The miscellaneous botanical works of Robert Brown/Volume 1/Observations on the herbarium collected by Professor Christian Smith

by Robert Brown (compiled by John Joseph Bennett)

OBSERVATIONS,

SYSTEMATICAL AND GEOGRAPHICAL,

ON

THE HERBARIUM

COLLECTED BY

PROFESSOR CHRISTIAN SMITH,

IN THE

VICINITY OF THE CONGO,

DURING THE EXPEDITION TO EXPLORE THAT RIVER,

UNDER THE

COMMAND OF CAPTAIN TUCKEY,

IN THE YEAR 1816.

BY
ROBERT BROWN, F.R.S.,
CORRESPONDING MEMBER OF THE ROYAL INSTITUTE OF FRANCE, AND OF THE
ROYAL ACADEMIES OF SCIENCES OF BERLIN AND MUNICH: HONORARY
MEMBER OF THE LITERARY AND PHILOSOPHICAL SOCIETY OF
NEW YORK, MEMBER OF THE WERNERIAN SOCIETY
OF EDINBURGH. LIBRARIAN TO THE
LINNEAN SOCIETY.

[Reprinted from a "Narrative of an Expedition to explore the River Zaire,"
pp. 420–485.]

LONDON:

---

1818.

OBSERVATIONS, &c.

[420

---

The Herbarium formed by the late Professor Smith and his assistant, Mr. David Lockhart, on the banks of the Congo, was, on its arrival in England, placed at the disposal of Sir Joseph Banks; under whose inspection it has been arranged; the more remarkable species have been determined; and the whole collection has been so far examined as the very limited time which could be devoted to this object allowed.

In the following pages will be found the more general results only of this examination; descriptions of the new genera and species being reserved for a future publication.

In communicating these results I shall follow nearly the same plan as that adopted in the Botanical Appendix to Captain Flinders's Voyage to Terra Australis:

1st. Stating what relates to the three Primary Divisions of Plants.

2dly. Proceeding to notice whatever appears most remarkable in the several Natural Orders of which the collection consists; and

3dly. Concluding with a general comparison of the vegetation on the line of the river Congo, with that of other equinoctial countries.

1. The number of species in the herbarium somewhat exceeds 600; the specimens of several of which are, indeed, imperfect; but they are all referable with certainty to the primary divisions, and, with very few exceptions, to the natural orders to which they belong.

Of the Primary Divisions, the Dicotyledonous plants amount to 460.

The Monocotyledonous to 113

And of the Acotyledonous, in which Ferns are included, there are only 33 species.

It is a necessary preliminary, with reference especially to the first part of my subject, to determine whether this herbarium, which was collected in a period not exceeding two months, and in a season somewhat unfavourable, can 421] warrant any conclusions concerning the proportional numbers of the three primary divisions, or of the principal natural orders in the country in which it was formed.

Its value in this respect must depend on the relation it may be supposed to have to the whole vegetation of the tract examined, and of the probability of the circumstances under which it was formed, not materially affecting the proportions in question.

Its probable relation to the complete Flora of the country examined, can at present be judged of only by comparing it with collections from different parts of the same coast of equinoctial Africa.

The first considerable herbarium from this coast, of which we have any account, is that formed by Adanson, on the banks of the Senegal, during a residence of nearly four years. Adanson himself has not given the extent of his collection, but as he has stated the new species contained in it to be 300,^[1] it may, I think, be inferred, that altogether it did not exceed 600, which is hardly equal to that from Congo. Limited as this supposed extent of Adanson's herbarium may appear, it is estimated on the most moderate calculation of the proportion that new species were likely to bear to the whole vegetation of that part of equinoctial Africa, which he was the first botanist to examine; allowance being at the same time made for the disposition manifested in the account of his travels, to reduce the plants which he observed to the nearly related species of other countries.

From the herbarium and manuscripts in the library of Sir Joseph Banks, it appears that the species of plants collected by Mr. Smeathman at Sierra Leone, during a residence of more than two years, amounted to 450.

On the same authority I find that the herbarium formed in the neighbourhood of Cape Coast by Mr. William Brass, an intelligent collector, consisted of only 250 species.

And I have some reason to believe, that the most extensive and valuable collection ever brought from the west coast of equinoctial Africa, namely, that formed by Professor Afzelius, during his residence of several years at Sierra Leone, does not exceed 1200 species; although that eminent naturalist, in the course of his researches, must have examined a much greater extent of country than was seen in the expedition to Congo.

From these, which are the only facts I have been able to meet with respecting the number of species collected [422 on different parts of this line of coast, I am inclined to regard the herbarium from Congo as containing so considerable a part of the whole vegetation, that it may be employed, though certainly not with complete confidence, in determining the proportional numbers both of the primary divisions and principal natural orders of the tract examined; especially as I find a remarkable coincidence between these proportions in this herbarium and in that of Smeathman from Sierra Leone.

I may remark here, that from the very limited extent of the collections of plants above enumerated, as well as from what we know of the north coast of New Holland, and I believe I may add of the Flora of India, it would seem that the comparative number of species in equal areas within the tropics and in the lower latitudes beyond them, has not been correctly estimated; and that the great superiority of the intratropical ratio given by Baron Humboldt, deduced probably from his own observations in America, can hardly be extended to other equinoctial countries. In Africa and New Holland, at least, the greatest number of species in a given extent of surface does not appear to exist within the tropics, but nearly in the parallel of the Cape of Good Hope.

In the sketch which I have given of the botany of New Holland, I first suggested the inquiry respecting the proportions of the primary divisions of plants as connected with climate; and I then ventured to state that "from the equator to 30° lat. in the northern hemisphere at least, the species of Dicotyledonous plants are to the Monocotyledonous as about 5 to 1, in some cases considerably exceeding, and in a very few falling somewhat short of this proportion, and that in the higher latitudes a gradual diminution of Dicotyledones takes place until in about 60° N., and 55° S. lat. they scarcely equal half their intratropical proportion."^[2]

Since the publication of the Essay from which this quotation is taken, the illustrious traveller Baron Humboldt, to whom every part of botany, and especially botanical geography, is so greatly indebted, has prosecuted this subject further, by extending the inquiry to the natural orders of plants; and in the valuable dissertation prefixed to his great botanical work,^[3] has adopted the same equinoctial proportion of Monocotyledones to Dicotyledones as that 423] given in the Paper above quoted; a ratio which seems to be confirmed by his own extensive herbarium.

I had remarked, however, in the Essay referred to, that the relative number of these two primary divisions in the equinoctial parts of New Holland appeared to differ considerably from those which I had regarded as general within the tropics; Dicotyledones being to Monocotyledones only as 4 to 1. But this proportion of New Holland very nearly agrees with that of the Congo and Sierra Leone collections. And from an examination of the materials composing Dr. Roxburgh's unpublished Flora Indica, which I had formerly judged of merely by the index of genera and species, I am inclined to think that nearly the same proportion exists on the shores of India.

Though this may be the general proportion of the coasts, and in tracts of but little varied surface within the tropics, it seems at the same time probable from Baron Humboldt's extensive collections, and from what we know of the vegetation of the West India islands, that in equinoctial America, in tracts including a considerable portion of high land, the ratio of Dicotyledones to Monocotyledones is at least that of 11 to 2, or perhaps nearly 6 to 1. Whether this or a somewhat diminished proportion of Dicotyledones exists also in similar regions of other equinoctial countries, we have not yet sufficient materials for determining.

Upon the whole, however, it would seem from the facts of which we are already in possession, that the proportions of the two primary divisions of phænogamous plants vary considerably even within the tropics, from circumstances connected certainly in some degree with temperature. But there are facts also which render it probable, that these proportions are not solely dependent on climate. Thus the proportion of the Congo collection, which is also that of the equinoctial part of New Holland, is found to exist both in North and South Africa, as well as in Van Diemen's Island, and in the south of Europe.

