convenient and it indicates that the specific group stands apart from all others.* Of course, as at present accepted, there are many different degrees of specific isolation, but it should be possible to gradually bring about greater uniformity in this regard. (15.) In certain cases groups, otherwise well defined, seem to be united by “intermediates” which cannot be joined to such groups or made into one or more species. Such “intermediates,” according to the teachings of genetics, may be assumed to be hybrids between microspecies, and their occurrence should not forbid the separation into species or varieties, as the case may be, of the distinct true-breeding groups (microspecies) which are connected by such presumably hybrid intermediates.† This has frequently been done in the New Zealand flora, but not because of any special biological explanation such as that of microspecific hybrids. Celmisia discolor and C. incana (Manual, pp. 304–5), Gnaphalium Lyallii and G. trinerve (Manual, p. 323), and many species of Veronica are cases in point. On the other hand, distinct microspecies are denied specific rank owing to their being connected by “intermediates.” Examples are: Epliobium pedunculare reduced to a var. of E. nummularifolium (Manual, p. 180), the treatment of the groups included under Hoheria populnea (Manual, p. 79), and the retention of vars. robusta, minor, and lanata as varieties of Craspedia uniflora (Manual, p. 348). As a botanical ecologist, endeavouring to define and classify the plant-communities of New Zealand and to learn something about the physiological requirements of the species and the physiology of form, I have keenly felt, for many years, the want of names for many well-marked groups of individuals which, though fitting fairly well into one or other of the recognized aggregate species, differ so greatly in their ecological requirements from other members of the species to which they are referred that to call them by the same name is most misleading, and in no few instances will cause incorrect ecological deductions.‡ Of what value would be an account of the leaf-anatomy or the rate of transpiration in the leaves of certain individuals of Pittosporum tenuifolium, Acaena Sanguisorbae, Aristotelia fruticosa, Geranium sessiliflorum, Celmisia coriacea, and Myosotis antarctica under the above specific names, unless a description of the actual plants dealt with were given—i.e., unless they were accorded for the time being the status of microspecies?

It may well be argued that the trend of botanical taxonomy the world over is to bestow specific names on the varieties, thus breaking up the long-recognized aggregates into so-called “valid species.” Certainly such groups distinguished by binomials are convenient for the ecologist working at synecology, but they are of far less use to the autecologist, the floristic phytogeographer, or the student of evolution or genetics than are aggregates with their varieties distinguished by trinomials. Once cease to emphasize the genetic aspect of taxonomy, away goes its philosophy—indeed, it ceases to be a science! To apply the principles enumerated above is far from easy; they probably represent ideals impossible of full attainment. Research is demanded in many directions; above all, living material is essential—field observations as accurate as possible must be made, and experiments in the garden must finally decide those doubtful points impossible to be solved either in the field or the herbarium. With regard to the species, & c., dealt with in the present paper I have received valuable assistance from various sources without which the work could not have been carried on. I must especially thank Mr. H. H. Allan, M.A., F.L.S. (Ashburton); Mr. B. C. Aston, F.I.C. (Wellington); Mr. H. Carse (Kaiaka); Mr. C. E. Christensen (Hanmer); Miss E. M. Herriott, M.A. (Christchurch); the Rev. J. E. Holloway, D.Sc. (Hokitika); Mr. R. M. Lamg, M.A., B.Sc. (Christchurch); Messrs. Nairn and Son (Christ-church); Mr. D. Petrie, M.A. (Auckland); Mr. R. H. Rockel, M.A. (New Plymouth); Professor A. Wall, M.A. (Christchurch); and Mr. J. Young (Christchurch) — all of whom have given me much-valued aid both in material and information. I must also acknowledge the kindness of Professor H. B. Kirk, M.A. (Wellington), who has afforded me every facility for using the herbarium of the late Mr. T. Kirk, F.L.S.; of Dr. J. A. Thomson for similar privileges with regard to Colenso's herbarium in the Dominion Museum; and of Mr. A. Turnbull, F.L.S., who has allowed me to consult his splendid library of Australasian and Pacific literature. As far as possible I have deposited in the herbarium of the Canterbury Museum, Christchurch, type specimens of all the species, & c., dealt with in this series of papers. II. Taxonomic. 25.* The numbers follow on consecutively in this series of papers. Carmichaelia Fieldii Cockayne sp. nov. Frutex parvus, glaber, afoliatus nisi juventute, prostratus. Rami usque ad 40 cm. longi sed saepe multo breviores, 2 mm. lati, arcuati, compressi, striati, pauciramosi, cortice luteo-viride obtecti. Racemi brevissimi, nunquam fasciculati, 2–5 flori; pedicelli ± 3 mm. longi, glabri. Flores. non visi. Legumen 3–4 mm. longum, oblique-ovoideum vel-oblongum, quam maxime turgidum, subrugosum, nigrum; rostrum basi crassum, curvatum, apiculatum. Semina 2–5 (plerumque 3–4), pallide brunnea. South Island: North-western Botanical District—Growing as a small colony on a wind-swept sandstone ledge on a small island rising, at low water, out of the mud-flat of Westhaven (West Wanganui). W. H. Field and B. C. Aston! The above description is drawn up from insufficient material. In many cases the capsules were much damaged. Carmichaelia Fieldii appears to come nearest in affinity to C. juncea Col., but it differs in its prostrate habit, broader always more or less

compressed branchlets, shorter fewer-flowered, racemes, glabrous pedicels, and smaller dehiscent pod, with much stouter beak, which contains not 1–3 but 2–5 seeds. From Carmichaelia prona, the only purely prostrate species yet described, it is distinguished at once by its leafless adult branches, dehiscing pod with longer beak and greater number of seeds. Except for the dehiscence of the pod, the species under consideration would come into the subgenus Huttonella. The plant was discovered by Mr. W. H. Field, M.P., to whom it is dedicated. Mr. Aston was with Mr. Field at the time of the discovery, and he kindly handed over to me for publication the material he had collected, gave me two living plants for cultivation and further observation, and supplied the information given above regarding the habitat and habit of the species. 26. Carmichaelia grandiflora (Benth.) Hook. f. var. alba T. Kirk. The var. alba of Carmichaelia grandiflora was established in 1899 by T. Kirk to accommodate a plant which grows abundantly near the outskirts of subalpine Nothofagus forest in the neighbourhood of the junction of the White River and the main branch of the River Waimakariri, not far from their sources (Western Botanical District). There it has been collected by Kirk himself, Cheeseman, Wall, myself, and others. Cheeseman (Manual, p. 115) recognizes only the “type”—obviously a mixture—and var. divaricata T. Kirk, but (Illustrations of the New Zealand Flora, facing pl. 33) he writes regarding C. grandiflora, “It is an exceedingly variable plant. Mr. T. Kirk in his ‘Students' Flora,’ enumerates three varieties, and there are other distinct-looking forms. These varieties* It is not clear whether the author means only Kirk's published varieties or these together with the “other distinct-looking forms,” but I think the latter are meant to be included. differ in size, in the mode of branching, and in the size and shape of the pod. But before their systematic position can be properly understood they all require careful study and examination in the field.” From this it is evident that, according to Cheeseman, the “varieties” of C. grandiflora differ from one another in virtually all the essential characters used to define the species of Carmichaelia. Coming to var. alba, this is probably now accepted by Cheeseman as a variety, since in the Illustrations, when criticizing Kirk's remarks about its odour, Cheeseman writes, “Mr. T. Kirk in the ‘Students' Handbook’ says that the flowers ‘smell disgustingly of mice.’ But this peculiarity, so far as my own observations go, is only noticeable when the plant is being dried. In the fresh state the odour of the flowers is decidedly pleasant.” Since there is blossoming just now in my garden (30th December, 1917) a plant of the variety under consideration, collected for me last year by Professor A. Wall, M.A., from the original locality of the plant, I am in a position to add a few details about the variety from living material, which, unless being slightly less luxuriant and blooming more scantily, is essentially the same as if gathered from a wild plant. Kirk describes his var. alba as follows: “Branchlets more robust, compressed, deeply grooved, fastigiate or nearly so. Flowers as in the typical form, but white. Ripe pods not seen. Smells disgustingly of mice”; and he adds that it is “possibly a distinct species.” The plant in my garden is certainly not “fastigiate”; on the contrary, the branches are wide-spreading, being 60 cm. in their spread, while the shrub is but 30 cm. high. The branches are dark green, flattened, about

3.5 mm. diameter throughout for their final 15 cm. of length, grooved but not nearly to the same extent as in dried material, and more or less arcuate. The branchlets vary from about 4 cm. to 15 cm. in length; they are inserted on the flanks of the branch at an angle of about 30° and at about 2.5 cm. distance from one another. They are straight, bright yellowish-green, striate, flat, 3 cm. wide more or less, and almost uniform in width throughout. The leaves are numerous where sheltered, and then in fascicles of 2–4 at the base of young stems; elsewhere they are inserted in the notch of the stem at an angle similar to that of the insertion of the branchlet; the largest are about 1.8 cm. long, 3-foliate, their petiole 7 mm. long and channelled above; the leaflets are uniform in size, rather dull green, obcordatecuneate, their midrib sunken above but slightly keeled beneath; other leaves have similar characters, but they gradually decrease in size towards the tips of the branchlets until they become only 5 mm. long, or even less, and may consist of one leaflet only. In the cultivated plant the leaves are glabrous, but wild specimens show a few hairs on the under-surface, especially on the midrib. By all previous authors C. grandiflora is described as having glabrous leaves, but in all my herbarium specimens collected in various localities, including the classical habitat, Milford Sound, the under-surface of the leaf is more or less hairy, and sometimes considerably so. The flowers are white, except for a distinct pale-purple blotch through the median line of the standard, and honey-scented but rather cloying; they are in lax-flowered racemes, about 16 mm. long, furnished with short peduncles 4 mm. long or less. The calyx is campanulate and 3 mm. long; its tube is green or mottled pale purple, and the teeth are acute, small, pale purplish-brown, and ciliolate with white hairs. The standard slightly exceeds the keel, being 6 mm. long by 6 mm. broad; the wings and keel are equal in length (5.5 mm.). The above description corresponds, as far as changes through drying allow a comparison to be made, with that supplied by dried specimens. The var. alba may therefore be defined as follows: A wide-spreading shrub with the branchlets situated on the flanks of the stems, the racemes numerous, 4–6-flowered, the flowers white with a pale-purple blotch down the centre of the standard and sweet-scented, the standard as broad as long and rather longer than the keel, which equals the wings. Up till now C. grandiflora var. alba has been recorded from its one original station only. But that it has so restricted a distribution seems highly unlikely. It is more than likely that through taxonomists working mainly with dried material the colour of the flowers has been frequently overlooked, and that specimens are now included in herbaria along with the “type,” or other possible varieties, which may agree in colour with var. alba. How greatly colour has been neglected in diagnoses of species of Carmichaelia is demonstrated by the facts that Kirk mentions colour specifically in only four out of twenty-three species (Huttonella included) and that Cheeseman refers to colour in only seven out of nineteen species. 27. Carmichaelia juncea Col. ex Hook. f. (var. from Upper Clarence Valley). Carmichaelia juncea Col. was described in the first place by J. D. Hooker in the Flora Novae-Zelandiae, vol. 1, p. 51, from specimens collected by Colenso from “east coast, Hawke's Bay and Taupo.” In the Handbook Hooker referred plants from the East Cape (coll. Sinclair), from Akaroa (coll. Raoul), and from the Canterbury Plains (coll. Travers) to this species.

