No. XXII.

BOTANICAL APPENDIX.

BY ROBERT BROWN, ESQ. F.R.SS. L. & E., F.L.S.

The Herbarium formed during the expedition, chiefly by the late Dr. Oudney, contains specimens, more or less perfect, of about three hundred species. Of these one hundred belong to the vicinity of Tripoli; fifty were collected in the route from Tripoli to Mourzuk, thirty-two in Fezzan, thirty-three on the journey from Mourzuk to Kouka, seventy-seven in Bornou, and sixteen in Haussa or Soudan.

These materials are too inconsiderable to enable us to judge correctly of the vegetable productions of any of the countries visited by the mission, and especially of the more interesting regions, Bornou and Soudan.

For the limited extent of the herbarium, the imperfect state of many of the specimens, and the very scanty information to be found respecting them, either in the herbarium itself, or in the Journal of the collector, it is unfortunately not difficult to account.

Dr. Oudney was sufficiently versed in Botany, to have formed collections much more extensive and instructive, had the advancement of natural history been the principal purpose of his mission. His time and attention, however, were chiefly occupied by the more important objects of the expedition: as a botanist he had no assistant; and the state of his health during his residence in Bornou must, in a great degree, have rendered him unable to collect or observe the natural productions of that country.

For the few specimens belonging to Soudan, we are indebted to Captain Clapperton, who, after the death of Dr. Oudney, endeavoured to preserve the more striking and useful plants which he met with. His collection was originally more considerable; but before it reached England, many of the specimens were entirely destroyed. It still includes several of the medicinal plants of the natives; but these being without either flowers or fruit, cannot be determined.

In the whole herbarium, the number of undescribed species hardly equals twenty; and among these not one new genus is found.

The plants belonging to the vicinity of Tripoli were sent to me by Dr. Oudney, before his departure for Fezzan. This part of the collection, amounting to one hundred species, was merely divided into those of the immediate neighbourhood of Tripoli, and those from the mountains of Tarhona and Imsalata.

It exceeds in extent the herbarium formed by Mr. Ritchie near Tripoli, and on the Gharian hills, which, however, though containing only fifty-nine species, includes twenty-seven not in Dr. Oudney’s herbarium.

The specimens in Mr. Ritchie’s collection are carefully preserved, the particular places of growth in most cases given, and observations added on the structure of a few; sufficient at least to prove, that much information on the vegetation of the countries he visited might have been expected from that ill-fated traveller.

In these two collections united, hardly more than five species are contained not already published in the works that have appeared on the botany of North Africa; particularly in the Flora Atlantica of M. Desfontaines, in the Flore d’Egypte of M. Delile, and in the Floræ Libycæ Specimen of Professor Viviani, formed from the herbarium of the traveller Della Cella.

The plants collected in the Great Desert and its oases, between Tripoli and the northern confines of Bornou, and which somewhat exceed a hundred, are, with about eight or ten exceptions, also to be found in the works now mentioned. And among those of Bornou and Soudan, which fall short of one hundred, very few species occur not already known as natives of other parts of Equinoctial Africa.

A complete catalogue of the herbarium, such as I have now described it, even if the number and condition of the specimens admitted of its being satisfactorily given, would be of but little importance, with reference to the geography of plants. Catalogues of such collections, if drawn up hastily, and from imperfect materials, as must here have been the case, are indeed calculated rather to injure than advance this department of the science, which is still in its infancy, and whose progress entirely depends on the scrupulous accuracy of its statements. To produce confidence in these statements, and in the deductions founded on them, it should in every case distinctly appear, that in establishing the identity of the species enumerated, due attention has been paid to the original authorities on which they depend, and, wherever it is possible, a comparison actually made with authentic specimens.

In the account which I am now to give of the present collection, I shall confine myself to a slight notice of the remarkable known plants it contains, to characters or short descriptions of the more interesting new species, and to some observations on such of the plants as, though already published, have either been referred to genera to which they appear to me not to belong, or whose characters require essential alteration.

In proceeding on this plan, I shall adopt the order followed in the botanical appendix to Captain Tuckey’s Expedition to the River Congo. And as there will seldom be room for remarks on the geographical distribution of the species I have to notice, I shall chiefly endeavour to make my observations respecting them of some interest to systematic botanists.

Cruciferæ. Fifteen species belonging to this family exist in the collection, one of which only appears to be undescribed, and of this the specimens are so imperfect, that its genus cannot with certainty be determined. Of those already published, however, the generic characters of several require material alterations, some of which suggest observations relative to the structure and arrangement of the natural order.