It is true indeed that from about 45° as far as to 60°, or perhaps even to 65° N. lat. there appears to be a gradual diminution in the relative number of Dicotyledones; but it by no means follows that in still higher latitudes a further reduction of this primary division takes place. On the contrary, it seems probable from Chevalier Giesecke's list of the plants of the west coast of Greenland,^[4] on different parts of which, from lat. 60° to 72°, he resided several years, that the relative numbers of the two primary divisions of phænogamous plants are inverted on the more northern parts of the coast;^[5] Dicotyledones being to Monocotyledones, in the list referred to, as about 4 to 1, or nearly as on the shores of equinoctial countries. And analogous to this inversion it appears, that at corresponding Alpine heights, both in the temperate and frigid zones, the proportion of Dictyledones is still further increased.

The acotyledonous or cryptogamous plants of the herbarium from Congo, are to the phænogamous as about 1 to 18. Some allowance is here to be made for the season, peculiarly unfavourable, no doubt, for the investigation of this class of plants. But it is not likely that Professor Smith, who had particularly studied most of the cryptogamous tribes, should have neglected them in this expedition; and the circumstance of the very few imperfect specimens of Mosses in the collection being carefully preserved and separately enveloped in paper, seems to prove the attention paid to, and consequently the great rarity of, this order at least; which, however, is not more striking than what I have formerly noticed with respect to some parts of the north coast of New Holland.^[6]

I have in the same place considered the Acotyledones of equinoctial New Holland, as probably forming but one thirteenth of the whole number of plants, while the general equinoctial proportion was conjectured to be one sixth. This general ratio, however, is certainly over-rated, though it is probably an approximation to that of countries containing a considerable portion of high land. Within the tropics, therefore, it would seem that the ratio of acotyledonous to phænogamous plants, varies from that of 1:15 to 1:5; the former being considered as an approximation to the proportion of the shores, the latter to that of mountainous countries.

425] II. The NATURAL ORDERS of which the herbarium from Congo consists are 87 in number; besides a very few genera not referable to any families yet established. More than half the species, however, belong to nine orders, namely, to Filices, Gramineæ, Cyperaceæ, Convolvulaceæ, Rubiaceæ, Compositæ, Malvaceæ, Leguminosæ, and Euphorbiaceæ; all of which have their greatest number of species in the lower latitudes, and several within the tropics.

I now proceed to make some observations on the orders above enumerated, and on such of the other families, included in the collection, as present anything remarkable, either in their geographical distribution, or in their structure; more especially where the latter establishes or suggests new affinities; and I shall take them nearly in the same order as that followed in the botanical appendix to Captain Flinders's Voyage.

ANONACEÆ. Only three species of this family are contained in the collection. One of these is Anona Senegalensis, of which the genus has been considered doubtful, even by M. Dunal in his late valuable Monograph of the order.^[7] That it really belongs to Anona, however, appears from the specimen with ripe fruit preserved in the collection. It is remarkable therefore as the only species of this genus yet known which is not a native of equinoctial America; for Anona Asiatica, of which Linnæus had no specimen in his herbarium when he first proposed it under this name, according to the original synonym, is nothing more than Anona muricata: and A. obtusiflora, supposed by M. Tussac^[8] to have been introduced into the American Islands from Asia, does not appear to differ from A. mucosa of Jacquin, which is known to be a native of Martinica.

The second plant of this order in the collection is very nearly related to Piper Æthiopicum of the shops, the Unona Æthiopica, and perhaps also Unona aromatica of Dunal:^[9] these with several plants already published, form a genus, which, like Anona, is common to America and Africa, but of which no species has yet been observed in Asia.

Of MALPIGHIACEÆ, an order chiefly belonging to equinoctial America, there are also three species from Congo.

One of these is Banisteria Leona, first described, from [426 Smeathman's specimens, by Cavanilles^[10] who has added the fruit of a very different plant to his figure, and quotes the herbarium of M. de Jussieu as authority for this species being likewise a native of America, which is, I believe, equally a mistake.

The two remaining plants of Malpighiaceæ, in the collection, with some additional species from different parts of the coast, form a new genus, having the fruit of Banisteria, but with sufficient distinguishing characters in the parts of the flower, and remarkable in having alternate leaves. From this disposition of leaves, in which the genus here noticed differs from all others decidedly belonging to the order, an additional argument is afforded, for referring Vitmannia to Malpighiaceæ, as proposed by M. du Petit Thouars;^[11] and the approximation, though perhaps not the absolute union of Erythroxylon to the same family is confirmed.

It may not be improper here to notice a very remarkable deviation from the usual structure of leaves in Malpighiaceæ, which is supposed to occur in a plant of equinoctial Africa, namely Flabellaria pinnata of Cavanilles (the Hirœa pinnata of Willdenow). It is certain, however, that the figure given by Cavanilles of this species is made up from two very different genera; the pinnated leaf belonging to an unpublished Pterocarpus; the fructification to a species of Hireea, having simple opposite leaves. The evidence respecting this blunder, which was detected by Mr. Dryander, is to be found in the herbarium of Sir Joseph Banks.

In Malpighiaceæ the insertion of the ovulum is towards its apex, or considerably above its middle; and the radicle of the embryo is uniformly superior. In these points Banisteria presents no exception to the general structure, though Gærtner has described its radicle as inferior, and M. de Jussieu does not appear to have satisfied himself respecting the fact.^[12] It appears, however, that M. Richard is aware of the constancy in the direction of the embryo in this order.^[13]

HIPPOCRATICEÆ. M. de Jussieu has lately proposed this as a distinct family,^[14] of which there are two plants in the collection. The first is a species of Hippocratea; the second is referable to Salacia.

In Hippocraticeæ, the insertion of the ovula is either [427 towards the base, or is central; the direction of the radicle is always inferior. In these points of structure, which are left undetermined by M. de Jussieu, they differ from Malpighiaceæ, but agree with Celastrinæ, to which, notwithstanding the difference in insertion and number of stamina, and in the want of albumen, they appear to me to have a considerable degree of affinity; especially to Elæodendrum, where the albumen is hardly visible, and to Ptelidium, as suggested by M. du Petit Thouars,^[15] in which it is reduced to a thin membrane.

SAPINDACEÆ. Only four plants of this natural family, which is almost entirely equinoctial, occur in the herbarium. Two of these are new species of Sapindus. The third is probably not specifically different from Cardiospermum grandiflorum of the West India Islands. And the fourth is so nearly related to Paullina pinnata, of the opposite coast of America, as to be with difficulty distinguished from it. M. de Jussieu,^[16] who probably intends the same plant, when he states P. pinnata to be a native of equinoctial Africa, has also described a second species from Senegal.^[17] No other species of this genus has hitherto been found, except in equinoctial America; for Paullinia Japonica of Thunberg, probably belongs even to a different natural order. The species from Congo, however, seems to be a very general plant on this line of coast; having been found by Brass near Cape Coast, and by Park on the banks of the Gambia.

In Sapindaceæ there is not the same constancy in the insertion of the ovulum and consequent direction of embryo, as in the two preceding orders. For although, in the far greater part of this family, the ovulum is erect and the radicle of the embryo inferior, yet it includes more than one genus in which both the seeds and the embryo are inverted. With this fact it would seem M. de Jussieu is unacquainted;^[18] and he is surely not aware that in his late Memoir on Melicocca^[19] he has referred plants to that genus differing from each other in this important point of structure.

TILIACEÆ. It is remarkable that of only nine 428] species belonging to this family in Professor Smith's herbarium, three should form genera hitherto unnoticed.

The first of these new genera is a shrub, in several of its characters related to Sparmannia, like which, it has the greater part of its outer stamina destitute of antheræ; in the structure of its fruit, however, it approaches more nearly to Corchorus.