Petrie in his list of Otago plants (Trans. N.Z. Inst., vol. 28, p. 546, 1896) recorded C. juncea from various localities in what I now call the “North Otago Botanical District.” Shortly afterwards, Kirk in the Students' Flora accepted Petrie's determinations, and, as will be seen, enlarged Hooker's original conception of the species. Finally, Cheeseman, in the Manual, followed Kirk but gave a fuller description of the species than had been published up to that time. Both Kirk and Cheeseman agree in considering that the Otago plant may belong to an undescribed species, basing their opinions chiefly on the size of the pod and position of its beak. On the 9th December, 1917, Mr. Christensen noted a Carmichaelia growing on the bank of the River Clarence (North-eastern Botanical District), between the roads leading to Jack's and Jollie's Passes, which he described as “a bush 2 ft. to 3 ft. in height, with the branches drooping over the water.” He very kindly sent me specimens, one showing immature flowers (for the most part) and the other abundance of leaves. Towards the end of the month he again went in quest of fully opened flowers, and sent me a large living specimen fully in bloom, which is now growing in my garden. The above specimens I have been able to compare with Petrie's North Otago plant and with Colenso's type specimens of the species. Below I give a full description of the Clarence Valley plant. It appears to come into the conception of the C. juncea of Cheeseman's Manual, but it differs from both the type and the Otago plant in the racemes never being in fascicles, the glabrous calyx, and the much longer calyx-teeth. I have not yet seen the pod. As for the calyx-teeth, they are different in the three forms. In the type they are so small as to be almost wanting; in the Otago plant they are small but quite distinct, and broad at the base; in the Hanmer plant they are comparatively long and narrow. Other distinctions between the three forms may be noted on comparing the following description with Hooker's, Kirk's and Cheeseman's diagnoses. Until fruiting specimens are received, and perhaps comparison made with living material from Hawke's Bay and Otago, which I am hoping to secure, it seems best to let the matter of Carmichaelia juncea remain as Cheeseman has left it. But there seems little doubt that the species as at present constituted is an aggregate, with distinct varieties of restricted distribution. As regards the Akaroa and Canterbury Plains plant I know nothing. Description of Carmichaelia juncea var. from the Upper Clarence Valley. A low shrub 60–90 cm. high, with abundant slender drooping branches and numerous short racemes of small sweet-scented flowers. Branchlets numerous, close-set, passing from stem at a very narrow angle, bright green, compressed, but oldest branches terete, 2–3 mm. broad or rather broader but gradually tapering to an extremely narrow apex, usually leafless, glabrous. Leaves on younger branches, 1–3-foliate, ± 16 mm. long, petioles up to 6 mm. long; leaflets variable in shape and size, frequently oblong or ovate-oblong but occasionally obovate, linear, & c., retuse, bright green, glabrous above but somewhat hairy beneath with short appressed hairs, lateral leaflets much smallest, terminal ± 9 mm. long. Racemes solitary, apparently never fascicled, ± 6 mm. distant, 4–12-flowered but not dense, up to 12 mm. long; pedicels and rhachis

slightly pubescent especially when young, pedicel at most equalling the calyx, pale-coloured. Flowers minute, about 3.5–4 mm. long; calyx campanulate, glabrous, 2 mm. long (to tip of tooth), pale yellowish-cream dotted pale purple, teeth narrow, acute, rather long, dark purple; standard about 3 mm. long by 4 mm. broad, upper surface cream-coloured on lower half, above dark purple marked with almost black lines passing obliquely from median line of leaf, paler beneath, slightly exceeding the keel; wings cream-coloured more or less tinged yellow, equalling keel, oblong, rather narrow; keel near apex dark purple, beneath yellowish to cream-coloured. 28. Carmichaelia Monroi Hook. f. Two years ago Mr. B. C. Aston, F.I.C., while investigating the flora of the Clarence River basin, collected a supposedly undescribed species of Carmichaelia a description of which is given below. The examination of Aston's material led me to a comparison with that in my herbarium of C. Monroi Hook. f., with the result that the Clarence Valley plant may be the true C. Monroi, and that the forms included by Kirk, Cheeseman, and others under that name may be either one or two undescribed species or very distinct varieties of C. Monroi. This conclusion is quite unexpected, for few species in the flora seemed to be better understood than is C. Monroi. Description of Aston's Clarence Valley Species of Carmichaelia. A stout rupestral much-branched shrub with spreading, more or less drooping, leafless branches, up to 60 cm. long. Branchlets rigid, green, flat, grooved, ± 12 cm. long and up to 9 mm. wide, hoary pubescent when young with short appressed hairs which also extend at times to the older branchlets. Racemes ± 5 cm. long, frequently 7-flowered; rhachis and pedicels densely pubescent with appressed white hairs; pedicels slender, about 6 mm. long. Flowers about 10 mm. long. Calyx campanulate, densely hairy, 5 mm. long; teeth 3 mm. long, narrow-triangular, acute, standard rather longer than the keel, 10 mm. long, 7 mm. broad, ? cream with large purple blotch in centre whence ? purple lines radiate to margin and apex; wings 6 mm. long, 2.5 mm. broad, marked with ? purple lines; keel 9 mm. long, blotched with ? purple near apex and marked with ? purple lines. Pods 12–17 mm. long, black when ripe, turgid; valves wrinkled; beak oblique (straight in one specimen), usually short but up to nearly 3 mm. long; seeds pale brown mottled with black, rather large, 3 mm. long. South Island: North-eastern Botanical District—On shaded faces of limestone cliffs in various gorges on the south-eastern side of the Inland Kaikoura Mountains. B. C. Aston! Had this limestone-cliff plant been the only species of Carmichaelia in its immediate locality, its altogether different habit, together with its much longer and wider branches and hoary branchlets, its 7-flowered racemes, and its larger flowers, would separate it from any forms of C. Monroi as at present understood. But on stony debris, in close proximity to the rock-plant, but in the open, Mr. Aston collected specimens of a Carmichaelia with short close branchlets like those of C. Monroi of the Manual except that they are pilose as in the rupestral plant. Unfortunately their pods, & c., are too immature in the specimens at my disposal for further comment. The type of Carmichaelia Monroi Hook. f. was collected by Monro “from half-way up to the summit of Macrae's Run” (Awatere River basin):

Handbook of the N.Z. Flora, p. 49. Hooker's description is very short and inadequate. But the branchlets are described as glabrous, while the flowers are smaller than in the above rupestral plant. Considering that Garmichaelia Monroi Hook. f. and the Clarence Valley plant grow on opposite sides of the same range of mountains, and that the ecological conditions of both areas are not very different, it seems fair to offer the suggestion that perhaps Hooker neglected to note the hairy branchlets of the Awatere plant, and that the groups here discussed are not distinct, but one and the same. But this question can alone be decided by comparing Awatere and Clarence Valley material and growing both rupestral and debris plants from seed and then cultivating the seedlings under identical conditions. Coming next to Carmichaelia Monroi in the sense used by Kirk and Cheeseman, this is invariably a low-growing shrub with dense erect branch-lets forming open flat cushions on stony ground. But an examination of the material of this species at my disposal and a comparison with the descriptions of Hooker, T. Kirk, and Cheeseman respectively have led me to the opinion that more than one varietal group is included. For instance, Petrie's Otago specimens have glabrous calyces—a marked contradistinction to Hooker's description of his Awatere specimens as having a “hoary” calyx. Also, specimens collected by me in the Eastern Botanical District have almost tomentose calyces, while Cheeseman describes the calyx as “silky, sometimes densely so,” but he does not suggest that it is ever glabrous. These Otago specimens, too, have triangular, but not narrow-triangular, calyx-teeth as given by Cheeseman, while the Eastern Botanical District plant has small calyx-teeth. From the above it seems clear that the Otago plant at least should be separated from its allies as a variety, but I do not propose to take this step until the taxonomy of the whole group is made clear. Again, there is an allied but much taller plant than the above cushion-form. This I have collected at Riversdale (Waimakariri River basin) and on the Waimakariri River bed on the Canterbury Plain near the protection-works. A specimen was planted by me in the gardens of the Biological Department, Canterbury College, but, unfortunately, before I could describe it, it was killed during the building of the new chemical laboratory. Another living plant was for many years in the old “native section” of the Christ-church Botanical Gardens, and it may still be there. Also, there is in my herbarium, under the MS. name Carmichaelia humilis, a specimen collected by Mr. Petrie in the North Otago Botanical District. To sum up, there is a group of more or less low-growing forms of Carmichaelia closely allied to and including Hooker's original C. Monroi which does not consist of a number of identical individuals, but of minor groups distinguished from one another by well-marked characters, so that the major group is either a collection of closely allied species one of which is Carmichaelia Monroi, or this latter should be treated as an aggregate species consisting of perhaps five quite distinct varieties. 29. Cassinia albida (T. Kirk) Cockayne. In Trans. N.Z. Inst., vol. 38, pp. 368–69, 1906, after considerable experience both with Cassinia Vauvilliersti Hook. f. and the var. albida T. Kirk, I proposed to rank the latter as a species. Cheeseman (Trans. N.Z. Inst., vol. 39, p. 446, 1907), criticizing my procedure, said that the course to be followed in this matter would “depend largely on the point of view and personal judgment of the observer, coupled, of course, with a full consider-