Savignya Ægyptiaca, (De Cand. Syst. 2. p. 283,) is the first of these. It was observed near Bonjem, by Dr. Oudney, whose specimens slightly differ from those which I have received from M. Delile, by whom this plant was discovered near the pyramid of Saqqârah, and who has well figured and described it in his Flore d’Egypte, under the name of Lunaria parviflora. By this name it is also published by M. Desvaux. Professor Viviani, in giving an account of his Lunaria libyca, a plant which I shall presently have occasion to notice more particularly, has remarked[87], that Savignya of De Candolle possesses no characters sufficient to distinguish it as a genus from Lunaria; and still more recently, Professor Sprengel has referred our plant to Farsetia[88]. The genus Savignya, however, will no doubt be ultimately established, though not on the grounds on which it was originally constituted; for the umbilical cords certainly adhere to the partition, the silicule, which is never absolutely sessile, is distinctly pedicellated in Dr. Oudney’s specimens, the valves are not flat, and the cotyledons are decidedly conduplicate. In describing the cotyledons of his plant as accumbent, M. De Candolle has probably relied on the external characters of the seed, chiefly on its great compression, its broad margin or wing, and on the whole of the radicle being visible through the integuments. It would appear, therefore, that the true character of the cotyledons of Savignya has been overlooked, chiefly from its existing in the greatest possible degree. To include this degree of folding, in which the margins are closely approximated, and the radicle consequently entirely exposed, a definition of conduplicate cotyledons somewhat different from that proposed in the “Systema Naturale” becomes necessary. I may here also observe, that the terms Pleurorhizæ and Notorhizæ, employed by M. De Candolle, to express the two principal modifications of cotyledons in Cruciferæ, appear to me so far objectionable, as they may seem to imply that in the embryo of this family, the position of the radicle is variable, and that of the cotyledons fixed. It is at least deserving of notice, that the reverse of this is the fact; though it is certainly not necessary to change these terms, which are now generally received.

On the subject of Savignya, two questions naturally present themselves. In the first place: Is this genus, solely on account of its conduplicate cotyledons, to be removed from Alyssineæ, where it has hitherto been placed, to Velleæ, its affinity with which has never been suspected, and to whose genera it bears very little external resemblance? Secondly: In dividing Cruciferæ into natural sections, are we, with M. De Candolle, to expect in each of these subdivisions an absolute uniformity in the state of the cotyledons? As far as relates to the accumbent and flatly incumbent states, at least, I have no hesitation in answering the latter question in the negative; and I believe that in one case, namely Hutchinsia, these modifications are not even of generic importance; for it will hardly be proposed to separate H. alpina from petræa, solely on that ground. I carried this opinion farther than I am at present disposed to do, in the second edition of Mr. Aiton’s Hortus Kewensis, where I united in the genus Cakile plants which I then knew to differ from each other, in having accumbent and conduplicate cotyledons; and I included Capsella bursa pastoris in the genus Thlaspi, although I was aware, both from my own observations, and from Schkuhr’s excellent figure[89], that its cotyledons were incumbent. I am at present, however, inclined to adopt the subdivision of both these genera, as proposed by several authors, and received by M. De Candolle; but to this subdivision the author of the Systema Naturale must have been determined on other grounds than those referred to; for in these four genera, in which the three principal modifications of cotyledons occur, he has taken their uniformity for granted.

As to the place of Savignya in the natural family, I believe, on considering the whole of its structure and habit, that it ought to be removed from Alyssineæ to a subdivision of the order that may be called Brassiceæ, but which is much more extensive than the tribe so named by M. De Candolle; including all the genera at present known with conduplicate cotyledons, as well as some others, in which these parts are differently modified.

There are two points in the structure of Savignya, that deserve particular notice. I have described the æstivation of the calyx as valvular; a mode not before remarked in this family, though existing also in Ricotia. In the latter genus, however, the apices of the sepals are perhaps slightly imbricate, which I cannot perceive them to be in Savignya.

The radicle is described by M. De Candolle as superior, with relation to the cotyledons. I am not sure that this is the best manner of expressing the fact of its being horizontal, or exactly centrifugal, the cotyledons having the same direction. This position of the seed is acquired only after fecundation; for at an earlier period the foramen of the testa, the point infallibly indicating the place of the future radicle, is ascendant. From the horizontal position of the radicle in this and some other genera, especially Farsetia, we may readily pass to its direction in Biscutella, where I have termed it descendant; a character which I introduced to distinguish that genus from Cremolobus. But in Biscutella the embryo, with reference to its usual direction in the family, is not really inverted, the radicle being still placed above the umbilicus. On the contrary, in Cremolobeæ, a natural tribe belonging to South America, and consisting of Cremolobus and Menonvillea, though the embryo at first sight seems to agree in direction with the order generally, both radicle and cotyledons being ascendant, it is, in the same sense, not only inverted, but the seed must also be considered as resupinate: for the radicle is seated below the umbilicus, and also occupies the inner side of the seed, or that next the placenta,—peculiarities which, taken together, constitute the character of the tribe here proposed. It appears to me singular that M. De Candolle, while he describes the embryo of these two genera as having the usual structure of the order, should consider that of Iberis, in which I can find no peculiarity, as deviating from that structure[90].

Lunaria libyca of Viviani[91] is the second plant of Cruciferæ, on which I have some observations to offer. This species was described and figured, by the author here quoted, in 1824, from specimens collected in 1817 by Della Cella. The specimens in the herbarium were found near Tripoli, where the plant had also been observed in 1819 by Mr. Ritchie, who referred it to Lunaria, and remarked that the calyx was persistent. Professor Sprengel, in his Systema Vegetabilium, considers it a species of Farsetia.

