form of Gigartina for reasons given that we shall presently consider. Now from Agardh's description it would appear that he had chiefly young specimens to deal with. True, the photographs of the type show certain fragments of old fronds; but evidently Agardh had not enough specimens of these and of the winter form to be able to form an adequate idea of their characters. We shall now discuss the questions raised by him. (1) We cannot find anything in his description of the microscopic structure of the frond to indicate that it should not be placed in the genus Gigartina. He refers to certain “maculas rotundas” in his sections. These are probably only cross sections of the strands. We have made a number of sections of juvenile, mature, and old fronds, and though in some cases the anastomosing threads are shrunken and in others stuffed with cell contents, we can find nothing in the structure that is not typical of Gigartina. The structure is of course quite different from that of Aeodes. (2) The tribe is founded on the supposed apod nature of the frond: “frondibus supra scutellum radicale immediate expansis.” While it is quite true that this is the case in young fronds superficially resembling Aeodes, this is not the case in mature fronds in which the stipe is as well developed as in G. circumcincta. In mature tetrasporic forms, the stipe often passes into a sub-cuneate base and in cystocarpic fronds it is often branched, each branch giving rise to an undivided frond. Divided fronds are rare, though it is occasionally possible to find one which is cleft. Stipes 1 cm. to 2 cm. in length are not uncommon, occasionally channelled (perhaps as a result of drying) and bearing several undeveloped leaflets. It will therefore be seen that no good characters can be founded on the form of the stipe in the mature plant, though certainly the juvenile form of it appears characteristic. Only those who have watched the plant through all seasons of the year would associate together all the forms included in the species. It is not therefore to be wondered at that Agardh, who had not seen the plant growing, should be unable to give a complete description of it. We have collected scores of specimens at various seasons from the back of the Gladstone Pier, Lyttelton, obviously of the one species, and a few from other localities. We should certainly have regarded the forms as belonging to different species if we had seen only a few specimens. (3) Agardh further remarks in his notes: “Attamen cystocarpia tantum in una pagina, provenientia mihi adparuerunt, ex qua adparentia forsan esse decumbentem, et poros, quos juniores vidi tantum caecos et in una pagina fructiferos aperta, margine vero recurvato hujus pagina cinctos, demum grandescentes et frondem rite pertusam linquentes-haec omnia speciem vario respectu ab aliis Gigartinis diversam indicare, non potui quin monerem.” We find it difficult to understand these remarks. (a) The cystocarps confined to one side of the frond. This is not an uncommon characteristic with certain species of Gigartina; but it is a character which again has to be used with discretion. In G. atropurpurea the cystocarps are sometimes fairly equally distributed on both sides of the frond; but often specimens may be found in which the distribution is quite unequal. In other species, specimens may be found in

which not a single cystocarp is to be found on the lower side of the frond, though the other side is closely covered with them; or again, in the same species, parts of each side of the frond may be cystocarpic, but these parts are not so opposite to each other, the plant being unable to produce cystocarps at the same time on different sides immediately opposite to each other. If prodúced, they would probably result in the destruction of that portion of the frond owing to the number of pores left by the fallen or decayed cystocarps. (b) Most of the foliose Gigartinas have perforated fronds when old, owing to the decay of the cystocarps. Probably Agardh had not come across such specimens. This is particularly the case with. G. atrpurpurea and G. circumcincta. We can see no reason for regarding the plant as in any way an aberrant species of Gigartina.