The second genus also agrees with Corchorus in its fruit; but differs from it sufficiently in the form and dehiscence of the antheræ; as well as in the short pedicellus, like that of Grewia, elevating its stamina and pistillum.

The third, of which the specimens are in fruit only, fortunately, however, accompanied by the persistent flower, is remarkable in having a calyx of three lobes, while its corolla consists of five petals; the stamina are in indefinite number; and the fruit is composed of five single-seeded capsules, connected only at the base. In the want of symmetry or proportion between the divisions of its calyx and corolla it resembles the Chlenaceæ of M. Du Petit Thouars,^[20] as well as Oncoba of Forskael and Ventenatia of M. de Beauvois.^[21] The existence of this new genus decidedly belonging to Tiliaceæ, and having a considerable resemblance to Ventenatia, whose place in the system is, indeed, not yet determined, but of which the habit is nearly that of Rhodolæna, seems in some degree to confirm M. du Petit Thouars's opinion of the near relation of Chlenaceæ to Tiliaceæ; though M. de Jussieu, in placing it between Ebenaceæ and Rhodoraceæ,^[22] appears to take a very different view of its affinities.

MALVACEÆ. Of this family 18 species were observed on the banks of the Congo. It forms, therefore, about one thirty-fourth part of the phænogamous plants of the collection; which is somewhat greater than the equinoctial proportion of the order, as stated in Baron Humboldt's dissertation,^[23] but nearly agrees with that of India, according to Dr. Roxburgh's unpublished Flora Indica.

The greater part of the Malvaceæ of the collection belong to Sida and Hibiscus; and certain species of both these genera are common to India and America. Urena Americana and Malachra radiata, hitherto supposed to be natives of America only, are also contained in the collection; and [429 the loftiest tree seen on the banks of the Congo, is a species of Bombax, which, as far as can be determined from the very imperfect specimens preserved in the herbarium, does not differ from Bombax pentandrum of America and India. I have formerly remarked^[24] that Malvaceæ, Tiliaceæ, Hermanniaceæ, Buttneriaceæ, and Sterculiaceæ, constitute one natural class; of which the orders appear to me as nearly related as the different sections of Rosaceæ are to each other. In both these, as well as in several other cases that might be mentioned, there seems to be a necessity for the establishment of natural classes, to which proper names, derived from the orders best known, and differing perhaps in termination, might be given.

It is remarkable that the most general character connecting the different orders of the class now proposed, and which may be named from its principal order Malvaceæ, should be that of the valvular æstivation of the calyx; for several, at least, of the genera at present referred to Tiliaceæ, in which this character is not found, ought probably, for other reasons likewise, to be excluded from that order: and hence perhaps also the Chlenaceæ, though nearly related, are not strictly referable to the class Malvaceae, from all of whose orders, it must be admitted, they differ considerably in habit.

LEGUMINOSÆ. According to Baron Humboldt,^[25] this family, or class, as I am rather disposed to consider it, constitutes one twelfth of the phænogamous plants within the tropics. Its proportion, however, is much greater in Professor Smith's herbarium, in which there are 96 species belonging to it, or nearly one sixth of the whole collection. And ample allowance being made for the lateness of the season when the collection was formed, which might be supposed to reduce the number of this family less than many of the others, Leguminosæ may be stated as forming one eighth of the Phænogamous plants on the banks of the Congo. In India, it probably forms about one ninth, which is also nearly the proportion it bears to Phænogamous plants in the equinoctial part of New Holland.

I have formerly proposed to subdivide Leguminosæ into three orders.^[26]

Of the first of these orders, MIMOSEÆ, there are only 430] eight species from Congo, seven of which belong to Acacia, as it is at present constituted; the eighth is a sensitive aculeated Mimosa very nearly allied to M. aspera of the West Indies, as well as to M. canescens of Willdenow, found by Isert in Guinea; and perhaps is not different from the species mentioned by Adanson as being common on the banks of the Senegal.

Of the second order, CÆSALPINEÆ, the collection contains 19 species, among which there are four unpublished genera. One of these is Erythrophleum of Afzelius, the Red Water Tree of Sierra Leone; another species of which genus is the ordeal plant, or Cassa of the natives of Congo. Guilandina Bonduc and Cassia occidentalis are also in the herbarium; the former, I beheve, is unquestionably common to India and America; whether Cassia occidentalis be really a native of India and equinoctial Africa, in both of which it is now at least naturalized, is perhaps doubtful.

Among PAPILIONACEÆ, which constitute the principal part of Leguminosæ in the collection, there is only one plant with stamina entirely distinct. This decandrous plant forms a genus very different from any yet established, but to which Podalyria bracteata of Roxburgh^[27] belongs.

The genera composing Papilionaceæ on the banks of the Congo have, upon the whole, a much nearer relation to those of India than of equinoctial America. To this, however, there is one remarkable exception. For of the only two species of Pterocarpus in the collection, one is hardly to be distinguished from P. Ecastaphyllum, unless by the want of the short acumen existing in the plant of Jamaica. The second agrees entirely with Linnaeus's original specimen of P. lunatus from Surinam, and seems to be not uncommon on the west coast of equinoctial Africa; having been observed by Professor Afzelius at Sierra Leone, and probably by Isert in Guinea;^[28] while no species of Pterocarpus related to either of these has hitherto been observed in India. On the other hand Abrus precatorius and Hedysarum triflorum, both of which occur in the collection, are common to equinoctial Asia and America.

TEREBINTACEÆ, as given by M. de Jussieu, appears to be made up of several orders nearly related to each other, and of certain genera having but little affinity to any of them. Of this, indeed, the illustrous author of the Genera Plantarum seems to have been aware, He pro- [431 probably, however, had not the means of ascertaining all their distinguishing characters, and therefore preferred leaving the order nearly as it was originally proposed by Bernard de Jussieu in 1759.

One of the orders included in Terebintaceæ, and which is proposed by M. de Jussieu himself, under the name of Cassuviæ, consists of Anacardium, Semecarpus, Mangifera, Rhus, and Buchanania, with some other unpublished genera.

The perigynous insertion of stamina in Cassuviæ (or Anacardeæ) may be admitted in doubtful cases from analogy, there being an unpublished genus belonging to it even with ovarium inferum. And the ovarium, though in all cases of one cell, with a single ovulum, may, at least in those genera in which the style is divided, be supposed to unite in its substance the imperfect ovaria indicated by the branches of the style, and which in Buchanania are actually distinct from the complete organ. The only plant belonging to this order in the herbarium, is a species of Rhus, with simple verticillate leaves, and very nearly approaching in habit to two unpublished species of the genus from the Cape of Good Hope.

AMYRIDEÆ, another family included in Terebintaceæ, and to which the greater part of Jussieu's second section belongs, may, like the former order, be considered as having in all cases perigynous insertion of stamina; this structure being manifest in some of its genera. Of Amyrideæ, there are two plants in the collection. The first of these is a male plant, probably of a species of Sorindeia;^[29] the second, which is the Safu of the natives, by whom it is cultivated on account of its fruit, cannot be determined from the imperfect state of the specimens; it is, however, probably related to Poupartia or Bursera.

CONNARACEÆ, is a third family which I propose to separate from Terebintaceæ; it consists of Connarus Linn. Cuestis Juss. and Rourea of Aublet or Robergia of Schreber. The insertion of stamina, in this family, is ambiguous; but as in a species of Cnestis from Congo, they originate from, or at least firmly cohere with, the pedicellus of the ovaria, they may be considered perhaps in all the [432 genera rather as hypogynous than perigynous. The most important distinguishing characters of Connaraceæ consist in the insertion of the the two collateral ovula of each of its pistilla being near the base; while the radicle of the embryo is situated at the upper or opposite extremity of the seed, which is always solitary. In Connarus there is but one ovarium, and the seed (figured by Gærtner under the name of Omphalobium) is destitute of albumen. Rourea or Robergia has always five ovaria, though in general one only comes to maturity. Its seed, like that of Connarus, is without albumen, and the æstivation of the calyx is imbricate.