That this plant ought not to be associated either with the original species of Lunaria, or with Savignya, as now constituted, is sufficiently evident. And if it is to be included in Farsetia, it can only be on the grounds of its having a sessile silicule, with compressed valves, an indefinite number of seeds in each cell, and accumbent cotyledons. But in these respects it accords equally with Meniocus, a genus proposed by M. Desvaux, and with some hesitation received by M. De Candolle, and with Schivereckia of Andrzejowski, which he has also adopted. It does not, however, agree with either of those genera in habit, and it is easily distinguished from both by its simple filaments and other characters, which I shall notice hereafter. Is this plant, then, sui generis? ought it to be united with Alyssum, the character of that genus being modified to receive it? or does not Alyssum require subdivision, and may not our plant be referred to one of the genera so formed? A brief result of the examination of these questions, so far as they are connected with the subject under consideration, will be found annexed to the character which is given of the genus formed by the union of Lunaria libyca with Alyssum maritimum, a plant also in the collection, from the neighbourhood of Tripoli.

Alyssum maritimum, which is described both as an Alyssum and as a Clypeola by Linnæus, is the Konig of Adanson, who founded his generic distinction on the monospermous cells and supposed want of glands of the receptacle; and M. Desvaux, admitting Adanson’s genus, has named it Lobularia. In the second edition of Hortus Kewensis I included this plant in Alyssum, which M. De Candolle has also done in his great work.

For the genus here proposed I shall adopt Adanson’s name, altering only the termination, and wishing it to be considered as commemorating the important services rendered to botany by my friend Mr. Konig, of the British Museum[92]. In comparing these two species of Koniga, their agreement is very striking in habit, in leaves, in the closely pressed bipartite pubescence, in the calyx, petals, stamina, and stigma. They correspond also in some other points, less obvious but equally important, which I shall separately notice. The first of these is in having eight glands on the receptacle; a character peculiar, I believe, to these plants, and which first suggested the generic name Octadenia. The glands in Alyssum maritimum were entirely overlooked by Adanson, are not noticed by M. Desvaux, and M. De Candolle has described only the four that subtend the longer stamina. These certainly are much more conspicuous than the remaining four, which, however, occupy the place of the only glands existing in several of the most nearly related genera.

The number and position of the glands in this genus give some support, perhaps, to the hypothesis which I have formerly advanced, of the divisions of an hypogynous disk being in most cases formed of abortive filaments; an opinion more strikingly confirmed, however, in this family of plants, by their form and texture in Alyssum calycinum, and minimum.

The second point in which the two species of Koniga agree is in the structure of the septum. On this, which I consider as a new source of character in Cruciferæ, I shall offer some remarks in speaking of Farsetia.

The third point of agreement is the adhesion of the funiculi umbilicales to the septum. This adhesion, though really existing, is not very obvious in the monospermous cells of Koniga maritima; but in the supposed variety of this species from Teneriffe, in which the cells are occasionally dispermous, it is manifest, and is very remarkable in all states of Koniga libyca.

I first introduced this adhesion of the funiculi to the septum, as a generic character, in distinguishing Petrocallis from Draba. It has since been advantageously employed in the character of Lunaria by M. de Candolle, who, however, supposes this structure of much rarer occurrence in Cruciferæ than it really is. According to my observations, it is neither unfrequent, nor always of generic importance. Thus, I find it to exist in some species only of Arabis, namely A. Turrita, pendula, and canadensis, and hence I did not introduce it into my generic character of Parrya, though I have noticed it in my description of the species.

The principal difference existing between these two species of Koniga is that the cells of the ovarium and silicula of K. maritima are monospermous, while those of libyca are polyspermous, the number being variable, apparently indefinite, but not exceeding six. There are, however, other instances in this family, in which the mere difference between definite and indefinite number of seeds is of specific importance only, as in Draba and Meniocus, in each of which a species exists with dispermous cells; and the objection arising from the apparently still greater difference between unity and indefinite number in the two species of Koniga is removed by a supposed third species or variety of K. maritima, in which two seeds are occasionally produced in each cell. It may even be observed, that from unity to the indefinite number in this case, where the ovula in the different cells are alternate, the transition is perhaps more easy than from the binary to the indefinite, in cases where, as in Alyssum properly so called, the ovula are placed opposite in the different cells, and are in the same cell equidistant from its apex; this symmetry, probably, admitting of addition only by fours.

The next genus of Cruciferæ to be noticed is Farsetia, a fragment of the original species of which is in the collection. There are also several specimens of a plant, found in the desert, supposed to be new, and which, though without flowers, and considerably different in the form of its stigma, I am inclined, from the resemblance in habit, in pubescence, in silicula, in seeds, and especially from the exact similarity in the structure of the septum, to refer to the same genus[93].

As the introduction of the structure of the dissepiment into the generic characters of Cruciferæ is now proposed for the first time, and as I believe that its texture and appearance should always be attended to in constituting genera in this family of plants, I shall here offer a few remarks respecting it.