Gigartina apoda J. Ag. (1) In the juvenile condition of G. apoda found chiefly in winter and spring, the stipe is only a millimetre or two in length and develops usually into a more or less orbicular undivided frond. The margin is at first entire but later dentate or irregularly serrate. The stipe is often naked but may bear one or more ligulate undeveloped fronds. The mature frond when dried may be as much as 40 cm. to 50 cm. across; but is usually smaller (15 cm. to 25 cm.), thin, in younger stages almost membranous and often rose-red. Though typically orbicular, it is sometimes elliptical or even ovate, flat or at other times cochleate with a thickened rim round the margin. (2) The mature cystocarpic form is found at all seasons of the year, except perhaps in early spring, when however, old decaying perforate cystocarpic fronds are not uncommon. The chief characteristics are (1) the narrow sterile intra-marginal band; (2) the cystocarps never appearing simultaneously on exactly opposite parts of the frond, and often confined more or less to one side only; (3) the pitted surface below the cystocarps; (4) the cystocarps solitary at first, but becoming bunched, fascicled, or borne on irregularly branched peduncles as the season advances; and finally (5) the usually orbicular or very broadly expanded form of the frond. The cystocarps are of the usual type, generally from 1 mm. to 1.5 mm. in diameter in dried specimens. In old fronds there is a cystocarpic fringe round the margins. The colour of the mature fronds is usually a fairly uniform dark brown when dry. Younger dried fronds are much lighter in colour, and often have a yellowish or greenish yellow tinge. Tetrasporic plants are much less abundant than cystocarpic, and do not show any definite intra-marginal sterile band. The few we have were collected in April, and are not sufficient in number to generalise upon. They show a definite tendency to the development of a subcuneate base, not present in the cystocarpic form. The margin is entire, becoming later coarsely toothed or even somewhat foliolate.

Mason Bay (Stewart Island) one slightly doubtful specimen; Timaru (North Mole), Lyttelton (Gladstone Pier, abundant), New Brighton (drift), Gore Bay, Breaker Bay (Wellington), R.M.L.; the Chathams (? Collector).

Tribe II.—Homaeopodae.

Gigartina grandifida J. Ag. Species excludenda.

This appears to us to be, in part at least, only one of the forms of G. lanceata J. Ag. We have a photograph of the type specimen and a single additional specimen. The plant was apparently described from specimens collected by Travers at the Chathams. The description appears in J. Agardh 1876 (pp. 199 and 200). It is quite evident that here again the author had an insufficiency of material, and so was unable to recognise the wide range of forms that may occur in a single species. The photograph shows G. lanceata split above the base into two widely divaricating fronds of the usual type, cystocarpic, and bearing along the margin at least the usual bent processes. Now though we do not happen to have any specimens of G. lanceata exactly matching these—and it is difficult to find two specimens exactly alike—yet we have several with widely divaricating branches some distance above the base, and we cannot see in this inconstant character any reason for separating out a fresh species, G. grandifida.

But G. lanceata and G. grandifida are placed by Agardh not only in different tribes, but in different sections of the genus. To us it would appear that they are so closely allied as to be collateral species at least, if not actually the same. G. lanceata is classified with G. stiriata, G. radula, and G. Burmanni because it sometimes has a channelled stipe. Yet though it is true that sometimes in dried specimens particularly, the stipe of G. lanceata is channelled, the character is obviously here valueless for the purpose of determination. This being so, there is no reason why G. grandifida and G. lanceata should be placed in different sections of the genus, because of alleged differences in this respect. In G. grandifida the stipe is stated to be flattened, not channelled. Such points are of course difficult to make out from a photograph, but to us the photographs of both species seem to show an equal amount of channelling of the stipe. The only other distinction that we can find between them in the descriptions of the species has already been alluded to. In G. grandifida the lamina is said to be bifid, simple, or forked; in G. lanceata it is linear lanceolate. We have seen hundreds of plants of G. lanceata, and amongst these are a few which correspond with Agardh's description of G. grandifida, but we do not regard these as being sufficiently numerous or distinctive to constitute a good species; and we therefore propose that G. grandifida be excluded from the list of the species of Gigartina.

It is also probable that certain forms of G. circumcincta and G. atropurpurea are included in Agardh's G. grandifida. In J. Agardh 1885, p. 50, it is stated of this species, “Plurimis affinibus tenuior et colore plerumque dilute rosea aut pallido plantae exsiccatae diagnoscenda.” As we have already pointed out, colour and texture must be used with discrimination in the determination of a species. There is in the Otago Museum a fragment of G. grandifida identified by Agardh; but it is sterile and quite insufficient to provide means of definite determination. It has the pale rose colour and thin

texture insisted upon in Agardh's description, and exactly matches certain plants of G. atropurpurea collected by the senior author at Akaroa.