Of Cnestis there are several new species in Professor Smith's herbarium. This genus has also five ovaria, all of which frequently ripen; the albumen forms a considerable part of the mass of the seed; and the æstivation of the calyx is valvular. The genera of this group, therefore, differ from each other, in having one or more ovaria; in the existence or absence of albumen; and in the imbricate or valvular æstivation of calyx. Any one of these characters singly is frequently of more than generic importance, though here even when all are taken together, they appear insufficient to separate Cnestis from Connarus.

In considering the place of the Connaraceæ in the system, they appear evidently connected on the one hand with Leguminosæ, from which Connarus can only be distinguished by the relation the parts of its embryo have to the umbilicus of the seed. On the other hand, Cnestis seems to me to approach to Averrhoa, which agrees with it in habit, and in many respects in the structure of its flower and seed; differing from it, however, in its five ovaria being united, in the greater number of ovula in each cell, in the very different texture of its fruit, and in some degree in the situation of the umbilicus of the seed.

But Averrhoa agrees with Oxalis in every important point of structure of its flower, and in most respects in that of its seed.

Oxalis, indeed, differs from Averrhoa in the texture of its fruits, in some respects in the structure of its seed; and very widely in habit, in the greater part of its species. The difference in habit, however, is not so great in some species of Oxalis; as for example, in those with pinnated and even ternate leaves from equinoctial America; and in that natural division of the genus including O. sensitiva, of which there are two species in the Congo herbarium. 433] This latter section of Oxalis^[30] agrees also with Averrhoa Carambola^[31] in the foliola, when irritated, being reflected or dependent, which is likewise their position in the state of collapsion or sleep, in all the species of both genera.

To the natural order formed by Oxalis and Averrhoa, the name of Oxalideæ may be given, in preference to that of Sensitivæ, under which, however, Batsch^[32] was the first to propose the association of these two genera, and to point out their agreement in sensible qualities and irritability of leaves.

M. de Jussieu, in a memoir recently published,^[33] has proposed to remove Oxalis from Geraniaceæ, to which he had formerly annexed it, and to unite it with Diosmeæ.

It appears to me to have a much nearer affinity to Zygophylleæ^[34], though it is surely less intimately connected with that order than with Averrhoa.

I am aware that M. Correa de Serra, one of the most profound and philosophical botanists of the present age, has considered Averrhoa as nearly related to Rhamneæ^[35] or rather to Celastrinæ; from which, however, it differs in the number and insertion of stamina and especially in the direction of the embryo, with respect to the pericarpium.

In all these characters Averrhoa agrees with Oxalis; its relation to which is further confirmed on considering the appendage of the seed or arillus, whose modifications in these two genera seem to correspond with those of their pericarpia.

CHRYSOBALANEÆ. The genera forming this order are Chrysobalanus, Moquilea, Grangeria, Coupea, Acioa, Licania, Hirtella, Thelira, and Parinarium, all of which are at present referred by M. de Jussieu to Rosaceæ, and the greater part to his seventh section of that family, namely, Amygdaleæ. If Rosaceæ be considered as an order merely, these genera will form a separate section, connecting it with Leguminosæ. But if, as I have formerly proposed, both these extensive families are to be regarded as natural classes, then they will form an order sufficiently distinct from Amygdaleæ, both in fructitication and habit, as well as in geographical distribution.

The principal distinguishing characters in the fructification of Chrysobalaneæ are the style proceeding from the base of the ovarium; and the ovula (which, as in Amyg- [434 daleæ, are two in number) as well as the embryo being erect. The greater part of Chrysobalaneæ have their flowers more or less irregular; the irregularity consisting in the cohesion of the foot-stalk of the ovarium with one side of the tube of the calyx, and a greater number, or greater perfection of stamina on the same side of the flower.

Professor Smith's herbarium contains only two genera of this order, namely, Chrysobalanus and Parinarium.^[36] One species of the former is hardly distinguishable from Chrysobalanus Icaco of America, and is probably a very common plant on the west coast of Africa; Icaco being mentioned by Isert^[37] as a native of Guinea, and by Adanson^[38] in his account of Senegal

Of Parinarium, there is only one species from Congo, which agrees, in the number and disposition of stamina, with the character given of the genus. In these respects M. de Jussieu^[39] has observed a difference in the two species found by Adanson at Senegal, and has moreover remarked that their ovarium coheres with the tube of the calyx. In that species most common at Sierra Leone, and which is probably one of those examined by M. de Jussien, the ovarium itself is certainly free, its pedicellus, however, as in the greater part of the genera of this order and several of Cæsalpineæ, firmly cohering with the calyx, may account for the statement referred to. I am not, indeed, acquainted with any instance among Dicotyledonous plants of cohesion between a simple ovarium, which I consider that of Chrysobalaneæ to be, and the tube of the calyx.

The complete septum between the two ovula of Parinarium, existing before fecundation, is a peculiar structure in a simple ovarium; though in some degree analogous to the moveable dessepiment of Banksia and Dryandra, and to the complete, but less regular, division of the cavity that takes place after fecundation in some species of Persoonia.^[40]

MELASTOMACEÆ. Four plants only of this order occur in the collection.

The first is a species of Tristemma, very nearly related to T. hirtum of M. de Beauvois.^[41]

435] The second is perhaps not distinct from Melastoma decumbens, of the same author.^[42]

The third and fourth are new species referable to Rhexia, as characterised by Ventenat,^[43] though not to that genus as established by Linnæus; and in some respects differing from the species that have been since added to it, all of which are natives of America.

In the original species of Tristemma^[44] there are, in the upper part of the tube of the calyx, two circular ciliated membranous processes, from which the name of the genus is derived; the limb of the calyx itself being considered as constituting the third circle. The two circular membranes are also represented as complete in T. hirtum.

But in the species from Congo, which may be named T. incompletum, only one circular membrane exists, with the unilateral rudiment of the second.

The rudiment of the inferior membrane in this species points out the relation between the apparently anomalous appendage of the calyx in Tristemma, and the ciliated scales irregularly scattered over its whole surface in Osbeckia; the analogy being established by the intermediate structure of an unpublished plant of this order from Sierra Leone, in Sir Joseph Banks's herbarium, in which the nearly similar squamæ, though distinct, are disposed in a single complete circle; and by Melastoma octandra of Linnæus, in which they are only four in number, and alternate with the proper divisions of the calyx.

The two species here referred, though improperly, to Rhexia, agree with a considerable part of the species published in the monograph of that genus by M. Bonpland, and with some other genera of the order, in the peculiar manner in which the ovarium is connected with the tube of the calyx. This cohesion, instead of extending uniformly over the whole surface, is limited to ten longitudinal equidistant lines or membranous processes, apparently originating from the surface of the ovarium; the interstices, which are tubular, and gradually narrowing towards the base, being entirely free.

The function of these tubular interstices is as remarkable as their existence.

In Melastomaceae, before the expansion of the corolla, the tops of the filaments are inflected, and the antheræ are pendulous and parallel to the lower or erect portion of the filament; their tips reaching, either to the line of complete cohesion between the calyx and ovarium, where that exists; or, where this cohesion is partial, and such as I have now [436 described, being lodged in the tubular interstices; their points extending to the base of the ovarium. From these sheaths, to which they are exactly adapted, the antheræ seem to be disengaged in consequence of the unequal growth of the different parts of the filament; the inflected portion ceasing to increase in length at an early period, while that below the curvature continues to elongate considerably until the extrication is complete, when expansion takes place.