ation of the evidence available.” More recently Cheeseman (Illustrations of the N.Z. Flora, facing pl. 107) accepts C. albida as a species. With my views as to the relations of species and variety greatly changed since 1906; I would now reverse my decision, were it not that both C. Vauvilliersii and C. albida embrace more than one microspecies, and that if the latter were reinstated as a variety of the former it would be necessary to establish subvarieties in addition to varieties, so overburdening the nomenclature. In Trans. N.Z. Inst., loc. cit., a var. canescens of Cassinia albida is defined by me. This is distinguished from the type, to which the distinguishing varietal name “typica” is here given,* Cassinia albida (T. Kirk) Cockayne var. typica Cockayne var. nov. Foliis supra pilis sparsissime obtectis. by the leaf being so densely covered on the upper surface with a mat of white hairs as to look as if powdered with dust or mildew. Some time ago Professor A. Wall sent me living plants of both varieties. These cultivated in my garden have put forth many young shoots, which maintain their distinguishing varietal characters, though in var. canescens the hoariness is somewhat less marked. Both varieties are confined to the North-eastern Botanical District, but recent observations of Wall show that possibly neither variety extends to its southern boundary. In addition to the two varieties dealt with above, Mr. Aston two years ago collected in the Clarence Valley a variety of Cassina albida which is woolly on the under-surface of the leaf and rather more hoary on the upper surface than is var. canescens. But, as I have only the one specimen, I merely call attention to this apparently distinct form. It is a matter of interest that on the Lord Auckland Islands the closely related Cassinia Vauvilliersii (Homb. & Jacq.) Hook. f. is represented by two varieties—viz., the type and one with a canescent upper surface to the leaf. These characters are so striking that the two varieties can be recognized at a distance.† See L. Cockayne, The Ecological Botany of the Subantarctic Islands of New Zealand, The Subantarctic Islands of New Zealand, vol. 1, p. 216, 1909. 30. Epilobium chloraefolium Hausskn. This extremely common subalpine species was first described by Haussknecht in 1879 from dried material (Oestr. bot. Zeitschrift, vol. 29, p. 149). Although, as Haussknecht points out, the species bears no resemblance to E. rotundifolium Forst. f., the dried material which he examined in various English herbaria showed him that it had been referred to the latter. But long after Haussknecht's subjecting the New Zealand Epilobia to a searching inquiry—indeed, up to the publication of Kirk's Students' Flora in 1899—with but few exceptions, the New Zealand Epilobia, now known to number at least thirty-eight species,‡ Since the publication of the twenty-eight species admitted by Cheeseman in the Manual the following have been described either as new or “restored”: Epilobium antipodum Petrie. E. arcuatum Petrie, E. cinereum A. Rich. (to replace E. junceum Sol. in part), E. Cockaynianum Petrie, E. erectum Petrie (to replace E. junceum var. macrophyllum Hausskn.), E. hirtigerum A. Cunn. (to replace E. junceum var. hirtigerum). E. nerterioides A. Cunn., E. pedunculare A. Cunn., E. rubro-marginatum Cockayne, E. tasmanicum Hausskn. (the last two to replace E. confertifolium Hook. f. so far as it applied to plants other than those of the New Zealand Subantarctic Botanical Province). as well as some strongly marked varieties, had been crammed into the Procrustean bed of Hooker's arrangement in the Handbook, where but seventeen species were admitted. This summary

treatment was for the most part due, I believe, to Hooker's statement in the Handbook (p. 76), as follows: “I have repeatedly studied the New Zealand ones [Epilobia], many of which completely puzzle me. The following descriptions represent in many cases perhaps prevalent forms rather than species; and the student will certainly find intermediates between most of them. It is useless attempting to name many species until copious suites of specimens are collected, the characters being to a great extent comparative.” Cheeseman's description of Epilobium chloraefolium in the Manual is excellent. However, he states (p. 178) that, though a well-marked plant, it is “at the same time a very variable one, especially in height, degree of branching, size of flowers and capsules, & c.” But this-variability depends, so far as my investigations go, not upon there being a number of true-breeding races (microspecies) included in either Cheeseman's or Haussknecht's groups, but rather upon true variability according to environment—shade- and sun-plants, for instance, differing greatly in certain particulars. Also, I rather suspect that certain hybrids are included by Cheeseman in his group. In this note I am suggesting an enlargement of the conception of the species by adding a distinct true-breeding group which, although it fits well into the original description if size of organs is ignored, far surpasses the type in this respect. I am also giving a varietal name to the “type,” so that to those accepting my conclusions the group E. chloraefolium will consist of the two varieties and of any other allied varieties which may be segregated from the individuals now constituting the species, or in course of time be discovered. (a.) Epilobium chloraefolium Hausskn. var. kaikourense Cockayne var. nov. Habitu robustiore, floribus duplo majoribus a typo differt. This well-marked variety is distinguished at a glance from any form of Epilobium chloraefolium by its exceedingly robust habit and large white flowers, which at times are quite 28 mm. in diameter when fully opened. The stems, decumbent at first, finally erect and woody in their older parts, are stout, purple, shining, smooth, and minutely bifariously pubescent. The leaves are numerous, moderately close-set, pilose especially on the margin at the base and on the petiole when young but finally glabrous, rather thick, coriaceous, somewhat glossy, bright- or yellowish-green above, often reddish beneath, and taper into a short, broad, channelled petiole about 4 mm. long; the lamina is more or less broadly oblong or even elliptic, about 20 mm. long by 12mm. wide and distinctly but minutely toothed, its apex is obtuse, the midrib is strongly keeled, and the lateral veins distinct. The flowers are few in the axils of the terminal leaves, white, invariably large, and often attain 28 mm. diameter. The plant continues blooming for more than six months. The capsule is about 3.3 cm. long, dark purple, minutely pubescent; its peduncle is only 3.5 mm. long—i.e., it is much shorter than the subtending leaf, which may be 19 mm. long or longer. The seeds are numerous, ± 1.75 mm. long, light-brown, and papillose. The great differences in appearance which the above – described vars. of E. chloraefolium present made me inclined at one time to consider the var. kaikourense a distinct species, especially as it came true from seed and occupied a special limited area of distribution. So long ago as 1892 Mr. T. Kirk wrote to me regarding a specimen (herb. L. Cockayne, No. 3668) I submitted for his opinion, “May prove distinct, but further specimens must be examined—a very interesting form.”

A careful comparison of living plants of both varieties in my garden shows that there is no important difference between them except size, while the structure of the flower is identical in both. The plant was found in the first instance by myself in 1892, growing in rather moist soil on cliffs a little distance from the sea at Kaikoura, and shortly after that I found it to be abundant under the subalpine scrub on Mount Fyffe (Seaward Kaikoura Mountains). Plants were cultivated in my New Brighton garden, where they, or their seedlings, remained for six years at least; and seeds were sent in 1897, and probably earlier, to various European botanical gardens under the name Epilobium Cockaynianum* E. Cockaynianum Petrie in Trans. N.Z. Inst., vol. 41, p. 140, 1909, has no relationship to this, but is related to E. alsinoides A. Cunn. Petrie ined., but which Petrie never published. In 1905 Mr. H. J. Matthews and myself again observed the plant on Mount Fyffe, and I recorded its occurrence in Trans. N.Z. Inst., vol. 38, p. 373, 1906, as Epilobium sp. aff. E. chloraefolium Hausskn. Since then the plant in question has been found by Mr. C. E. Foweraker and myself in the Awatere Valley, by Mr. B. C. Aston in the Clarence Valley, and by Professor A. Wall on the Seaward Kaikoura Mountains; while a plant from Mount Isabel, at Hanmer, collected by Mr. C. E. Christensen probably is var. kaikourense. In other words, the variety is confined to the North-eastern Botanical District, where it is of wide distribution from sea-level to at least 900 m. altitude, and grows on rock, beneath shrubs, and probably in shady tussock grassland. Phytogeographically the distribution of E. chloraefolium var. kaikourense is an interesting case of a true-breeding race of a species of wide distribution being confined to a limited area which possesses a special ecological character, as reflected in the great number of locally endemic plants. From the horticultural standpoint, the ease of culture of the plant, its general habit, beautiful long-blooming flowers, and purplish foliage and stems render it worthy of any rock-garden; nor is there fear of its becoming a weed, as in the case of certain New Zealand Epilobia. (b.) Epilobium chloraefolium Hausskn. var. verum Cockayne var. nov. This equals E. chloraefolium as described by Haussknecht in Monographie der Gattung Epilobium, p. 299, Taf. 19, fig. 81, 1884. No further description is needed. The differences between var. verum and var. kaikourense are given above. 31. Epilobium pedunculare A. Cunn. var. brunnescens Cockayne var. nov. Caulibus pallidis saepe brunneis tinctis. Foliis ovato-oblongis vel oblongo-rotundatis, supra pallide viridibus saepe brunnescentibus, margine remote dentatis, subtus purpurascentibus. Capsulis glaberrimis, pallide bruneis, ± 5.3 cm. longis; pedicellis colore capsulis etiam, multo elongatis, 6.6 cm. longis. This variety forms large more or less circular patches. The leaves vary in size, but about 9 mm. long is frequent; the petioles are about 3 mm. long. Where exposed to bright light the leaves assume a brownish tinge. The flowers are small, white, and about 5 mm. diameter. The calyxsegments are narrow-oblong, 3 mm. long, brownish, and end in a swollen purplish apex. The capsule is more than three times the length of the ovary, and the peduncle increases from about 2.2 cm. to 5.7 cm. as the capsule develops.