According to the particular view which I briefly but distinctly published in 1818, and which M. de Candolle first adopted in 1821, of the composition of the pistillum in Cruciferæ[94], the dissepiment in this family is necessarily formed of two lamellæ, derived from the parietes of the fruit. These lamellæ are in many cases easily separable, and where their union is more intimate, their existence is still evident from the want of correspondence, and consequent decussation, of their areolæ. The lamellæ, which are usually very thin and transparent, have their surface divided into areolæ, in different genera of very different forms, some of which may, with sufficient clearness, be described. In many cases no other appearance exists; in some, however, the axis of the septum resembles either a single nerve, or two distinct parallel nerves; and from this axis, whether formed of one or two nerves, tubes having the appearance and ramification of the veins of a leaf, and which generally terminate within the margin, not unfrequently proceed. This is remarkably the case in Farsetia, as I here propose to limit that genus; the central vessels in both its species being closely approximated, so as to form a single cord, extending from the apex to the base of the septum, and the veins being numerous and uncommonly distinct. Approaches more or less manifest to this structure of Farsetia exist in several other genera, as in Parrya, Savignya, and Koniga. But in this last mentioned genus the nerve, which originates, as in all cases, at the apex, hardly extends, even in the polyspermous species, beyond the middle of the septum, and the veins, which are much less distinct, are descendent.

As far as my observations on this subject at present extend, I expect, with great confidence, uniformity in the structure of the septum of strictly natural genera, and in many cases, though certainly not in all, I have found a resemblance in this respect in more extensive groups. Thus Draba, Arabis, and Aubrietia, agree in having amorphous areolæ, bounded by flexuose tubes or lines; while Alyssum, Berteroa, and Fibigia, have narrow linear areolæ, bounded by parallel or slightly arched lines. Capsella bursa differs from Thlaspi and Æthionema, as Draba from Alyssum, and agrees with Lepidium procumbens, Linn. improperly referred to Hutchinsia, and which equally has incumbent cotyledons. Cochlearia differs in like manner from Kernera. And numerous other examples of the same agreement in nearly related plants, and of differences where the usual sources of distinction are less available, might be noticed.

Hesperis nitens of Viviani is sparingly in the herbarium, both in flower and fruit. The seeds, though not ripe, are sufficiently advanced to show that the direction of the cotyledons is in this stage accumbent; and, as I have found in Cruciferæ generally that the ultimate agrees with the early state of cotyledons, I conclude they are likewise accumbent in the ripe seed. The plant is also abundantly different from Hesperis in other respects, and does not appear to be referrible to any genus yet published. This new genus[95] I have dedicated to the memory of Dr. Oudney, who found the present species in many of the wadeys between Tripoli and Mourzuk, and remarks that camels and mules eat it.

Hesperis ramosissima, which is also in the herbarium, was found in Fezzan. This plant differs in aspect from most of the other species of Hesperis, approaching in some points to Malcomia, in others to Mathiola; and as its cotyledons are very obliquely incumbent, it may form a section or subgenus, with a name, Hesperis (Plagiloba) ramosissima, indicating that character.

Capparideæ, of which eight species occur in the collection, is the family next to be noticed. I consider this order as belonging to the same natural class with Cruciferæ; and that this class includes also Resedaceæ, Papaveraceæ, and Fumariaceæ.

M. de Candolle, in defining Capparideæ, appears to regard the ovarium as having in all cases only two placentæ, and therefore formed of two pistilla or carpella. But to this, which is certainly the more usual number, there are many exceptions. These exceptions occur chiefly in the genus Capparis, which, as it is at present constituted, includes species differing from each other in having an ovarium with from two to eight placentæ, and, consequently, composed of an equal number of pistilla. Capparis spinosa is the most decided instance of the increased number of placentæ, and this, as well as some other nearly related species, are also remarkable in having septa subdividing the placentæ, and uniting in the centre of the compound ovarium.

In the herbarium there are three species of the genus Cleome. Two of these, C. pentaphylla and arabica, are in many respects well known plants; the third I believe to be an undescribed species, but nearly related to monophylla.

If the very natural group, formed by the Linnæan genus Cleome, is not to be preserved entire, its subdivision must be carried much farther, and established on other grounds, than has been done by M. de Candolle, whose genera and sections appear to me to have been equally founded on partial considerations. Thus, his Polanisia, uniting all the Cleomes whose stamina exceed six, contains in its first section, in addition to the species from which the genus was formed, at least two sets of plants, having very little affinity either with each other, or with the original species, whose only congener is placed in a second section.

Gynandropsis also consists of two groups not very intimately connected: the first is composed of species belonging to South America, and having the usual æstivation of the family: the second, of which C. pentaphylla may be taken as the type, is chiefly African, and is readily distinguished by its very different æstivation,—the great peculiarity of which consists in the petals not covering the stamina at any period. To this mode of æstivation of petals, which has never before been noticed, though it equally exists in Crateva and in Resedaceæ, I shall apply the term aperta. It is constantly conjoined, and, perhaps, necessarily connected with the early opening of the calyx, whose segments are originally connivent and slightly imbricate: for it may here be remarked, that in all the modifications of what I have termed imbricate æstivation of petals, they are, I believe, in the very early stage in like manner erect, and the sexual organs equally exposed.