It is singular that this mode of cohesion between the ovarium and calyx in certain genera of Melastomaceæ, and the equally remarkable æstivation of antheræ accompanying it, should have been universally overlooked, especially in the late monograph of M. Bonpland; as both the structure and economy certainly exist in some, and probably in the greater part, of the plants which that author has figured and described as belonging to Rhexia.

On the limits, structure, and generic division of Melastomaceæ, I may remark—

1st. That Memecylon, as M. du Petit Thouars has already suggested,^[45] and Petaloma of Swartz^[46] both belong to this order, and connect it with Myrtaceæ, from which they are to be distinguished only by the absence of the pellucid glands of the leaves and other parts, existing in all the genera really belonging to that extensive family.

2ndly. There are very few Melastomaceæ in which the ovarium does not in some degree cohere with the tube of the calyx; Meriana, properly so called, being, perhaps, the only exception.

And in the greater number of instances where, though the ovarium is coherent, the fruit is distinct, it becomes so from the laceration of the connecting processes already described.

3rdly. That the generic divisions of the whole order remain to be established. On examination, I believe, it will be found that the original species of the Linnean genera, Melastoma and Rhexia, possess generic characters sufficiently distinguishing them from the greater part of the plants that have been since added to them by various authors. In consequence of these additions, however, their botanical history has been so far neglected, that probably no genuine species of Melastoma, and certainly none of Rhexia, has yet been published in M. Bonpland's splendid and valuable monographs of these two genera. Of RHIZOPHOREÆ,^[47] as I have formerly proposed to [437 limit it, namely, to Rhizophora, Brugniera, and Carallia, the collection contains only one plant, which is a species of Rhizophora, the Mangrove of the lower part of the river, and probably of the whole line of coast, but very different both from that of America, and from those either of India or of other equinoctial countries that have been described. There is, however, a plant in the collection which, though not strictly belonging to this order, suggests a few remarks on its affinities.

I referred Carallia^[48] to Rhizophoreæ, from its agreement with them in habit, and in the structure of its flower. It is still uncertain whether its reniform seed is destitute of albumen; the absence of which, however, does not seem necessary to establish its affinity with the other genera of this order; for plants having the same remarkable economy in the germination of the embryo as that of Rhizophora, may belong to families which either have or are destitute of albumen.

The plant referred to from Congo may be considered as a new species of Legnotis having its petals less divided than those of the original species of that genus, and each cell of its ovarium containing only two pendulous ovula. The genus Legnotis agrees with Carallia in habit, especially in having opposite leaves with intermediate stipules; in the valvular æstivation of its calyx, and in several other points of structure of its flower. It differs in its divided petals; in its greater number of stamina, disposed, however, in a simple series; and in its ovarium not cohering with the calyx. It is therefore still more nearly related to Richœia of M. du Petit Thouars,^[49] from which perhaps it may not be generically distinct. The propriety of associating Carallia^[50] with Rhizophoreæ is not perhaps likely to be disputed; and its affinity to Legnotis, especially to the species from Congo, appears very probable. It would seem, therefore, that we have already a series of structures connecting Rhizophora on the one hand with certain genera of Salicariæ, particularly with Antherylium, thongh that genus wants the intermediate stipules; and on the other with Cunoniaceæ,^[51] especially with the simple leaved species of 438] Ceratopelatum. While Loranthus and Viscum, associated with Rhizophora by M. de Jussieu, appear to form a very distinct family, and which, as it seems to me, should even occupy a distant place in the system.

HOMALINÆ. In the collection from Congo a plant occurs evidently allied, and perhaps referable, to Homalium, from which it differs only in the greater number of glands alternating with the stamina, whose fasciculi are in consequence decomposed: the inner stamen of each fasciculus being separated from the two outer by one of the additional glands. This plant was first found on the banks of the Gambia, by Mr. Park, from whose specimens I have ascertained that the embryo is enclosed in a fleshy albumen.

The same structure of seed may be supposed, from very obvious affinity, to exist in Astranthus of Loureiro, to which Blackwellia of Commerson ought perhaps to be referred; in Napinoga of Aublet, probably not different from Homalium; and in Nisa,^[52] a genus admitting of subdivision, and which M. du Petit Thouars has referred to Rhamneæ. All these genera appear to me sufficiently different from Rosaceæ, where M. de Jussieu has placed them, and from every other family of plants at present established.

Their distinguishing characters as a separate order are, the segments of the perianthium disposed in a double series, or an equal number of segments nearly in the same series; the want of petals; the stamina being definite and opposite to the inner series of the perianthium, or to the alternate segments where they are disposed apparently in a simple series; the unilocular ovarium (generally in some degree coherent with the calyx) having three parietal placentæ, with one, two, or even an indefinite number of ovula; and the seeds having albumen, as inferred from its existence in the genus from Congo. The cohesion of the ovarium with the tube of the perianthium, though existing in various degrees in all the genera above enumerated, is probably a character of only secondary importance in Honialinæ. For an unpublished genus found by Connnerson in Madagascar, which in every other respect agrees uith this family, has ovarium superum. This genus at the same time seems to establish a considerable affinity between Homalinæ and certain genera, either absolutely belonging to Passifloreæ, especially Paropsia of M. du Petit Thouars,^[53] or nearly related to them as Erythrospermum, well de- [439 scribed and figured by the same excellent botanist.^[54]

The increased uumber of stamina in Homalium, and particularly in the genus from Congo, instead of presenting an objection to this affinity, appears to me to confirm it. It may be observed also that there are two genera referable to Passifloreæ, though they will form a separate section of the order, which have a much greater, and even an indefinite, number of perfect stamina, namely, Smeathmania, an unpublished genus of equinoctial Africa, agreeing in habit, in perianthium, and in fruit, with Paropsia; and Ryania of Vahl,^[55] which appears to me to belong to the same family.

In Passifloreæ the stamina, when their number is definite, which is the case in all the genera hitherto considered as belonging to them, are opposite to the outer series of the perianthium; a character which, though of general importance, and here of practical utility in distinguishing them from Homalinæ, is not expressed in any of the numerous figures or descriptions that have been published of the plants of this order.

Passifloreæ and Cucurbitaceæ, though now admitted as distinct families, are still placed together by M. de Jussieu; and he considers the floral envelope in both orders as a perianthium or calyx, whose segments are disposed in a double series.^[56]

These views of affinity and structure are in some degree confirmed by Homalinæ, in which both ovarium inferum and superum occur; and in one genus of which, namely, Blackwellia, the segments of the perianthium, though the complete number, in relation to the other genera of the order, be present, are all of similar texture and form, and are disposed nearly in a simple series. If the approximation of these three families be admitted, they may be considered as forming a class intermediate between Polypetalæ and Apetalæ, whose principal characters would consist in the segments of the calyx being disposed in a double series, and in the absence of petals; the different orders nearly agreeing with each other in the structure of their seeds, and to a considerable degree in that of the ovarium.

The formation of this class, however, connected on the 440] one hand with Apetalæ by Samydeæ,^[57] and on the other, though as it seems to me less intimately, with Polypetalæ by Violeæ, would not accord with any arrangement of natural orders that has yet been given. While the admission of the floral envelope being entirely calyx; and of the affinity of the class with Violeæ, would certainly be unfavorable to M. de Candolle's ingenious hypothesis of petals in all cases being modified stamina.

VIOLEÆ.^[58] This order does not appear to me so nearly related to Passifloreæ as M. du Petit Thouars is disposed to consider it; for it not only has a genuine polypetalous corolla, which is hypogynous, but its antheræ differ materially in structure, and its simple calyx is divided to the base. The irregularity both of petals and stamina in the original genera of the order, namely, Viola, Pombalia,^[59] and Hybanthus, though characters of considerable importance, are not in all cases connected with such a difference in habit as to prevent their union with certain regular flowered genera, which it has lately been proposed to associate with them.