Epilobium pedunculare var. brunnescens has a wide range, but this I cannot at present define, nor its ecological distribution. However, I have plants in my garden identical in every particular collected from localities far distant from one another—viz., Mount Egmont (coll. L. C.) (EgmontWanganui Botanical District) and Four Peaks (coll. A. Wall) (south of the Eastern Botanical District). Haussknecht describes a var. laxa of E. pedunculare, and it may be that my new variety is the same. But without actually comparing the material on which Haussknecht founded his variety it is impossible to come to a conclusion, so it seems to me better to risk the establishment of a synonym, which for a time will serve a definite phytogeographic purpose, than to withhold publication or refer the group to var. laxa, which it may not be after all. The further question arises, am I right to uphold the species Epilobium pedunculare A. Cunn. rather than follow Hooker, Kirk, and Cheeseman, and deal with it as a variety of E. nummularifolium? Haussknecht—relying only, however, upon dried material—keeps the two species distinct, and strongly supports his position by the two fine figures 94 and 96 (Monographie der Gattung Epilobium, Taf. 22 and 23). He also states that in herbaria E. pedunculare is frequently found mixed with E. nummularifolium, but that they are readily distinguished by E. pedunculare having the leaves smaller, more close-set, thicker, entire, and with shorter stalks; the capsule glabrous and its peduncle more slender, and the seeds covered much more thickly with papillae (i.e., p. 303—freely translated). My own experience, after many years' observation of various groups included under the specific names nummularifolium and pedunculare, both growing in many parts of New Zealand and also cultivated by me, has convinced me that the two species are absolutely distinct, and separated by well-marked unchangeable characters. Epilobium nummularifolium, in one form at any rate—and the species may quite well contain only the one form—is common throughout the North and South Islands, but absent in Stewart Island; it appears to be mainly a lowland plant, but there is no exact record of its distribution, such being confused with that of E. pedunculare, which ascends at least to the subalpine belt. E. nummularifolium may be best distinguished from the aggregate E. pedunculare by its orbicular or suborbicular bright-green leaf with at times a more or less truncate base, its rather long petiole which is winged above, its capsule not glabrous but closely covered with a short cinerescent pubescence, its fruiting peduncle not lengthening so greatly as in E. pedunculare, and its less papillose seeds. Even the leaves alone of living specimens enable the two species to be identified in an instant. 32. Epilobium pedunculare A. Cunn. var. minutiflorum Cockayne var. nov. Varietas distinctissima, caulibus gracilibus rubro-purpureis, foliis parvis rotundis subrotundis vel ovatis viridibus, floribus minutis, pedunculis statu fructu solum 3 cm. longis et capsulis purpurascentibus brevibus 11 mm. longis facile distinguenda. South Island: Eastern Botanical District — (1) Trelissick Basin, but details regarding habitat wanting: A. Wall ! (2) Rakaia River bed not far from mouth of river: H. H. Allan !

The above variety is described from a plant which has been in my garden for only a few weeks, collected by Professor Wall as above. The plant forms matted patches. The leaves vary in size from less than 3 mm. long up to about 5mm. The reddish-purple stems, petioles, and peduncles contrast with the bright-green leaves. The flowers are white; the calyx is pale brown tinged and margined with reddish-purple; the petals are white; the slender petiole only increases from 2 cm. to 2.5 cm. when the capsule is ripe. 33. Epilobium pedunculare A. Cunn. var. viride Cockayne var. nov. Caulibus teneris pallide viridibus; foliis oblongis vel rotundatis remote et obscure dentatis, laminis usque ad 7 mm. diam.; floribus 6.5 mm. diam., pedunculis brevibus 5–11 mm. longis; capsulis circ. 2.8 cm. longis; viridibus secundum suturam brunneo tinctis. North Island: North Auckland Botanical District—On river-bed near Fairburn, Mangonui County. H. Carse ! This variety is readily distinguished from E. pedunculare var. brunnescens by its green leaves and stem and by the much shorter peduncle of the flower, which does not elongate to nearly the same length as that of var. brunnescens in the fruiting stage. The flowers and capsules are also smaller. The plants now growing in my garden, from which the above diagnosis is drawn up, were collected specially for this paper by Mr. H. Carse as an example of the form of Epilobium pedunculare in his neighbourhood. Whether the above variety is identical with the plant originally described by Allan Cunningham I cannot say, for the original description is quite general and would fit almost any variety of the species. 34. Gunnera densiflora Hook. f. In 1864 Hooker published his Gunnera densiflora, basing his description on specimens collected by W. T. L. Travers at an altitude of 4,000ft. in the Acheron and Clarence Valleys—i.e., it must have been collected not far from the sources of these rivers if the altitude as given is correct. For some thirty-two years no Gunnera was discovered that could be referred to the above species until, in 1896, I collected a species of Gunnera in the Craigieburn Mountains (Eastern Botanical District), at the headwaters of the Hogsback Creek, at an altitude of rather more than 900 m. Specimens were sent by me to Kirk, who referred them, apparently without hesitation, in his Students' Flora (1899) to G. densiflora. Cheeseman, in the Manual, working with the same material as Kirk, drew up a new diagnosis of the species based partly upon Hooker's original brief description and partly upon my not-too-well-prepared specimens. Regarding these latter Cheeseman writes that they “are the only ones I have seen that can be referred the species.” So the matter remained until, in 1911, Mr. R. M. Laing, M.A., B.Sc., during a botanical excursion towards the headwaters of the Rivers Clarence and Waiau, discovered, in abundance, on the western side of Lake Tennyson a species of Gunnera which, in my opinion, is equivalent to the plant on which Hooker founded G. densiflora. Mr. Laing submitted his material first of all to Cheeseman, who suggested that it might quite well be Gunnera cordifolia Hook. f., hitherto thought to be confined to Tasmania. Later, Mr. Laing submitted specimens for my opinion, telling me also what Cheeseman had said. G. cordifolia

is well illustrated in Das Pflanzenreich (IV. 225. Halorrhagaceae, p. 108, fig. 31). After examining Mr. Laing's specimens, and comparing them with the above-cited figure, & c., I came to the conclusion that the species was either G. cordifolia or a variety of that species, and so dealt with it in my unpublished Vegetation of New Zealand. Laing (Trans N.Z. Inst., vol. 44, pp. 65–66, 1912) drew up a detailed description of the Gunnera in question in the field with the living plant before him, and his and my original opinions are considered below. Regarding the Craigieburn Mountains plant, Laing stated (i.e., p. 66) that Cheeseman had informed him it was distinct from the Lake Tennyson plant, and that it had been identified by the Kew authorities as Gunnera densiflora Hook. f. Learning recently that Professor Wall intended paying a botanical visit to the Trelissick Basin, I explained to him as exactly as I could—no easy matter—the precise spot where I had collected the Gunnera in 1896, and urged him to make a thorough search. This he most willingly did, and not only found the plant in quantity in the locality indicated, but discovered other stations for it in the neighbourhood. He secured ample material, of which he sent me abundance both living and dried, some of the former being now growing in my garden. As it is a matter of considerable phytogeographical importance to get detailed knowledge of this rather critical species of Gunnera I am publishing a description. Description of the Gunnera from the Craigieburn Mountains. Rhizome short, ± 19 mm. long by 5.5 mm. thick, rooting with straight roots about 8 mm. long, and giving off stout, terete, dark-brown, more or less strigose-pilose stolons each about 3 cm. long and 2 mm. diam. Leaves in rosettes of about 4 or 5 ± 3.5 cm. across; petiolate with petiole variable in length from about 2–2.8 cm., fleshy, pale often tinged pink, terete or channelled above or only near junction with lamina, pilose with strigose white hairs on back and margin but variable in this respect as to density of hairs on different leaves of same plant; lamina moderately bright green, coriaceous, usually more or less cordate at base, sometimes truncate, auricled at base with two small toothed appendages ± 2 mm. long which are bent upwards, orbicular or broadly ovate-orbicular, hairy above and on margin, glabrous beneath, rounded at apex or occasionally almost subacute, rather coarsely but sharply toothed with about 9 teeth ± 1 mm. long on each side, veins evident above and beneath, midrib stout and keeled beneath. Flowers unisexual, numerous. Male flowers in spikes about 2.5 cm. long terminating rather stout scapes about 2.8 cm. long arising from axils of leaves and densely covered with brown strigose hairs; pedicels very short, subtended by a small narrow subulate bract about 2mm. long; calyxlobes 2, narrow-triangular, about 1 mm. long; petals 2, transverse, narrow linear-spathulate, much exceeding anthers, 4 mm. long and 0.5 mm. wide near the black acute apex; stamens 2, situated on base of petals, broadly ellipsoid, 2 mm. long, rounded at apex, filaments extremely short. Female crowded into a dense globose head about 6 mm. long terminating a stout, fleshy, pale or pale-brown scape 10 mm. long, pilose with numerous white hairs; calyx-tube urceolate, pale green, smooth, about 2 mm. long; calyxlobes 2, subulate, purple with black tip, about 0.75 mm. long. Styles 2, wide-spreading, pale brown, 4–5 mm. long, stigmatic throughout. On compa ing, detail by detail, the above description with that of Laing (for the Lake Tennyson plant), of Schindler (for the Tasmanian plant),

and of Hooker (for the type of Gunnera densiflora), my opinion is that the New Zealand plants are all one and the same, and that the Tasmanian may also belong to the same species. On the other hand, the New Zealand Gunnera, accepting Schindler's description and illustration of the Tasmanian, differs from the latter in the petals, which are much longer and narrower in the New Zealand than in the Tasmanian plant, and in the shape of the drupe, pyriform in the New Zealand, ovoid according to Hooker ex Schindler in the Tasmanian plant. Also Schindler's figure shows the margin of the leaf of his plant as strongly ciliated, whereas in the New Zealand plants the ciliation is virtually confined to young leaves. Bearing the above in mind, it seems best to maintain Gunnera densiflora as a species, but to remember that it is extremely closely related to G. cordifolia Hook. f. of Tasmania, and is a further link between the floras of the eastern Australian and New Zealand regions. 35. Haastia recurva Hook. f. var. Wallii Cockayne var. nov. Foliis et capitulis quam illa typi minoribus; plerumque pilis albidis munitis sed eis prope ramulorum apices interdum subfulvidis tinctis; bracteis involucri apiculatis. The variety differs from any example of Haastia recurva that I have seen hitherto in the much smaller size of all its parts, in its dense wool being white nearly everywhere and only slightly fulvous near the apices of the branchlets, and in the apiculate apex of the involucral bracts. The leaves are generally less than 10 mm. long, or only half the length of good-sized leaves of the type. The flower-heads are 7 mm. diam., or not half the size of medium-sized heads in the type. The plant was collected by Professor A. Wall on a shingle-slip near the summit of Mount Fyffe, Seaward Kaikoura Mountains. Unfortunately, only the one plant was noted. On receiving the specimen I thought this plant with white wool and slender branches might be the common form of the Kaikoura Mountains in general, but upon comparison with Mr. Aston's specimens from Mount Tapuaenuku (Inland Kaikoura Mountains) this apparently is not the case. Specimens from Shingly Range (Awatere) also belong to the type. 36. Haastia Sinclairii Hook. f. Judging from specimens in my herbarium, there appear to be two distinct groups of plants included under Haastia Sinclairii Hook. f. by Cheeseman (Manual, p. 321). One of these groups is figured in Cheeseman's Illustrations of the N.Z. Flora, pl. 100, and this appears characteristic of the species so far as the North-western and Western Botanical Districts are concerned; but the Fiord Botanical District group appears to differ in certain particulars as compared with the more northern plant, especially in its smaller leaves, which are covered beneath much more thinly with fulvous (not white) wool, and above are thinly covered with wool or, at times, almost glabrous. Also, the heads of the Fiord plant are much smaller. I do not propose here to separate the species into two varieties, the intention of this note being to call the attention of collectors in the area of the species to probable differences in plants of this species which they may find. In the Illustrations of the N.Z. Flora Cheeseman comments upon the distribution of Haastia Sinclairii as follows: “H. Sinclairii, which is a true ‘shingle-slip’ plant, never found away from the slopes of dry shingle which form such a prominent feature on the eastern side of the Southern Alps.”