If the expediency of preserving the genus Cleome entire were admitted, a question which I do not pretend at present to decide, it would still be of the greatest importance to arrange its numerous species according to their affinities, and carefully to distinguish the subordinate groups that compose it. To such inferior groups, whether termed subgenera or sections, names, in fact, have been of late years very generally assigned, both by zoologists and botanists.

It has not yet been proposed, however, that these subgeneric names should form an essential part of the name of the species; although, by employing them in this manner, while the principal groups would be kept in view, their subdivision would be carried to the same extent, and the subordinate groups as well expressed as if they had been actually separated into distinct genera.

The adoption of this method, which would not materially disturb names already existing, would probably lead to a greater consistency in the formation of genera, with reference to the natural orders of which they are subdivisions. In this way also the co-operation of two classes of naturalists, at present opposed to each other on the question of the construction of genera, might to a certain extent be expected, and greater uniformity in nomenclature consequently secured.

These advantages appear to me so important, that some expedient for obtaining them will, I am persuaded, at no distant period, be generally adopted.

In favour of the present plan it may be remarked, that it is analogous to the method followed by the Romans in the construction of the names of persons, by which not only the original family, but the particular branch of that family to which the individual belonged was expressed. Thus the generic name corresponds with the nomen (Cornelius), the name of the section with the cognomen (Scipio), and that of the species with the prænomen (Publius).

Without attempting at present to obviate the objections to which the proposed innovation is no doubt liable, I shall proceed to apply it to Cleome pentaphylla. According to my view the genus Cleome would include Gynandropsis, a name which, as that of a section, may be continued to those species of M. de Candolle’s genus belonging to equinoctial America, and having the common æstivation of the family: while Gymnogonia, derived from its remarkable æstivation, may be employed for the section that includes C. pentaphylla, of which the name might be given in the following manner:—

Cleome (Gymnogonia) pentaphylla. This plant, the earliest known species of Cleome, and that on which the genus was chiefly constituted, was found in Bornou. The species is regarded by M. de Candolle as a native of the West India islands, and he doubts whether it may not also belong to Egypt and India. On the other hand I consider it a native of Africa and India, and am not satisfied with the evidence of its being also indigenous to the American islands, where, though now very common, it has probably been introduced by the negroes, who use it both as a potherb and in medicine. It is not unlikely that M. de Candolle, in forming his opinion of the original country of this plant, has been in part determined by finding several species of his Gynandropsis decidedly and exclusively natives of the new continent. But if I am correct in separating these species from the section to which Cleome (Gymnogonia) pentaphylla belongs, this argument, which I have formerly applied to analogous cases[96], would be clearly in favour of the opinion I have here advanced; those species of the section with which I am acquainted being undoubtedly natives of Africa or of India.

Cleome (Siliquaria) Arabica, (Linn. sp. pl. ed. 2. p. 939. De. Cand. prodr. 1. p. 240), a supposed variety of which was found both in the neighbourhood of Tripoli and in Soudan, belongs to another subdivision of the genus, equally natural, and readily distinguishable. The species of this subdivision are included in M. De Candolle’s second section of Cleome, but are there associated with many other plants, to which they have very little affinity.

All the species of Cleome Siliquaria are indigenous to North Africa and Middle Asia, except violacea, which is a native of Portugal. Cleome deflexa of M. De Candolle (prodr. 1. p. 240.), founded on specimens in Mr. Lambert’s herbarium, which were sent by Don Joseph Pavon as belonging to Peru, seems to present a remarkable exception to this geographical distribution of the section. But on examining these specimens I find them absolutely identical with some states of violacea. I think it probable, therefore, either that they are erroneously stated to have come from Peru, or that this species may have been there introduced from European seeds.

Cadaba farinosa (Forsk. Arab. p. 68. De Cand. prodr. 1. p. 244) is in the herbarium from Bornou. The specimen is pentandrous, and in other respects agrees with all those which I have seen from Senegal, and with Strœmia farinosa of my catalogue of Abyssinian plants, collected by Mr. Salt, and published in his travels. M. De Candolle, who had an opportunity of examining this Abyssinian plant, refers it to his C. dubia, a species established on specimens found in Senegal, and said to differ from farinosa, slightly in the form of the leaves, and in being tetrandrous. Of the plant from Abyssinia I have seen only two expanded flowers, one of which is decidedly pentandrous, the other apparently tetrandrous. Mr. Salt, however, from an examination of recent specimens, states it to be pentandrous. It is probably, therefore, not different from C. farinosa of Forskal, whose specimens M. De Candolle has not seen. And as the form of leaves is variable in the specimens from Senegal, and not elliptical, but between oval and oblong, in those of Abyssinia, C. dubia is probably identical with, or a variety merely of farinosa, as M. De Candolle himself seems to suspect.