The collection from Congo contains two plants belonging to the section of Violeæ with regular flowers. One of these evidently belongs to Passalia, an unpublished genus in Sir Joseph Banks's herbarium, and described in the manuscripts of Solander from a plant found by Smeathman at Sierra Leone, which is perhaps not specifically distinct from that of Congo, or from Ceranthera dentata of the Flore d'Oware. But Ceranthera,^[60] which M. de Beauvois, being unacquainted with its fruit, has placed in the order Meliaceæ, is not different from Alsodeia, a genus published somewhat earlier, and from more perfect materials, by M. du Petit Thouars,^[61] who refers it to Violeæ. The latter generic name ought of course to be adopted, and with a change in the termination (Alsodinæ) it may also denote the section of this order with regular flowers.

Physiphora of Sir Joseph Banks's herbarium, discovered by himself in Brazil, differs from Alsodeia only in its filaments being very slightly connected at base, and in the form and texture of its capsule, which is membranaceous, and, as the name imports, inflated.

Five species belonging to this section of Violeæ occur in Aublet's History of the Plants of Guiana, where each of [441 them is considered as forming a separate genus. Of three of these genera, namely, Conohoria, Rinorca, and Riana the flowers alone are described; the two others, Passura and Piparea, were seen in fruit only.

From the examination of flowers of Aublet's original specimens of the three former genera, in Sir Joseph Banks's herbarium, and of the fruit of Conohoria, which entirely agrees with that of Passura, and essentially with that of Piparea, I have hardly a doubt of these five plants, notwithstanding some differences in the disposition of their leaves, actually belonging to one and the same genus; and as they agree with Physiphora in every respect, except in the texture and form of the capsule, and with the Passalia of Sierra Leone and Congo, except in having their stamina nearly or entirely distinct, it appears that all those genera may be referred to Alsodeia.

I have also examined, in Sir Joseph Banks's herbarium, a specimen of Pentaloba sessilis of the Flora Cochinchinensis, which was sent so named, by Loureiro himself, and have found it to agree in every important point with Alsodeia, even as to the number of parietal placentæ. Loureiro, however, describes the fruit of Pentaloba as a five-lobed, five- seeded berry, and if this account be correct, the genus ought to be considered as distinct; but if, which is not very improbable, the fruit be really capsular, it is evidently referable to Alsodeia; with the species of which, from Madagascar and the west coast of equinoctial Africa, it agrees in the manifest union of its filaments.

It appears therefore that the ten genera now enumerated, and perhaps also Lauradia of Vandelli, may very properly be reduced to one; and they all at least manifestly belong to the same section of Violeæ, though at present they are to be found in various, and some rather distant, natural orders.

M. de Jussieu, in adopting Aublet's erroneous description of the stamina of Rinorea and Conohoria, has referred both these genera to Berberides,^[62] to which he has also annexed Riana, adding a query whether Passura may not 442] belong to the same genus. With M. de Beauvois, he refers Ceranthera to Meliaceæ; and Pentaloba of Loureiro he reduces also to the same order.^[63] Piparea is, together with Viola, annexed to Cistinæ in his Genera Plantarum, and is therefore the most correctly placed, though its structure is the least known, of all these supposed genera.

An unpublished genus of New Holland, which I have named Hymenanthera, in Sir Joseph Banks's herbarium, agrees with Alsodeia in its calyx, in the insertion, expans:on, and obliquely imbricate æstivation of its petals, and especially in the structure of its antheræ, which approach more nearly to those of Violeæ properly so called. It differs, however, from this order in having five squamæ alternating with the petals; and especially in its fruit, which is a bilocular berry, having in each cell a single pendulous seed, whose internal structure resembles that both of Violeæ and Polygaleæ, between which I am inclined to think this genus should be placed.

CHAILLETEÆ. The genus Chailletia was established by M. de Candolle^[64] from a plant found by Martin in French Guiana, and which, as appears by specimens in Sir Joseph Banks's herbarium, had been many years before named Patrisia by Von Rohr, who discovered it in the same country. At a still earlier period, Solander, in his manuscripts, preserved in the library of Sir Joseph Banks, described this genus under the name of Mestotes, from several species found by Smeathman at Sierra Leone. Both Dichapetalum and Leucosia of M. du Petit Thouars^[65] appear to me, from the examination of authentic specimens, to belong to the same genus; and in Professor Smith's herbarium there is at least one additional species of Chailletia different from those of Sierra Leone.

Of the two generic names given by M. du Petit [443 Thouars, and published somewhat earlier than M de Candolle's Memoir, Leucosia will probably be considered inadmissible, having been previously applied by Fabricius to a genus of Crustacea; and Dichapetalum is perhaps objectionable, as derived from a character not existing in the whole genus, even allowing it to be really polypetalous. It seems expedient, therefore, to adopt the name proposed by M. de Candolle, who has well illustrated the genus in the memoir referred to. It appears to me that Chailletia, a genus nearly related to it from India with capsular fruit, and Tapura of Aublet (which is Rohria of Schreber), form a natural order, very different from any yet established. The principal characters of this order may be gathered from M. de Candolle's figure and description of Chailletia, to which, however, must be added that the cells of the ovarium, either two or three in number, constantly contain two collateral pendulous ovula; and that in the regular flowered genera there exist within, and opposite to, the petal-like bodies an equal number of glands, which are described by M. du Petit Thouars in Dichapetalum, but are unnoticed by him in Leucosia, where, however, they are equally present.

It may seem paradoxical to associate with these genera Tapura, whose flower is irregular, triandrous, and apparently monopetalous. But it will somewhat lessen their apparent differences of structure to consider the petal-like bodies, which, in all the genera of this order, are inserted nearly or absolutely in the same series with the filaments, as being barren stamina; a view which M. de Candolle has taken of those of Chailletia, and which M. Richard had long before published respecting Tapura.^[66] It is probable also that M. de Candolle at least will admit the association here proposed, as his Chailletia sessiliflora seems to be merely an imperfect specimen of Tapura guianensis.

The genera to which Chailleteæ most nearly approach appear to me to be Aquilaria of Lamarck^[67] and Gyrinops of Gærtner. But these two genera themselves, which are not referable to any order yet established, may either be regarded as a distinct family, or perhaps, to avoid the too great multiplication of families, as a section of that at present 441] under consideration, and to which I should then propose to apply the name of Aquilarinæ in preference to Chailleteæ.

The genus Aquilaria itself has been referred by Ventenat to Samydeæ. From this order, however, it is sufficiently distinct, not only in the structure of its ovarium and seeds, but in its leaves being altogether destitute of glands, which are not only numerous in Samydeæ, but consisting of a mixture of round and linear pellucid dots, distinguish them from all the other families^[68] with which there is any probability of their being confounded.

Sir James Smith^[69] has lately suggested the near affinity of Aquilaria to Euphorbiaceæ. But I confess it appears to me at least as distinct from that order as from Samydeæ; and I am inclined to think, paradoxical as it may seem, that it would be less difficult to prove its affinity to Thymeleæ than to either of them; a point, however, which, requiring considerable details, I do not mean to attempt in the present essay.

Of EUPHORBIACEÆ there are twenty species in the collection, or one twenty-eighth part of its Phænogamous plants. This is somewhat greater than the intratropical proportion of the order as stated by Baron Humboldt, but rather smaller than that of India or of the northern parts of New Holland.

The most remarkable plants of Euphorbiaceæ in the Congo herbarium are: a new species of the American genus Alchornea; a plant differing from Ægopricon, a genus also belonging to America, chiefly in its capsular fruit; two new species of Bridelia, which has hitherto been observed only in India; and an unpublished genus that I have formerly alluded to,^[70] as in some degree explaining the real structure of Euphorbia, and from the consideration of which also it seems probable that what was formerly described as the hermaphrodite flower of that genus, is in reality a compound fasciculus of flowers.^[71] From the same species of this unpublished genus a substance resembling caoutchouc is said to be obtained at Sierra Leone.