And farther on—“But it was soon found to have a wide distribution on the eastern side of the Southern Alps, and is now known to extend from the northern portions of the Mount Arthur Range southwards through the Canterbury Alps to the south-west of Otago…. I am not aware, owever, that it occurs in any locality well on the western side of the watershed of the Alps.” The above statements, though topographically true in our present state of knowledge, neglect the ecological viewpoint. There are two distinct classes of shingle-slips—the one very dry on the surface and situated beyond the average limit reached by the western rainfall, and the other not by any means so dry a station, since it lies within the wet area. These two classes of shingle-slip are clearly defined by their plant inhabitants. A dry or eastern shingle-slip contains that wonderful assemblage of which the following, to cite only a few, are characteristic: Craspedia alpina, Notothlaspn rosulatum, Poa sclerophylla, Ranunculus Haastii, Stellaria Roughii, Veronica epacridea. On the other hand, the western shingle-slip contains none of the above species; in fact, there are but few plants common to both—e.g., Epilobium pycnostachyum is one. The species of Haastia, too, are an especially good index. On an eastern shingle-slip (using the term ecologically and not as used in the quotation above) Haastia recurva is alone to be found, but on a western shingle-slip it is absent, being represented by H. Sinclairii. So, too, the dry east gives Veronica Haastii, but once well into the area of excessive rain it is V. Haastii var. macrocalyx. As for Haastia Sinclairii not having been found to the west of the actual Divide, I suspect this is chiefly due to the fact that shingle-slip is not much in evidence on the west, speaking comparatively, and also that, the North-western Botanical District excepted, few collections have been made on mountains possessing shingle-slips, these true western mountains being almost entirely unexplored botanically. 37. Hymenanthera crassifolia Hook. f. Hymenanthera crassifolia Hook. f. was originally a mixture of Scaevola novae-zelandiae A. Cunn., now known as Hymenanthera novae-zelandiae (A. Cunn.) Hemsley, and plants from Cape Palliser and Nelson. These latter, along with certain other plants, form H. crassifolia in its restricted sense. But the distribution of the species is uncertain, owing to lack of knowledge as to the limits of polymorphy to be allowed or the variation which takes place not only in H. crassifolia but in the species next dealt with—H. obovata T. Kirk. At any rate, so far as I know, true H. crassifolia is found in the Ruahine-Cook Botanical District, on the coast both of the Wellington and Sounds Subdistricts, whence it extends, but not in an unbroken line, to the coast of the South Otago Botanical District. Certain inland plants have been referred to this species, as also a Stewart Island plant, but all these determinations must be received with caution. Although there is an admirable plate of H. crassifolia in the Flora Novae-Zelandiae, there is no description easily available which deals with the colour of the flower, and as this is an important character for identification purposes the following description of the flower may prove useful. Description of Flowers of Hymenanthera crassifolia Hook. f. Flowers inserted on under-surface of twigs, numerous but quite hidden from view on living plant, very small, usually solitary but close-set, hermaphrodite,

slightly sweet-scented, pedunculate with decurved or straight green peduncle rather shorter than the flower, furnished with two minute, broadly triangular, scarious brown bracts inserted a little below its centre. Sepals orbicular, about one-half length of petals, green with broad, purplish, minutely fimbriate margin. Petals waxy in appearance, lemon – yellow, oblong, ± 3 mm. long, obtuse, recurved at apex which on margin is sometimes purplish; in bud deeply stained purple. The pollen is shed just before the flower opens or shortly afterwards and so easily falls on the stigma. There is no honey. Abundance of pollen reaches the stigma. After pollination the ovary, & c., rapidly enlarges. In the neighbourhood of Wellington H. crassifolia commences to bloom some time during the first two weeks of September. 38. Hymenanthera obovata T. Kirk. Hymenanthera obovata, as established by T. Kirk in 1895 (Trans. N.Z. Inst., vol. 27, p. 350), and upheld by Cheeseman in 1906 (Manual, p. 50), is based on material from two sources—the Trelissick Basin (Canterbury) and various localities in Nelson. An examination of the type material in Kirk's herbarium shows that the Trelissick and Nelson material look very different, and the feeling at once arises that the species as at present constituted is a combination of two distinct groups of individuals, each of which is entitled to rank as a species. The above difficulty is increased, firstly, by the imperfect knowledge of the flowers of either the Trelissick or Nelson plants, and, secondly, by the discovery by Mr. B. C. Aston, some years ago, of another group of individuals with, it is now known, a local distribution along the shores, & c., of Cook Strait from the French Pass and Kapiti Island to Somes Island in Wellington Harbour. This last-named group far more closely resembles the Nelson than the Trelissick group—indeed, when the flowers of the Nelson group are investigated it possibly will be found either that the two groups are identical, or that they are microspecies which must be united under one name. Likewise, judging from Kirk's type specimens, from my personal knowledge of the genus Hymenanthera in the Trelissick Basin, and from fresh specimens of the plant in question recently collected in the above locality by Professor Wall, it seems not unlikely that the Trelissick group may eventually be referred to H. crassifolia. The present state of knowledge regarding Hymenanthera obovata, which I have attempted to concisely indicate, demands that any further knowledge should at once be made available for students and collectors, so that the real status of the species, and of the groups cited above, may be established. Thanks to Mr. Aston, who last year (1916) put me in the way of seeing the Wellington plant in more than one locality, and who assisted me in collecting ample flowering material, I am in a position to describe the Wellington plant. Further, Mr. Aston at the close of the year 1917 collected material of the Nelson plant from the Riwaka-Takaka hills, which he has placed in my hands. Finally, Professor Wall has procured for me living and dried material of the Trelissick plant; while Miss Herriott (Biological Laboratory, Canterbury College) sent me some time ago from Cass (Waimakariri River basin) seedlings of the Hymenanthera of that locality, which must be either H. obovata T. Kirk (in part) or H. dentata var. alpina, another group of quite uncertain position.

Description of the Wellington Coastal Hymenanthera (= ? H. obvata T. Kirk in its restricted sense). A low shrub more or less flattened to the substratum into which its prostrate stems root, but its height varies with regard to degree of exposure of the plant. Branches more or less divaricating and interlacing, with younger twigs clad with pale bark covered with a fine pubescence, but older twigs having grey bark dotted freely with lenticels. Leaves obovate or oblong-obovate, varying greatly in size according to situation, but from 1 cm. long by 7 mm. wide to 3 cm. long by 1.6 cm. wide are a fair average, though there are others both larger and even smaller, very dark green above, whitish-green beneath, very thick and coriaceous, obtuse, often emarginate, generally entire but occasionally there is a coarse tooth on either side,* The Manual description reads, “quite entire,” but even Kirk's type specimens show some leaves not entire. above veins obscure, beneath evident but not numerous; petiole short, about 3 mm. long. (In shade reversion-shoots occur with leaves thinner, larger, more irregular in shape, sometimes rhomboid, 1–2-toothed on either side.) Flowers numerous, about 4 mm. diam., mostly on the naked branches in the axils of former leaves, solitary or in fascicles of about 4, apparently hermaphrodite, almost twice as large as those of H. crassifolia (see above, No. 37), paler yellow, more urceolate, and margins of petals edged with a bright-purple line; pedicels about 4 mm. long, pale green, fleshy; bract broadly triangular, acute; sepals 4–5 times shorter than petals, much broader than long, green at base but strongly margined with purple, rounded at apex which is fimbriate; petals linear-oblong or narrow ovate-oblong, about 5 mm. long and 2 mm. broad, pale yellow, obtuse, sometimes emarginate, strongly recurved; stamens with orange staminal process; scale (nectary) obovate, slightly praemorse at fimbriate apex, and abundance of honey at base. Apart from the much greater size of the leaves, and, where not exposed to the most powerful wind, the much more open character of growth, the above species is distinguished at once from H. crassifolia by the flowers, which are twice as large, the sepals not half length of petals but only one-third or one-fourth as long, the narrower, longer, pale-yellow not lemon-yellow petals. With regard to Nelson specimens of undoubted Hymenanthera obovata (in the restricted sense), Aston's specimens are from two sources. The first grows “in crevices of limestone rock at from 2,500ft. to 2,700ft. on the Riwaka Hill, and 1–3 ft. high” (fide Aston). Specimens of this plant show (as described for the Wellington plant) the leaves linear-obovate to occasionally almost linear, very numerous, alternate or fascicled, from more than 4 cm. long to 1.5 cm. or even less, not thick, probably rather dark green above, pale beneath, tapering into a short petiole, entire, rounded at apex. The second was taken from one plant growing at Golden Bay, near the cement-works. It was a “shrub with trunk about 10 ft. high and pendulous branches growing in shade on limestone country” (fide Aston). This specimen has leaves up to 7.5 cm. long, some are 3 cm. broad, quite small leaves are rare. They are bright green above, pale beneath, entire, occasionally emarginate, not in fascicles. Although the two plants just noted differ so far as the leaf is concerned in some particulars from the Wellington plant, such differences are probably

Plate IX Leptospermum scoparium Leonard Wilson, growing naturally near Port Levy, Banks Pennsula.