Further, the photograph of the type that we have seen is scarcely sufficiently definite to show the texture of the plant photographed. If thin, the type specimen might probably be a form of G. circumcincta. We come therefore to the conclusion that G. grandifida as a species probably does not exist; but that the group as defined by Agardh probably includes forms of G. lanceata and possibly also of G. circumcincta and G. atropurpurea.

Gigartina rubens J. Ag.

Agardh in 1876, p. 685, described two varieties of G. grandifida. The second of these, var. b. latifolia he subsequently (1885), p. 31, distinguished with the specific name of rubens, describing it in (1899) p. 34. The specimens are from the west coast of New Zealand, doubtless collected by Captain Fairchild and presented to Agardh by Baron F. von Mueller. There is little or nothing in the appearance of the photographs of the type to distinguish it from G. circumcincta except possibly the cordate base. Specimens of G. circumcincta with a cordate base are, however, not lacking. Why Agardh should in the first place have associated this plant with G. grandifida is not clear. Indeed, he himself says (loc. cit.), p. 34: “Ejusmodi formae et ramificationis differentias, quibus omnes ad species dignoscendas conatus—si quidem characteribus ex frondis forma et ramificationis norma petitis insisterent—repulsos fere diceres, non potui quin stupens adverterem.” Now the characters given in the diagnosis are a cordate base, passing into an ovae, reniform frond, branching into two or three forks, with lobes growing sub-pinnately (from the margins). Yet the photograph of the type does not show any sub-pinnate lobes, and of the three fronds shown, two are once divided and the other is practically undivided. However, the fronds in the photographs do not show an intra-marginal sterile band under a lens, and so the plant would seem to differ from G. circumcincta, though it is questionable how much stress should be laid on this peculiarity. We have certain non-circumcinct specimens found growing on empty shells at Purau (Lyttelton) which seem to match well with Agardh's description, and fairly well with his photographs, but as they are few in number and the form has not again been collected, we hesitate to identify them definitely with G. rubens. The species must therefore be left in our list of species inquirendae.

Agardh's diagnosis is as follows:—“Stipite brevi juvenili aut fere immediate in laminam 2–3 furcam abeunte, aut evolutione tum laminae a cordata basi in folium ovato reniforme abeuntis, tum lobis excrescentibus frondem subpinnatim decompositam gerentibus.”

Tribe III.—Brachypodae.

These species are characterised by a short petiole surmounted usually by a broad, little divided frond, lanceolate to orbicular. On the short petiole there are generally several small undeveloped

fronds. They are further characterised in the cystocarpic specimens at least by a sterile inter-marginal band. Cystocarps are developed more or less evenly on both sides of the frond or confined to different parts of opposite sides. Agardh recognises three species, G. circumcincta, G. orbitosa, and G. gigantea. Of these, G. circumcincta is an abundant well defined species. G. orbitosa is separated from G. circumcincta chiefly by the form of the frond, the size, the more cartilaginous texture, and the colour tending to purple. Now none of these characters is distinctive, and we propose to include G. orbitosa with G. circumcincta. Indeed, we cannot separate it even as a variety. We have a large series of forms of G. circumcincta collected between Timaru and Wellington; and these include specimens which match the photograph of the type specimen of G. orbitosa, yet they all apparently belong to one species. Agardh originally included his specimens of G. orbitosa with G. circumcincta, and only made the separation in his latest writings. We have further the photograph of the type specimen of G. gigantea, labelled “Banks Peninsula, New Zealand, Berggren.” Now Agardh (1899), p. 37, does not record G. gigantea from New Zealand, but from Tasmania and Australia, apparently overlooking the fact that his type was from Banks Peninsula. The character which chiefly separates G. gigantea from G. circumcincta is the development of marginal folioles and laciniae. G. circumcincta is usually entire or only once or twice divided. We have, however, several such Gigartinas with marginal folioles and laciniae, such as are represented in the photograph of the type of G. gigantea. These specimens were found at Lyall Bay and Kaikoura, in company with undoubted specimens of G. circumcincta, and we cannot believe that they represent a separate species, but only such modifications of the form of G. circumcincta as may well be expected in such a polymorphic genus. If the name G. gigantea is retained at all, it should only be as a form of G. circumcincta. We have also a small fragment—insufficient for identification—of a Gigartina 18 inches across, from Eagle Hawk Neck, Tasmania. This was given us by Mr A. H. S. Lucas, and by him labelled G. gigantea, but though cystocarpic, there is insufficient material for a satisfactory comparison with New Zealand forms.