According to Mr. Lockhart a frutescent species of 445] Euphorbia, about eight feet in height, with cylindrical stem and branches, was observed, planted on the graves of the natives near several of the villages; but of this, which may be what Captain Tuckey has called Cactus quadrangularis in his Narrative (p. 115), there is no specimen in the herbarium.

COMPOSITÆ. It is unnecessary here to enter into the question whether this family of plants, of which upwards of 3000 species are already known, ought to be considered as a class or as an order merely; the expediency of subdividing it, and affixing proper names to the divisions, being generally admitted. The divisions or tribes proposed by M. Cassini, in his valuable dissertations on this family, appear to be the most natural, though as yet they have not been very satisfactorily defined.

The number of Compositæ in the collection is only twenty-four, more than half of which are referable to Heliantheæ and Vernoniaceæ of M. Cassini. The greater part of these are unpublished species, and among them are five new genera. The published species belong to other divisions, and are chiefly Indian: but one of them, Ageratum conyzoides, is common to America and India; the Struchium (or Sparganophorus) of the collection does not appear to me different from that of the West Indies; and Mikania chenopodifolia, a plant very general on this line of coast, though perhaps confined to it, belongs to a genus of which all the other species are found only in America. Baron Humboldt has stated^[72] that Compositæ form one sixth of the Phænogamous plants within the tropics, and that their proportion gradually decreases in the higher latitudes until in the frigid zones it is reduced to one thirteenth. But in the herbarium from Congo Compositæ form only one twenty-third, and both in Smeathman's collection from Sierra Leone and in Dr. Roxburgh's Flora Indica, a still smaller part, of the Phænogamous plants. In the northern part of New Holland they form about one sixteenth; and in a manuscript catalogue of plants of equinoctial America, in the library of Sir Joseph Banks, they are nearly in the same proportion.

In estimating the comparative value of these different materials, I may, in the first place, observe that though the herbarium from Congo was collected in the dry season of the country, there is no reason to suppose on that account that the proportion of this family of plants, in particular, is materially or even in any degree diminished, nor can [446 this objection be stated to the Sierra Leone collection, in which its relative number is still smaller.

To the Compositæ in Dr. Roxburgh's Flora Indica, however, a considerable addition ought, no doubt, to be made; partly on the ground of his having apparently paid less attention to them himself, and still more because his correspondents, whose contributions form a considerable part of the Flora, have evidently in a great measure neglected them. This addition being made, the proportion of Compositæ in India would not differ very materially from that of the north coast of New Holland, according to my own collection, which I consider as having been formed in more favorable circumstances, and as probably giving an approximation of the true proportions in the country examined. Baron Humboldt's herbarium, though absolutely greater than any of the others referred to on this subject, is yet, with relation to the vast regions whose vegetation it represents, less extensive than either that of the north coast of New Holland, or even of the line of the Congo. And as it is in fact as much the Flora of the Andes as of the coasts of intratropical America, containing families nearly or wholly unknown on the shores of equinoctial countries, it may be supposed to have several of those families which are common to all such countries, and among them Compositæ, in very different proportion. At the same time it is not improbable that the relative number of this family in equinoctial America, may be greater than in the similar regions of other intratropical countries; while there seems some reason to suppose it considerably smaller on the west coast of Africa. This diminished proportion, however, in equinoctial Africa would be the more remarkable, as there is probably no part of the world in which Compositæ form so great a portion of the vegetation as at the Cape of Good Hope.

RUBIACEÆ. Of this family there are forty-three species in the collection, or about one fourteenth of its Phaenogamous plants. I have no reason to suppose that this proportion is greater than that existing in other parts of equinoctial Africa; on the contrary, it is exactly that of Smeathman's collection from Sierra Leone.

Baron Humboldt, however, states the equinoctial proportion of Rubiaceæ to phænogamous plants to be one to twenty-nine, and that the order gradually diminishes in relative number towards the poles.

417] But it is to be observed that this family is composed of two divisions, having very different relations to climate; the first, with opposite, or more rarely verticillate, leaves and intermediate stipules, to which, though constituting the great mass of the order, the name Rubiaceæ cannot be applied, being chiefly equinoctial; while the second, or Stellatæ, having verticillate or very rarely opposite leaves, but in no case intermediate stipules, has its maximum in the temperate zones, and is hardly found within the tropics, unless at great heights.

Plence perhaps we are to look for the minimum in number of species of the whole order, not in the frigid zone, but, at least in certain situations, a few degrees only beyond the tropics.

In conformity to this statement, M. Delile's valuable catalogue of the plants of Egypt^[73] includes no indigenous species of the equinoctial division of the order, and only five of Stellatæ, or hardly the one hundred and sixtieth part of the Phænogamous plants. In M. Desfontaines' Flora Atlantica, Rubiaceæ, consisting of fifteen Stellatae and only one species of the equinoctial division, form less than one ninetieth part of the Phænogamous plants, a proportion somewhat inferior to that existing in Lapland.

In Professor Thunberg's Flora of the Cape of Good Hope, where Rubiaceæ are to Phænogamous plants as about one to one hundred and fifty, the order is differently constituted; the equinoctial division, by the addition of Anthospermum, a genus peculiar to southern Africa, somewhat exceeding Stellatæ in number. And in New Holland, in the same parallel of latitude, the relative number of Stellatæ is still smaller, from the existence of Opercularia, a genus found only in that part of the world, and by the addition of which the proportion of the whole order to the Phænogamous plants is there considerably increased.

More than half the Rubiaceæ from Congo belong to well known genera, chiefly to Gardenia, Psychotria, Morinda, Hedyotis, and Spermacoce.

Of the remaining part of the order, several form new genera.

The first of these is nearly related to Gardenia, which itself seems to require subdivision.

The second is intermediate between Rondeletia and Danais, and probably includes Rondeletia febrifuga of Afzelius.^[74]

The third has the inflorescence and flowers of Nauclea, [443 but its ovaria and pericarpia are confluent, the whole head forming a compound spherical fleshy fruit, which is, I suppose, the country-fig of Sierra Leone, mentioned by Professor Afzelius.^[75]

The fourth is a second species of Neurocarpæa, a genus which I have named, but not described, in the catalogue of Abyssinian plants appended to Mr. Salt's Travels.^[76]

The fifth genus is intermediate between Rubiaceæ and Apocineæ. With the former it agrees in habit, especially in its interpetiolary stipules; and in the insertion and structure of its seeds, which are erect, and have the embryo lodged in a horny albumen forming the mass of the nucleus; while it resembles Apocineæ in having its ovarium entirely distinct from the calyx; its capsule in appearance and deliisccnce is exactly like that of Bursaria.

The existence of this genus tends to confirm what I have formerly asserted respecting the want of satisfactory distinguishing characters between these two orders, and to prove that they belong to one natural class; the ovarium superum approximating it to Apocineæ; the interpetiolary stipules and structure of seeds connecting it, as it appears to me, still more intimately with Rubiaceæ.

The arguments adduced by M. de Jussieu^[77] for excluding Usteria from Rubiaceæ and referring it to Apocineæ, are, its having ovarium superum, an irregular corolla, fleshy albumen, and only one stamen; there being no example of any reduction in the number of stamina in Rubiaceæ, (in which Opercularia and Pomax are not included by M. de Jussieu) while one occurs in the male flowers of Ophioxylum, a genus belonging to Apocineæ. From analogous reasoning he at the same time decides in referring Gærtnera of Lamarck^[78] to Rubiaceæ, though he admits it to have ovarium superum; its flowers being regular, its albumen more copious and horny, and its embryo erect. But all these characters exist in the new genus from Congo. These two genera therefore, together with Pagamea of Aublet, Usteria, Geniostoma of Forster (which is Anasser of Jussieu) and Logania,^[79] might, from their mere agreement in the situation of ovarium, form a tribe inter- 449] mediate between Rubiaceæ and Apocineæ. This tribe, however, would not be strictly natural, and from analogy with the primary divisions admitted in Rubiaceæ, as well as from habit, would require subdivision into at least four sections: but hence it may be concluded that the only combining character of these sections, namely, ovarium superum, is here of not more than generic value; and it must be admitted also that the existence or absence of stipules is in Logania^[80] of still less importance.