Plate X Flowering branch of Leptospermum scopanum Leonard Wilson, showing the double white flowers

entirely environmental. The most interesting point is the greatly reduced leaves present with much larger ones on the Riwaka plant, and such suggest that perhaps the Trelissick Basin plant is, after all, a reduced form. An examination of flowers and fruit can alone settle this interesting point, but I am still inclined to agree with my opinion as stated above — that the Trelissick plant is one species, and that the Riwaka and other Nelson plants should be united with the Wellington plant either as a polymorphic or an epharmonic group. 39. Leptospermum scoparium Forst. (forms with double flowers). (Plates IX and X.) In New Zealand Plants and their Story, p. 149 (1910), I have called attention to a form of Leptospermum scoparium with double flowers which was discovered by Mr. E. Phillips Turner, F.R.G.S., in the Volcanic Plateau Botanical District. A second plant with double flowers was found some four years ago at Torrent Bay, Nelson, by a lady residing at Motueka. This information I received from Messrs. Nairn and Sons, nurserymen, of Christchurch. A third plant with double flowers must now be recorded. This was found recently by Mr. Leonard H. Wilson on his property at Port Levy, Banks Peninsula. I am indebted to Mr. J. Young, Curator of the Christ-church Botanical Gardens, for calling my attention to this interesting plant and for supplying the fine photograph (see Plate IX) of the wild plant in its original habitat, the photograph being taken by his son, Mr. James E. Young. Cuttings from the Port Levy plant were struck by Mr. Young, so that there is now a vigorous specimen in the collection of New Zealand plants in the Christchurch Botanical Gardens. Since the doubling of flowers is essentially a teratological phenomenon, one cannot look on such a race, capable only of being reproduced artificially from cuttings or layers, as equivalent to a taxonomic variety. I would propose for it the garden name of “Leonard Wilson,” the plant to be known therefore as Leptospermum scoparium Leonard Wilson. 40. Myrtus Ralphii Hook. f. This species was founded by J. D. Hooker on specimens collected by Dr. Ralph near the City of Wellington in the very early days of the province, and on the east coast of the North Island by Colenso, and it was first published in the Flora Novae-Zelandiae in 1853. Later, in the Handbook of the New Zealand Flora, Hooker suggested that it might be a variety of Myrtus bullata Sol. The species was accepted by T. Kirk (Students' Flora, p. 165) and by Cheeseman (Manual, p. 169), both authors agreeing that it is closely allied to M. bullata. During the last few years I have had ample opportunity for examining the “species” in the field, and in consequence have come to the conclusion that it is a polymorphic hybrid between M. bullata Sol. and M. obcordata (Raoul) Hook. f. My reasons for the above conclusion are (1) that the “species” is never to be found unless both Myrtus bullata and M. obcordata are present, and (2) that the individuals are strongly polymorphic even when growing in close proximity, some closely approaching M. bullata and others M. obcordata, while leaves of the obcordata and bullata types occur frequently on the same individual. I do not think a much better example can be found of the often mentioned “series of intermediate forms” connecting two species than is to

be seen in the multiplicity of forms assumed by M. Ralphii and connecting M. bullata and M. obcordata. Thus, to those believing that “intermediates” obliterate the distinctions between groups which if not so connected would be species, the only logical course to take would be either to unite all three species of Myrtus under the earliest name, “bullata,” or to uphold M. bullata and M. obcordata, which form the unlike poles of the series, and to treat the intermediates—i.e., M. Ralphii—as unnamed varieties of whichever of the two species they most resembled. For action of this kind the New Zealand and many other floras offer ample precedent; indeed, one or other of the methods suggested above would be the orthodox taxonomic course to pursue. All the same, the most inveterate “lumper” could not bring himself to unite groups so absolutely different as those represented by M. bullata and M. obcordata. Some exact details regarding the polymorphy of Myrtus Ralphii may here be given in support of my contention that it is of unfixed hybrid origin. On and near the saddle joining the Kaukau Range and Mount Crowsnest, near Wellington City, there is a remarkable scrub-association which owes its presence to excess of wind. In certain places near its outskirts there is abundance of the three species of Myrtus mentioned above growing side by side. Although frequently somewhat stunted in habit, M. bullata can be recognized at a glance; so too, generally, with the individuals of M. obcordata. But on examining the bushes of M. Ralphii it is seen at once that there is no uniformity amongst the individuals, some coming somewhat near to M. bullata in colour, shape, and blistering of leaf, while others are far more of the obcordata type—some, indeed, being almost identical with that species. Thus a hybrid origin is at once suggested, and close examination for and against such a supposition demanded. Happily for such an investigation, Myrtus bullata and M. obcordata possess certain well-marked distinguishing characters. Thus, taking the leaves alone, even were the flowers of the two species identical, so different are the leaves that no taxonomist would unite the species. For bullata there is (1) the large leaf, (2) its bullate surface, (3) its usually acute apex, (4) its power of becoming reddish-brown when exposed to the sun, and (5) the base of the lamina not narrowed into the petiole. True, the bullate surface may be strongly flattened in a plant grown in complete shade, but it is always present more or less and is a marked unit-character. Then, for obcordata there is (1) the small leaf, (2) the rounded emarginate apex, (3) the flat surface, (4) the tapering base of the lamina, and (5) the more feeble response to coloration by intense light. A number of specimens were collected of the three “species” of myrtle growing side by side on the outskirts of the wind-swept scrub, each specimen being taken from one individual. Of these, after examination, some were put on one side as true Myrtus bullata, others as true M. obcordata, and twenty-two were considered to be M. Ralphii. A closer examination of these twenty-two showed that the specimens fell into two classes—the one with large brownish-red, more or less bullate leaves, and the other with much smaller, greener, and flatter leaves. But these differences affected only the bulk of the leaves of each specimen. From even the largest-leaved specimens with leaves of bullata type small leaves of obcordata type were picked. Any of the fundamental characters mentioned above might be present, absent, or even combined. The twenty-two M. Ralphii, specimens were each numbered and examined in detail. Here are a few results:—

No. 1 had leaves about 17 mm. long by 15 mm. broad, with reddish slightly bullate surface, the apex subacute, and the lamina-base not tapering —i.e., they were almost pure bullata type. But other leaves were only 10 mm. long by 9 mm. broad, and had the obcordata lamina-base, while one leaf was slightly emarginate. No. 4 had many obovate leaves, but others tapered at both apex and base of lamina, and in one case the apex was acute; some were almost flat and some slightly bullate. No. 5 had its slightly bullate leaves generally emarginate with tapering lamina-base; some leaves were almost rotund. This specimen might well have been taken for M. obcordata were it not for the slightly bullate leaves. No. 7 had leaves up to 19 mm. long by 15 mm. broad, their apex generally rounded or subacute, but the bases of the laminae generally tapered and a few leaves had an emarginate apex. No. 8 showed little sign of any obcordata character, but a few leaves had their bases strongly tapering. No. 9 was a distinct form with the large (20 mm. by 15 mm.) leaves pale green, but here again actual obcordate leaves were present. No. 10 strongly approached M. obcordata, but traces of the bullate surface were present, while in some leaves the emarginate apex was wanting and in many not strongly developed. No. 14 was a distinct form with quite small leaves, but these were distinctly of the bullata type, though even here one obcordate leaf was noted. No. 16 is specially worthy of mention. It had bright-green flat leaves, small, but larger than in M. obcordata of the locality. There was hardly a trace of emarginate apex, and the base of the lamina did not markedly taper. Here the most characteristic feature of the series of intermediates—the more or less bullate surface—was absent, but so was the emarginate apex of M obcordata. No. 17 was similar to No. 16, and were it not for the number of leaves with non-emarginate apex and a very slight trace of a bullate surface here and there the specimen would be M. obcordata pure and simple. My contention of the hybrid origin of Myrtus Ralphii does not rest upon the above examples alone. At the “Meeting of the Waters,” near New Plymouth, there, not in a wind-swept habitat, but in the moist, still atmosphere of the forest-interior, I saw an astonishing series of intermediates between M. bullata and M. obcordata, both these species being also present. It was seeing these plants, indeed, which first suggested the theory of hybrid origin, and which led me to carefully examine the plants of the Wellington wind-scrub. Unfortunately, I was not in a position at the time to secure material for a searching examination. At Kenepuru Inlet, Marlborough Sounds, I collected specimens of Myrtus Ralphii. At that time I had no suspicion of its hybrid origin, or I should have collected far more copiously and taken special notes. But the specimens did strike me as not typical M. Ralphii. Examining them (five specimens) now I see that most of the leaves are more or less rotund, large, ± 15 mm. long, and have little trace of a bullate surface. But an emarginate apex is present in a good many cases, and occurs on leaves even 10 mm. long. One specimen has much smaller leaves, obcordate or broadly obovate, and it may be true M. obcordata. In Colenso's herbarium, now to be consulted at the Dominion Museum, Wellington, there are a number of Hawke's Bay specimens of Myrtus Ralphii. Amongst these there is no uniformity, and they present features such as already described.

The distribution of M. Ralphii needs a few words. As Cheeseman has pointed out, it is a local plant and occurs throughout the North Island, Ahipara being the most northerly locality yet recorded. This is the only locality in Mangonui County known to Carse (Trans. N.Z. Inst., vol. 43, p. 210, 1911), but M. obcordata and M. bullata grow in the same locality, which, like M. Ralphii, is according to Carse the only locality for the former. Here is indeed a crucial case. M. bullata is common throughout Mangonui County, according to Carse (a most trustworthy observer); M. obcordata is only known at Ahipara, but this, too is the sole station for M. Ralphii! Without going into further details, I think all must agree that a strong case is made out for the hybrid origin of Myrtus Ralphii. How far all the individuals are actually the F1 generation between M. bullata and M. obcordata it is impossible to say, but the extreme polymorphy suggests that F2 and perhaps other generations are present where individuals are abundant. It is also interesting to see how certain characters may appear unchanged (pure), how others are obvious mixtures, and how others are suppressed; but without actual breeding experiments no conclusions can be drawn. Taxonomically the only change demanded if M. Ralphii be considered a hybrid is to write the name thus: X M. Ralphii. Appendix. Since writing the above I have received from Mr. R. H. Rockel, M.A., a representative collection of the myrtles growing in the forest at the “Meeting of the Waters,” near New Plymouth. Each specimen of the sixty-one specimens sent was taken from a different individual, and so great has been the care exercised by Mr. Rockel and a friend of his who assisted in the work that probably nearly every form of the area in question is represented. A casual glance at the collection shows that my former word “astonishing” used with regard to the polymorphy of Myrtus Ralphii in that locality is no exaggeration. A more detailed examination of the material shows every gradation between typical Myrtus obcordata and typical M. bullata; there are, in fact, specimens which could be called “giant obcordata” and “dwarf bullata.” There is also a series of specimens which match the figure (No. 94) of M. Ralphii in Kirk's Forest Flora. Though, taken as a whole, the specimens can be arbitrarily divided into a number of groups, the majority of the specimens has each its special characteristics. The following call for particular mention:— (1.) Leaves stained brown, slightly bullate, apex rounded usually but some with obtuse apex and one or two slightly emarginate, bases rounded or tapering; lamina generally large—e.g., 2.5 cm. by 2.3 cm. (2.) Leaves similar to those in No. 1 but very pale green in colour. (3.) Leaves quite of obcordata type so far as size, base, and colour go, but none are obcordate. (4.) Leaves similar to those of No. 3 but considerably larger, but one or two of the smallest leaves are true obcordata. (5.) Leaves for the most part very deeply stained with purple, base much tapering, apex rounded or subacute, blistering considerable, lamina frequently about 1.9 cm. by 1.2 cm. (6.) Many leaves almost rotund but all with emarginate apex, lamina averages about 11 mm. by 11 mm.