Crateva Adansonii (De Cand. prodr. 1. p. 243) is in the collection from Bornou. This species is established by M. De Candolle upon a specimen in M. de Jussieu’s herbarium, found in Senegal by Adanson, and is supposed to differ from all the other species in having its foliola equal at the base. I have examined the specimen in M. de Jussieu’s herbarium, in which, however, the leaves not being fully developed, I was unable to satisfy myself respecting their form. But in a specimen, also from Senegal, which I received from M. Desfontaines, the lateral foliola, though having manifestly unequal sides, are but slightly unequal at the base, and the inequality consists in a somewhat greater decurrence of the lamina on the anterior or inner margin of the footstalk. As well as can be determined, in very young leaves, this is also the case in the specimen from Bornou; and it is manifestly so in my specimen of C. læta, which appears to belong to the same species.

Crateva læta was founded by M. De Candolle on a plant from Senegal, communicated by M. Gay, from whom I also received a specimen in 1824, with the remark, that it was not different from C. Adansonii. In that specimen the flowers are male with an imperfect pistillum; in the plant from Bornou they are hermaphrodite, with elongated filaments; and in the specimen received from M. Desfontaines they are also hermaphrodite, but the stamina, though apparently perfect, are fewer in number and shorter than the stipes of the ovarium. I have observed, however, the flowers to be in like manner polygamous in some other species of Crateva, belonging both to India and America, a fact which materially lessens the dependence to be placed on characters taken from the number and length of the stamina in this genus.

Crateva Adansonii, it would appear, then, is the only known species of the African continent, for C. fragrans does not belong to the genus. And it will be difficult to distinguish this African Crateva from a plant which seems to be the most general species of India; except that in the latter, as in all the other species of the genus, the inequality of the lateral foliola, which is also more marked, consists in the greater decurrence of the lamina being on the outer or posterior margin of the footstalk. This Indian species, which may be named C. Roxburghii, is the Capparis trifoliata of Dr. Roxburgh’s manuscripts, but not Niirvala of Hortus Malabaricus (vol. 3. p. 49. t. 42), as he considers it. I have little doubt of its being also the plant described as C. Tapia, by Vahl, (symb. 3. p. 61.) his specific character well according with it, and not applying, as far as relates to the petals, to any known species of America. But as this character is adopted by Sir James Smith (in Rees’s Cyclop.), it may likewise be C. Tapia of the Linnæan herbarium; a conjecture the more probable as Linnæus has distinguished his Tapia by its ovate petals from gynandra, in which they are said to be lanceolate (Sp. pl. ed. 2. p. 637). This celebrated herbarium, however, is here of no authority, for Linnæus was never in possession of sufficient materials to enable him to understand either the structure and limits of the genus Crateva, or the distinctions of its species; and the specific name in question, under which he originally included all the species of the genus, ought surely to be applied to an American plant, at least, and if possible to that of Piso, with whom it originated. It is hardly to be supposed that the plant intended by Piso can now with certainty be determined; the only species from Brazil, however, with which I am acquainted, well accords with his figure and short description. This Brazilian species is readily distinguishable both from C. Adansonii and Roxburghii, by the form of its petals, which, as in all the other American species, are narrow-oblong or lanceolate; and from C. gynandra by the shortness of its stipes genitalium, or torus.

Crateva Tapia so constituted, is, on the authority of a fragment communicated by Professor Schrader, the Cleome arborea of that author, (in Gœtt. Anzeig. 1821, p. 707. De Cand. Prodr. 1. p. 242.); nor is there any thing in the character of C. acuminata of De Candolle (Prodr. 1. p. 243) which does not well apply to our plant.

C. Tapia, as given by M. De Candolle (op. cit.), is characterized chiefly on the authority of Plumier’s figure, in the accuracy of which, either as to the number or length of stamina, it is difficult to believe, especially when we find it also representing the petals inserted by pairs on the two upper sinuses of the calyx.

The genus Crateva agrees, as I have already stated, in the remarkable æstivation of its flower with Cleome Gymnogonia, by which character, along with that of its fruit, it is readily distinguished from every other genus of the order. Although this character of its æstivation has never before been remarked, yet all the species, referred to Crateva by M. De Candolle, really belong to it, except C. fragrans, which, with some other plants from the same continent, forms a very distinct genus, that I shall name Ritchiea, in memory of the African traveller, whose botanical merits have been already noticed.

Capparis sodada nob. Sodada decidua, Forsk. Arab. p. 81. Delile, Flore d’Egypte, p. 74. tab. 26. De Cand. Prodr. 1. p. 245.

The specimen in the herbarium is marked by Dr. Oudney as belonging to a tree common on the boundaries of Bornou. It is probably the Suag, mentioned in his journal, observed first at Aghedem, and said to be “a tetrandrous plant having a small drupa, which is in great request in Bornou and Soudan, for removing sterility in females: it is sweetish and hot to the taste, approaching to Sisymbrium Nasturtium;” and that “in passing the plant a heavy narcotic smell is always perceived.”

I have here united Sodada with Capparis, not being able to find differences sufficient to authorise its separation even from the first section of that genus, as given by De Candolle.

Forskal describes his plant as octandrous, and M. De Candolle has adopted this number in his generic character. M. Delile (op. cit.), however, admits that the stamina vary from eight to fifteen; and, in the specimen which I received from M. Jomard, I have found from fourteen to sixteen. But were the number of stamina even constantly eight, this alone would not justify its separation from Capparis, several octandrous species of which, belonging to the same section, are already known.