1. Fam. des Plant. 1, p. cxvi.
2. Flinders' Voyage to Terra Australis, 2, p. 538. (Antè, p. 8.)
3. Nova Genera et Species Plantarum, quas in perigrinatione orbis novi collegerunt, &c. Amat. Bonpland et Alex. de Humboldt, ex. sched. autogr. in ord. dig. C. S. Kunth, 1815, Parisiis.
4. Article "Greenland," in Brewster's 'Edinburgh Encyclopædia.'
5. That some change of this kind takes place on that coast might perhaps have been conjectured from a passage in Hans Egede's 'Description of Greenland,' where it is stated, that althongh from lat. 60° to 65° there is a considerable proportion of good meadow land, yet in the more northern parts, "the inhabitants cannot gather grass enough to put in their shoes, to keep their feet warm, but are obliged to buy it from the southern parts." (English Translation, pp. 44 and 47.)
6. Flinders' Voyage, 2, p. 539. (Antè, pp. 9, 10.)
7. Monogr. de la famille des Anonacées, p. 76.
8. Flore des Antilles, 1, p. 193.
9. Anonae, p. 113 et 112.
10. Dissert. 424, t. 247.
11. In Nov. gen. Madagasc. n. 46 (Biporeia).
12. Annal. du Mus. d'Hist. Nat. 18, p. 4S0.
13. Mem. du Mus. d'Hist. Nat. 2, p. 400.
14. Annal. du Mus. d'Hist. Nat. 18, p. 183.
15. Hist. des Véget. des Isles de l'Afrique, p. 183.
16. In Annal. du Mus. d'Hist. Nat. 4, p. 317.
17. Loc. cit., p. 348.
18. Annal. du Mus. d'Hist. Nat. 18, p. 476.
19. Mém du Mus. d'Hist. Nat. 3. p. 179.
20. Hist. des. Véget. des Isles de l'Afrique, p. 46.
21. Flore d'Oware, 1, p. 29, t. 17.
22. Mirbel, Elem. de Physiol. Veg. et de Bot. 2, p. 855.
23. Prolegomena, p. xviii. De Distrib. Geogr. Plant., p. 43.
24. Flinders's Voy. 2, p. 540. (Antè, p. 11.)
25. Op. citat.
26. Flinders's Voy. 2, p. 551. (Antè, p. 22.)
27. Coromand. Plants, 3 tab.
28. Reise nach Guinea, p. 116.
29. Aubert du Petit Thouars, nov. gen. Madagas. n. 80.
30. Herba sentiens, Rumph. Amboin. 5, p. 301.
31. Bruce in Philos. Transact. 75, p. 356.
32. Tab. affin. p. 23.
33. Mém. du Mus. d'Hist. Nat. 3, p. 448.
34. Flinders's Voy. 2, p. 545. (Antè, p. 16.)
35. Annal. du Mus. d'Hist. Nat. 8, p. 72.
36. Juss. Gen. 342. Parinari, Aublet Guian. 514. Petrocarya Schreb. Gen. 629.
37. Reise nach Guinea, p. 54.
38. Voyage au Senegal, 175.
39. Gen. Plant. 342.
40. Linn. Soc. Transact. 10, p. 35.
41. Flore d'Oware, 1, p. 94, t. 57.
42. Op. citat. 1. p. 69, t. 49.
43. Mém. de l'Institut. sc. phys. 1807, prem. semest. p. 11.
44. Tristemma virusana, Vent. Choix de Plantes, 35.
45. Mélanges de Botanique; Observ. address, à M. Lamarck, p. 57.
46. Flor, Ind. Occid. 2, p. 831, tab. 14.
47. Flinders's Voy. 2, p. 549. (Antè, p. 20.)
48. Roxburgh. Coromand. 3, p. 8, t. 211.
49. Nov. Gen. Madagasc. n. 84.
50. Or Barraldeia, Du Petit Thouars, Nov. Gen. Madagasc. n. 82.
51. Flinders's Voy. 2, p. 548. (Antè, p. 20.)
52. Nov. Gen. Madagasc. n. 81.
53. Hist. des Véget. des Isles de l'Afrique, 59.
54. Op. citat. 65.
55. Eclog. 1, p. 51, t. 9.
56. Annal. du Mus. d'Hist. Nat. 6, p. 102.
57. Ventenat in Mém. de l'Instit. Sc. Phys. 1807, 2 sem. p. 142.
58. Juss. Gen. Pl. 295. Ventenat Malmais, 27.
59. Vandelli Fasc. Pl. p. 7, t. 1. Ionidium, Venten. Malmais. 27
60. Flore d'Oware, 2, p. 10.
61. Hist. des Véget. des Isles de l'Afrique, 55.
62. The genera belonging to Berberideæ are Berberis (to which Ilex Japonica of Thunberg belongs); Leontice {including Caulophyllum, respecting which see Linn. Soc. Transac. 12, p. 145) Epimedium; and Diphylleia of Michaux. Jeffersonia may perhaps differ in the internal structure of its seeds, as it does in their arillus, from true Berberideæ, but it agrees with them in the three principal characters of their flower, namely, in their stamina being equal in number and opposite to the petals; in the remarkable dehiscence of antheræ; and in the structure of the ovarium. Podophyllum agrees with Diphylleia in habit, and in the fasciculi of vessels of the stem being irregularly scattered; essentially in the floral envelope, and in the structure of the ovarium; its stamina, also, though numerous, are not altogether indefinite, but appear to have a certain relation both in number and insertion to the petals: in the dehiscence of antheræ, and perhaps also in the structure of seeds, it differs from this order, to which, however, it may be appended. Nandina ought to be included in Berberideæ, differing only in its more numerous and densely imbricate bracteæ, from which to the calyx and even to the petals, the transition is nearly imperceptible; and in the dehiscence of its antheræ.
63. Mém. du Mus. d'Hist. Nat. 3, p. 440.
64. Annal. du Mus. d'Hist. Nat. 17, p. 153.
65. Nov. Gen. Madagasc. n. 78 et 79.
66. Dict. Elem. de Botanique par Bulliard, revu par L. C. Richard, ed. 1802, p. 34.
67. Or Ophiospermum of the Flora Cochinchinensis, as I have proved by comparison with a specimen from Loureiro himself.
68. The only other genus in which I have observed an analogous variety of form in the glands of the leaves, is Myroxylon (to which both Myrospermum and Toluifera belong), in all of whose species this character is very remarkable, the pellucid lines being much longer than in Samydeæ.
69. Linn. Soc. Transact. 11, p. 230.
70. Flinders's Voy. 2 p. 557. (Antè), p. 29.)
71. Linn. Soc. Transact. 12, p. 99.
72. In op. citat.
73. Flor. Egypt. Illustr. in Descript. de l'Egypt, Hist. Nat. v. 2, p. 49.
74. In Herb. Banks. This is the "New sort of Peruvian Bark" mentioned in his Report, p. 174; which is probably not different from the Bellenda or African Bark of Winterbottom's Account of Sierra Leone, vol. 2, p. 243.
75. Sierra Leone Report for 1794, p. 171, n. 32.
76. Voyage to Abyssinia, append. p. lxic. (Antè, p. 94.)
77. Annal. da Mus. d'Hist. Nat. 10. p. 323.
78. Illustr. Gen. tab. 167.
79. Prodr. Flor. Nov. Holl. 1, p. 455.
80. Prodr. Flor. Nov. Holl. 1, p. 455.