(7.) Most of the leaves large, deeply coloured with purplish-brown, somewhat bullate, and suddenly narrowed into an acute apex, but some of the smaller leaves have a rounded apex. (8.) Leaves strongly bullate, apex rounded but occasionally emarginate in the smaller leaves, average lamina about 1.7 cm. by 1.5 cm. (9.) Average lamina only 8 mm. by 6 mm., surface flat, hardly a trace of emarginate apex. (10.) Leaves tapering both at apex and base, most rather large, flat, one or two rather large obcordate leaves present. Comparing the series of specimens just dealt with, on the whole each specimen shows more uniformity than in the Kaukau-Crowsnest series, but this may be due, in part, to the specimens being smaller. There does not seem to be any evidence of a fixed race being present. 41. Senecio (the coastal species of West Wanganui Inlet, hitherto referred to S. rotundifolius (Forst. f.) Hook. f. In Trans. N.Z. Inst., vol. 39, p. 446, 1907, Cheeseman records, without special comment, Senecio rotundifolius as growing at West Wanganui Inlet. Earlier (Manual, p. 383, 1906) the same botanist had referred the shrub common near Cape Foulwind to the same species. More recently Petrie (Trans. N.Z. Inst., vol. 46, p. 30, 1914) has referred, “without hesitation,” the Cape Foulwind plant to Senecio elaeagnifolius Hook. f., stating, “I examined a great many specimens of this shrub [the Cape Foulwind plant], and found that the leaves, though more coriaceous than usual, are nearly always longer than broad, in outline more or less ovate or ovateoblong, and not rarely waved or repandly toothed at the margin. At an elevation of 2,300 ft. or so on the Longwood Range, Southland, specimens of S. elaeagnifolius may be seen in the forest with leaves approaching those of S. rotundifolius much more closely than do any to be found near Cape Foulwind.” Recently I have received from Mr. B. C. Aston a living specimen, but not in flower, of the West Wanganui Senecio. This specimen I should certainly hesitate to refer either to S. rotundifolius or to S. elaeagnifolius. For instance, it differs at once from both species in that the branchlets, the very youngest excepted, are covered with a smooth, not grooved, purple bark altogether lacking tomentum but, in part, most minutely pubescent, whereas similar, or indeed older, branchlets of the other two species are densely covered with pale-buff tomentum. Also, the tomentum, of the under-surface of the leaves is rather thin, white and not buff as in S. rotundifolius and S. elaeagnifolius, and puts one in mind of that of Olearia arborescens (Forst. f.) Cockayne and Laing. The youngest branchlets are covered more or less with a white pellicle of hairs, through which the purple bark soon becomes visible. In addition, the shrub, according to Mr. Aston's notes, is only a few feet high, and not almost a tree as is Senecio rotundifolius. But this low stature may be caused by the plant growing on cliffs, the only habitat where it was noted. As for shape, none of the leaves are rotund, but obovate and oblong are the commonest forms. The leaf-base is slightly unequal and cuneate, but such a base is shown in specimens of Senecio rotundifolius which I collected at Anita Bay (Fiord Botanical District). The midrib is much keeled, sometimes glabrous or almost so for its entire length, or it may be covered by a thin pellicle of white hairs.

From the above it seems clear that the taxonomic position of the West Wanganui Inlet plant is quite uncertain; it may indeed be, as Mr. Aston suggests, an undescribed species. It is certainly not typical Senecio rotundifolius, the northern limit of which appears to be Jackson's Bay, as stated in the Manual. Neither is it typical S. elaeagnifolius. At present it seems best to look upon it as belonging to a group confined to the North-western Botanical District which may be either a distinct species or a variety of either S. elaeagnifolius or S. rotundifolius. 42. Sophora prostrata Buch. Buchanan described Sophora prostrata as a “rigid prostrate rambling shrub 12–18 in. high” (Trans. N.Z. Inst., vol. 16, p. 395, 1884). Kirk and Cheeseman both describe it as prostrate. That it frequently is quite prostrate and but a few centimetres high is true enough, but this extreme degree of depression appears to be due entirely to its environment. The really important characters which distinguish the species from any of the other species of Sophora are its divaricating growth-form, the smaller flowers solitary or in pairs, the standard almost equalling the wings, and the small pod with few seeds. Specimens more than 1.8 m. high, and probably much higher, are common enough, but they never grow out of the divaricating growth-form stage of development. The species reproduces itself true from seed. The actual southern limit of Sophora prostrata has not been ascertained, but it does not seem to extend into the North Otago Botanical District, although the conditions there are ideal for its requirements. Personally, I have not seen it on the Canterbury Plain south of the Rakaia River, nor in the mountainous area to the south of the Waimakariri River basin. According to Cheeseman (Manual p. 123), S. prostrata is found in the mountains. But it also occurs in the lowland belt, nor do I think it ascends much above 900 m. 43. Veronica salicifolia Forst. f. var. angustissima Cockayne var. nov. Foliis lineari-lanceolatis, racemis longis gracilibus laxifloris, calycis lobis anguste lanceolatis acutis. North Island: Ruahine-Cook Botanical District—Otaki Gorge, base of Tararua Mountains, on more or less stony ground. L. C. This variety is distinguished at once from any other described variety of Veronica salicifolia by the following combination of characters: Linearlanceolate thin leaves ± 10 cm. long by ± 10 mm. broad, slender racemes ± 17 cm. long with the flowers rather distant, the rhachis and pedicels most minutely pubescent, the deeply-cut calyx almost as long as the corolla-tube with long narrow lanceolate acute segments, the white corolla, and the far-exserted purple anthers. When not in bloom the plant might be easily confused with broad-leaved forms of Veronica angustifolia A. Rich., but from that species it is at once separated by the much longer racemes, the larger calyx, and the narrow acute not obtuse calyx-segments. III. Phytogeographic. Although the localities given below have for the most part not been recorded hitherto, in this series of papers the station of a species, though already published, if it is not generally known, will in certain cases be

given. Another class of species, the distribution of which requires defining with much more detail, includes those mentioned in the Manual as “often local,” “probably not uncommon but easily overlooked,” and similar statements. The distribution of varieties needs far greater attention than is usually accorded them. The variety and not the aggregate species being the real entity, until varietal distribution is put on a sounder footing any truly scientific discussion of the distribution in general of the New Zealand flora will not be possible. What are really wanted from the evolutionary and historical standpoints are accurate records both of the distribution of the aggregate species and of the microspecies which compose them; but before such a study is possible many so-called “valid species” must be joined together as aggregates, while many more varieties of the present aggregates must be constituted. The crying need of New Zealand floristic botany on the phyto-geographical side is undoubtedly a much more intensive study of distribution. Full lists of species, & c., for unbotanized localities are demanded. The critical points, where two botanical districts join one another, require a most intensive study. These boundaries at present are nearly all highly problematical, even in areas apparently well known. The botanical hunt should not be merely for “something new”: the presence or absence of the commonest species is generally a matter of far greater importance than the finding of a rarity. 1. Acaena saccaticupula Bitter. South Island: North-eastern Botanical District—(1.) Eastern part of Hurunui River basin: A. Wall! (2.) Trelissick Basin, Canterbury, at about 900 m. altitude: A. Wall! This species is Acaena adscendens, in part, of the Manual. It is apparently common on the eastern side of the Southern Alps generally, but is not usually found in great abundance in any station. It appears to occur chiefly in wettish ground. Exact information as to its ecological requirements, distribution, and polymorphy is wanted. 2. Acaena Sanguisorbae Vahl. var. viridior Cockayne. South Island: (1.) Sounds Subdistrict of Ruahine – Cook Botanical District — Near the Nelson City waterworks: L. C. (2.) North-eastern Botanical District—In the Kaikoura neighbourhood: A. Wall! Most likely this well-marked variety is of wide distribution, but so far it has not been recorded to the north of the neighbourhood of the city of Wellington or to the south of Banks Peninsula. 3. Angelica geniculata (Forst. f.) Hook. f. South Island: Eastern Botanical District—(1.) On rock, Malvern Hills: A. Wall! (2.) On limestone rock at junction of the River Porter and the Broken River, Trelissick Basin: A. Wall! This species has now been reported far inland in various parts of New Zealand, and can no longer rank as a special coastal plant. 4. Apium prostratum Labill. var. South Island: Eastern Botanical District — By side of stream on wet bank near Scargill, Canterbury, at about seven miles from the sea. A. Wall and L. C.

5. Asperula perpusilla Hook. f. North Island: Volcanic Plateau Botanical District—Tall tussock grassland (“grass-steppe” of my Report on a Botanical Survey of the Tongariro National Park), Waimarino Plain. H. Carse. 6. Asplenium Colensoi Hook. f. North Island: Ruahine-Cook Botanical District — Common on moist shady banks near streams in forest of Mount Crowsnest, Wellington. L. C. 7. Carex dipsacea Berggren. North Island: Volcanic Plateau Botanical District—By the roadside, Waimarino. H. Carse. 8. Carex Solandri Boott. North Island: Volcanic Plateau Botanical District—Margin of forest, Waimarino. H. Carse. 9. Carmichaelia grandiflora Hook. f. South Island: Eastern Botanical District—(1.) Mount Torlesse: A. Wall! (2.) Mount Hutt: A. Wall! Professor Wall's specimens were too small for me to refer them to the special variety of the species to which they belonged. 10. Celmisia Armstrongii Petrie. South Island: Western Botanical District—Mount Tuhua, in herb-field, subalpine. J. E. Holloway! Mount Tuhua is a peak, 1,093 m. high, situated to the east of Lake Kaniare, near Hokitika, and distant from the sea about fifteen miles. Its flora was unknown until the Rev. Dr. Holloway sent me a small collection of plants collected by him mostly above the forest-line. This collection shows the flora to be much the same as that at similar altitudes on the actual Divide. 11. Celmisia intermedia Petrie. South Island: Western Botanical District—Mount Tuhua, in herb-field, subalpine. J. E. Holloway! 12. Claytonia australasica Hook. f. South Island: Western Botanical District — Styx Valley. J. E. Holloway! 13. Coprosma foetidissima Forst. North Island: Ruahine – Cook Botanical District — Near Makerua Railway-station, on sandstone bluff. L. C. 14. Coprosma serrulata Hook. f. South Island: Western Botanical District—Browning's Pass. J. E Holloway!