Another species of Capparis, also from Bornou, exists in the herbarium. It appears to be undescribed, and to belong to M. De Candolle’s first section of the genus; but the specimen is too imperfect to be satisfactorily determined.

Both these species have aculei stipulares, and it may here be remarked that all the plants belonging either to Capparis, or to any of the genera of the order whose fruit is a berry, in which these aculei are found, are indigenous either to Asia, Africa, or Europe; while all the aculeated Cleomes, with the exception of perhaps a single African species, are natives of equinoxial America.

Mærua rigida. This plant, of which flowering specimens were collected at Aghedem, certainly belongs to Forskal’s genus Mærua, adopted by Vahl and De Candolle; and I believe it to be a species distinct from the three already published. It is very nearly related, however, to a fourth species (M. Senegalensis nob.), of which I received a specimen from M. Desfontaines. M. De Candolle has placed the genus Mærua at the end of Capparideæ, between which and Passifloreæ he considers it intermediate. This view of its relation to these two orders I cannot adopt. To me it appears truly a Capparidea, having very little affinity with Passifloreæ, to which it seems to approach in one point only, namely, the corona of the calyx. But of a similar corona rudiments exist in several other African Capparideæ, and from some of these the genus Mærua is with difficulty distinguished[97].

Resedaceæ. The herbarium contains two species of Reseda. The specimens of one of these are too imperfect to be determined. The other is probably undescribed, though very nearly related to R. suffruticulosa, and undata of Linnæus. This supposed new species (Reseda propinqua) was found near Tripoli by Mr. Ritchie, and between Tripoli and Mourzuk by Dr. Oudney. It is remarkable in having the ungues of all the petals simple; that is, neither dilated, thickened, nor having any process or appendage at the point of union with the trifid lamina, into which they gradually pass. We have here therefore a species of Reseda with petals not different in any respect from those of many other families of plants; and, although this is an exception to their usual structure in the genus, I shall endeavour to show that all the deviations existing, however complex in appearance, are reducible to this more simple state of the organ.

Resedaceæ, consisting of Reseda, divisible into sections or subgenera, and Ochradenus, which may perhaps be regarded as only one of these subdivisions, I consider very nearly related to Capparideæ, and as forming part of the same natural class. It differs, in the variable number of the parts of its floral envelopes, from the other orders of the class, in which the quaternary or binary division is without exception; and it is especially remarkable in having the ovarium open even in its earliest state. From Cruciferæ and Capparideæ, the two families of the class to which they most nearly approach, Resedaceæ also differ in the apparent relation of the stigmata to the placentæ. The stigmata in this order terminate the lobes of the pistillum, and as these lobes are open sterile portions of the modified leaves, from the union of which in the undivided part I suppose the compound ovarium to originate, they necessarily alternate with the placentæ. I have generally found, however, the upper part of each placenta covered by a fleshy or fungous process, which is connected with the margins of the lobes, and therefore with the stigmata, and is probably essential to the fecundation of the ovula. The singular apparent transposition of the placentæ in Sesamoides of Tournefort, so well described by M. Tristan in his ingenious Memoir on the Affinities of Reseda[98], appears to me necessarily connected with the extreme shortness of the undivided base of the ovarium; for in supposing this base to be elongated, the placentæ would become parietal, and the ovula, which are actually resupinate, would assume the direction usual in the order.

M. De Jussieu, in his Genera Plantarum, has included Reseda in Capparideæ, and to this determination I believe he still adheres. M. Tristan, in the memoir referred to, is inclined to separate it as a family intermediate between Passifloreæ and Cistineæ, but more nearly approaching to the latter. M. De Candolle, who first distinguished Reseda as an order under the name here adopted, in 1819[99] placed it between Polygaleæ and Droseraceæ, and consequently at no great distance from Capparideæ. He must, since, however, have materially altered his opinion respecting it; for the order Resedaceæ is not included in the first or second part of his Prodromus, and I can find no observation respecting it in these two volumes. It is probable, therefore, that he may either intend to place it near Passifloreæ, as suggested by M. Tristan, or, which is more likely, that he has adopted the hypothesis lately advanced, and ingeniously supported, by Mr. Lindley, respecting its structure and affinities[100].

According to this hypothesis, in Reseda the calyx of authors is an involucrum, its petals neutral flowers, and the disk or nectary becomes the calyx of a fertile floret in the centre: and, as a deduction from this view of its structure, the genus has been placed near Euphorbiaceæ.

The points in the structure of Reseda, which appear to have led Mr. Lindley to this hypothesis, are the presence and appearance of the hypogynous disk, the anomalous structure of the petals, and the singular æstivation of the flower; but it is no slight confirmation of the correctness of M. de Jussieu’s opinion, that all these anomalies occur in a greater or less degree in Capparideæ, and have been found united in no other family of plants. The remarkable æstivation of Reseda equally exists in Crateva, and in more than one subdivision of the genus Cleome; the hypogynous disk is developed in as great a degree in several Capparideæ; and an approximation to the same kind of irregularity in the petals occurs in two sections of Cleome.