22. Gigartina circumcincta J. Ag. (Pl. 19: Figs. 11–12).

This species was founded by J. Agardh (1876), p. 202, to replace the G. radula from New Zealand of other authors. We have already considered their relationships under G. radula (sp. excl.), and the general characters of G. circumcincta have just been given under the section Brachypodae. It now remains to consider them in somewhat more detail. As in other forms, a considerable amount of polymorphy is shown, but the species is usually recognisable. It grows (at Kaikoura) up to 1 metre in length and 20 cm. in breadth. From the scutellum it usually passes at once into a short, terete stipe which may be only 2 or 3 millimetres in length, then into a short cuneate base which expands into the frond. The stipe usually carries several ligular expansions; these

however in some cases are developed into folioles. At times too, the stipe is almost wanting and the frond expands immediately above the scutellum; or again the terete petiole may be several centimetres in length; but the base is rarely if ever so well developed as in G. atropurpurea or G. lanceata. The frond is typically lanceolate to ovate or elliptical, more rarely rotundate, usually narrowed and rounded towards the apex. It is frequently undivided, but sometimes forked. The margin may be irregularly toothed, lobed, or entire. Processes such as are characteristic of G. lanceata and G. radula are usually wanting. We have, however, several old specimens having linear processes on the surfaces and margin, 3 mm. to 5 mm. long, with cystocarps at the apices. The cystocarps are, however, solitary normally sub-sessile, sometimes abundant on both sides of the frond, but more often in patches with a sterile area immediately opposite on the other side. The sterile marginal band is usually only 2 mm. to 3 mm. in width and is sometimes lacking in old and worn fronds. Cystocarps up to 1 mm. in diameter are produced in immense numbers without terminal or lateral processes, almost sessile or shortly petiolate, very rarely with divided petioles. Tetrasporic forms are similar to the cystocarpic, usually with entire, sometimes with erose margins. The fronds of both cystocarpic and tetrasporic plants are thin and membranous, rather than cartilaginous. The colour is usually dark brown to reddish brown. Hybrids probably occur between this and adjacent species, such as G. atropurpurea and G. apoda where both species are present. This is one of the commonest of the foliose species, and has been found at many points on the east coast between Dunedin and the Bay of Islands. Plants from the Auckland and the Campbell Islands recorded by the senior author (1909), p. 506, as G. radula apparently belong to this species, but are in poor condition. Cystocarpic and tetrasporic plants, September to May.

Dunedin, Timaru, Lyttelton, Double Corner (Amberley), Gore Bay, Kaikoura, Wellington Heads, R.M.L. and H.W.G.; Waipu Cove (E. W. Blackwell!), Bay of Islands (A. McMillan!), Muriwai Bay (M. C. Crookes!).

Tribe IV.—Heteropodae.

23. Gigartina longifolia J. Ag. (Pl. 19a: Figs. 13–15).

This species was based by Agardh on a few specimens only (pauca tantum specimina) and placed by him close to the species G. microphylla. Now, thanks to the good offices of Professor W. A. Setchell, we have several specimens of G. microphylla from California. According to Agardh, the terminal fronds of G. longifolia are scarcely different from those of G. microphylla, though possibly somewhat shorter, and the substance of the leaves somewhat more cartilaginous. The tribe Heteropodae contains those members of the genus which bear laminae in a great variety of forms and which are clothed with lingulate appendicules. Further, the inramarginal region is stated to be sterile. Now we have a few speci-