15. Corallospartium crassicaule (Hook. f.) J. B. Armstg. South Island: Eastern Botanical District — Mount Hutt, subalpine. A. Wall! 16. Donatia novae-zelandiae Hook. f. South Island: Western Botanical District—Mount Tuhua, subalpine. J. E. Holloway! 17. Dracophyllum Kirkii Berggren. South Island: Western Botanical District — Browning's Pass. J. E. Holloway! 18. Drapetes villosa (Berggren) Cheesem. South Island: Western Botanical District—Mount Tuhua, subalpine. J. E. Holloway! 19. Epilobium chionanthum Hausskn. South Island: Eastern Botanical District — Swamp on Waimakariri River bed, on the Craigieburn Run. A. Wall! 20. Euphrasia cuneata Forst. f. North Island: Ruahine-Cook Botanical District—(1.) Near Plimmerton, in a remarkable subassociation of a Typha-Phormium swamp where Leptocarpus simplex is dominant: L. C. (2.) Waikanae, on Sphagnum: W. H. Field. 21. Forstera Bidwillii Hook. f. South Island: Western Botanical District—Mount Tuhua, subalpine. J. E. Holloway! 22. Gaultheria perplexa T. Kirk. South Island: Eastern Botanical District — Near the Rakaia Gorge. A. Wall! 23. Gentiana serotina Cockayne. South Island: Eastern Botanical District—Hills near “The Point,” Rakaia Gorge. A. Wall! 24. Gleichenia Cunninghamii Heward. South Island: Eastern Botanical District — Forest at base of Mount Hutt. A. Wall! The localities previously known for this fern in the Eastern Botanical District are: Mount Peel; Alford Forest; Banks Peninsula, especially near Port Levy (T. H. Potts, Out in the Open, p. 53, 1882). 25. Isotoma fluviatilis (R. Br.) F. von Muell. South Island: (1.) Eastern Botanical District—Shore of Lake Rubicon, Mount Torlesse: A. Wall! (2.) North-eastern Botanical District—Awatere River basin, subalpine: C. E. Foweraker and L. C. This species, first recorded for New Zealand in the Manual, p. 401, is now known to occur in all the botanical districts of the South Island excepting the Western and Fiord Districts.

26. Lobelia anceps L. f. North Island: Ruahine-Cook Botanical District—On drained ground of the Makurerua Swamp, but apparently not common. L. C. 27. Lycopodium cernuum L. South Island: North-western Botanical District—Near Westhaven (West Wanganui). B. C. Aston! This discovery of Mr. Aston's is of considerable phytogeographical importance, since it extends the southern range of L. cernuum from the neighbourhood of Lake Taupo (Volcanic Plateau Botanical District) for a distance of about eighty-four miles. It also adds another species to the following remarkable list of plants which occur in the North-western Botanical District, but which otherwise are confined to the Northern Botanical Province or extend only a short distance beyond its southern boundary: Astelia Banksii, Adiantum aethiopicum, Blechnum Fraseri, Dracophyllum latifolium, Schoenus tendo, and Pterostylis puberula. 28. Lygodium articulatum A. Rich. North Island: Volcanic Plateau Botanical District—In forest, Waimarino; apparently rare. H. Carse. The Manual gives the Bay of Plenty and Kawhia as the southern limits of this fern. The above record extends its southern range considerably. 29. Metrosideros lucida (Forst. f.) A. Rich. South Island: Eastern Botanical District — In patches of forest in gullies on hills near “The Point,” Rakaia Gorge. A. Wall! 30. Myosotis Townsoni Cheesem. South Island: Western Botanical District — Browning's Pass. J. E. Holloway! This well – marked species has hitherto been recorded only from the Brunner Range and Lyell Mountains, in the North-western Botanical District; its known range is thus extended about sixty miles to the south. 31. Notospartium torulosum T. Kirk. South Island: Eastern Botanical District—In the vicinity of the Rakaia Gorge. A. Wall! I have only a mere scrap, but it seems identical with Kirk's type. The species has now been recorded from the above locality, from Mount Peel (where it was recently rediscovered by Mr. R. M. Laing, B.Sc.), the Waikari Hills, the Hanmer Plains area, and the vicinity of the River Mason. Mr. D. Petrie, M.A., suggested to me some time ago that the Clarence Valley plant discovered by Mr. Aston was possibly neither the above nor Notospartium Carmichaeliae; and he may quite well be right, as its much-swollen pod looks very distinct. 32. Olearia Colensoi Hook. f. South Island: Western Botanical District — Mount Tuhua; an important member of the subalpine scrub. J. E. Holloway!

33. Ourisia macrocarpa Hook. var. calycina (Col.) Cockayne. South Island: Western Botanical District—Mount Tuhua, in herb-field. J. E. Holloway! This variety has been recorded by Mr. D. L. Poppelwell from as far south as the mountains near the Haast Pass, but its exact southern limit—i.e., where it is replaced by var. cordata (the type of the species)—is not yet known. 34. Plagianthus cymosus T. Kirk. South Island: South Otago Botanical District—Banks of Waihopai Stream, near Invercargill. J. Crosby Smith! As Mr. Crosby Smith's record of this interesting plant in his list of Southland plants (Trans. N.Z. Inst., vol. 46, p. 223, 1914) may be easily overlooked, I am calling attention to this station. 35. Pseudopanax lineare (Hook. f.) C. Koch. South Island: Western Botanical District—In subalpine scrub of Mount Tuhua. J. E. Holloway! 36. Ranunculus chordorhizos Hook. f. South Island: Eastern Botanical District—Mount Hutt, on subalpine shingle-slip. A. Wall! 37. Ranunculus Enysii T. Kirk. South Island: Eastern Botanical District—Mount St. Bernard. H. Wall! R. Enysii, according to the Manual, is said to occur not only in the Waimakariri River basin, but also, without there being any stations intermediate, on the East Taieri Hills (South Otago Botanical District) and near Lake Harris (Fiord Botanical District). The Taieri station is given on the authority of Buchanan, his Ranunculus tenuis from that locality being considered by Cheeseman as a form of R. Enysii with the leaves more pinnately divided than usual. But R. tenuis Buch. includes not only the Taieri plant but one from Masterton (Ruahine-Cook Botanical District), while the figure (Trans. N.Z. Inst., vol. 20, pl. xii, 1888) does not match any form of R. Enysii from its original habitat; therefore I think the Taieri habitat should not be accepted. I would also exclude the Lake Harris and Masterton (probably Tararua Mountains) habitats. Should it eventually be proved that I am right, then the species under consideration is, on our present knowledge, confined to the Waimakariri River basin and to the south-eastern portion of the Hurunui River basin, where Professor Wall recently collected it. Two very distinct forms of the species were collected by Wall on Mount St. Bernard, and I have also in my herbarium and garden several well-marked forms, but I await cultivation tests before going into the matter of varieties in this rather puzzling aggregate species. J. B. Armstrong (Trans. N.Z. Inst., vol. 12, p. 336, 1880) includes Ranunculus geraniifolius Hook. f. in his catalogue of the plants of Canterbury, but is seems almost certain that the plant he had in mind was R. Enysii, to which R. geraniifolius bears no small resemblance.

38. Ranunculus insignis Hook. f. North Island: Volcanic Plateau Botanical District—Mount Ngauruhoe, on wet lava cliffs. H. Carse. This species is not mentioned in my Report on a Botanical Survey of the Tongariro National Park, but it has since been noted by Mr. Allison, of Wanganui, on the south-eastern side of Ruapehu; by Mr. E. Phillips Turner in the bed of the Maungaturuturu River; and by Mr. Carse as above. All the same, it appears to be an uncommon plant for the central group of volcanoes in general and the adjacent part of the Volcanic Plateau. 39. Raoulia glabra Hook. f. North Island: Ruahine-Cook Botanical District—On summit and other stony exposed places on the Kaukau Range. L. C. In Aston's catalogue of Wellington plants (Trans. N.Z. Inst., vol. 43, p. 235, 1911) the only localities given for R. glabra are the Rimutaka and Tararua Mountains. 40. Rubus parvus Buchanan. South Island: Western Botanical District—Styx and Arahura Valleys. J. E. Holloway! In the Manual the Taramakau Valley is given as the southern limit of Rubus parvus. It is, however, now known to extend almost to the Fox Glacier, and probably it extends still farther to the south. It appears, indeed, to be fairly common on old river-bed, though perhaps somewhat local, throughout the North-western and Western Botanical Districts. Poppelwell does not record its occurrence in the neighbourhood of the Haast Pass or the River Haast. 41. Scirpus inundatus Poir. var. North Island: Volcanic Plateau — Wet ground on Waimarino Plain. H. Carse. 42. Selliera radicans Cav. South Island: Eastern Botanical District—(1.) Near the junction of the Porter River and the Broken River, Trelissick Basin: A. Wall! (2.) On the shores of certain of the small lakes (Marymere, & c.) and the slopes adjacent, near Mount St. Bernard: A. Wall! The Manual gives only quite general information regarding the inland distribution of this extremely common coastal plant. Aston (loc. cit., p. 236) states that it ascends to 3,500 ft. on the Kaimanawa Mountains. Petrie records it (Trans. N.Z. Inst., vol. 28, p. 565, 1896) as “rare inland [Otago] and much reduced in size, as at Lakes Wanaka and Te Anau. Ascends to 1,000 ft.”; and the Manual, p. 395, “ascending to over 2,500 ft. at the base of Ruapehu.” Wall's specimens agree with Petrie's remarks as to reduction in size. 43. Urtica ferox Forst. f. North Island: Volcanic Plateau Botanical District—On bank of river, Makatote Gorge. H. Carse.

44. Urtica linearifolia (Hook. f.) Cockayne. North Island: Ruahine-Cook Botanical District — Makurerua Swamp, where drained, climbing over shrubs. L. C. 45. Veronica amplexicaulis J. B. Armstg. South Island: Eastern Botanical District — Mount Peel, subalpine. H. H. Allan! Previously I have only known this species from the cultivated plants in the Christchurch Botanical Gardens, which were probably cuttings from the original plant. Also, in the Manual Armstrong's original habitat, “Upper Rangitata,” is the only one given. Mr. Allan's specimens have not yet bloomed in my garden, but they seem to match exactly the not-readily-mistaken vegetative form of V. amplexicaulis, of which I have also a cultivated example. 46. Veronica Haastii Hook. f. var. macrocalyx (J. B. Armstg.) Cheesem. South Island: Western Botanical District—Browning's Pass neighbourhood. J. E. Holloway! I am inclined to think it would be better to treat this variety as a species. If not, then V. epacridea should also be united to V. Haastii. The plant in question has so far been recorded only from Mount Rolleston, the vicinity of the Waimakariri glaciers, the Rangitata Valley, and the above locality.