The analogical argument alone then might, perhaps, be regarded as conclusive against the hypothesis. But the question, as far as relates to the petals, and consequently to the supposed composition of the flower, may be decided still more satisfactorily on other grounds. Both M.M. Tristan and Lindley regard the upper divided membranaceous part of the petal as an appendage to the lower, which is generally fleshy. On the other hand, I consider the anomaly to consist in the thickening, dilatation, and inner process of the lower portion, and that all these deviations from ordinary structure are changes which take place after the original formation of the petal. To establish these points, and consequently to prove that the parts in question are simple petals, and neither made up of two cohering envelopes, as M. Tristan supposes, nor of a calyx and abortive stamina, according to M. Lindley’s hypothesis, I shall describe their gradual development, as I have observed it in the common Mignonette; a plant in which all the anomalies that have led to this hypothesis exist in a very great degree.

The flower-bud of Reseda odorata, when it first becomes visible, has the divisions of its calyx slightly imbricate and entirely enclosing the other parts. In this stage the unguis of each of the two upper petals is extremely short, not broader than the base of the lamina, and is perfectly simple; there being no rudiment of the inner process so remarkable in the fully expanded flower. The lamina at the same period may be termed palmato-pinnatifid, its divisions are all in the same plane, the terminating or middle segment is whitish or opake, and several times longer than the lateral segments, which are semitransparent.

Of the remaining four petals, the two middle are dimidiato-pinnatifid, their lateral segments existing only on the upper side; and the two lower are undivided, being reduced to the middle segment or simple lamina. All the petals are erect, and do not cover the stamina in the slightest degree, either in this or in any other stage. The disk is hardly visible. The Antheræ are longer than their filaments, of a pale-green colour; those on the upper or posterior side of the flower being manifestly larger, and slightly tinged with brown. The Pistillum is very minute and open at the top. In the next stage, the calyx is no longer imbricate, but open: the petals have their segments in nearly the same relative proportions; the interior margin of the unguis is just visible; but the transition from unguis to lamina is still imperceptible; the apex of the former not being broader than the base of the latter. It is unnecessary to follow the development through the more advanced stages of the flower, the facts already stated being, in my opinion, absolutely conclusive as to the real nature of the parts in question: and I may remark, that similar observations on certain genera of Caryophylleæ, especially Dianthus, Lychnis and Silene, clearly establish the analogy between their petals and those of Reseda.

I am aware that it has lately been proposed to include Datisca in Resedaceæ, to which it is nearly similar in the structure of its ovarium, as M. de Jussieu has long since remarked. But this is the only point of resemblance between them; for the calyx of Datisca is certainly adherent, and in most of its other characters it differs widely both from Reseda, and from every other genus yet published. Among the numerous discoveries made by Dr. Horsfield in Java, there is a genus, (Tetrameles nob.) however, manifestly related to Datisca, and remarkable in the regular quaternary division of every part of its diœcious flowers. These two genera form an order very different from every other yet established, and which may be named Datisceæ.

Caryophylleæ. Five species only of this family were collected near Tripoli, none of which are new.

Of Zygophylleæ, six species exist in Dr. Oudney’s herbarium, namely, Tribulus terrestris, found in Bornou; Fagonia cretica, from Tripoli to Benioleed; Fagonia arabica, at Aghedem; Fagonia Oudneyi nob. with Zygophyllum simplex in Fezzan; and Zygophyllum album every where in the desert.

This family, so distinct in habit from Diosmeæ or Rutaceæ, with which it was formerly united, is not easily characterized by any very obvious or constant peculiarities in its parts of fructification.

The distinguishing characters in its vegetation or habit are the leaves being constantly opposite with lateral or intermediate stipulæ, being generally compound, and always destitute of the pellucid glands, which universally exist in true Diosmeæ, though not in all Rutaceæ properly so called.

M. Adrien de Jussieu, in his late very excellent Memoir on the great order or class Rutaceæ, in distinguishing Zygophylleæ from the other subdivisions of that class in which he has included it, depends chiefly on the endocarp, or inner lamina of the pericarp, not separating from the outer lamina or united epicarp and sarcocarp, and on the texture of the albumen. His first section of Zygophylleæ, however, is characterised by the want of albumen; and in his second section I find exceptions to the remaining character, especially in Fagonia Mysorensis, in which the two laminæ of the ripe capsule separate as completely as in Diosmeæ. Another plant, in my opinion referrible to the same order, and which, in memory of a very meritorious African traveller, I have named Seetzenia africana, has in its ripe capsule the epicarp, or united epicarp and sarcocarp, confined to the dorsal carina of each cell, the endocarp being the only membrane existing on the sides, which are exposed long before the bursting of the fruit. The plant in question has indeed many other peculiarities, some of which may, perhaps, be considered sufficient to authorise its separation from the order to which I have referred it; for the æstivation of its calyx is valvular, it has no petals, its five styles are distinct to the base, and the cells of its ovarium appear to me to be monospermous. It completely retains, however, the characters of vegetation on which I chiefly depend in distinguishing Zygophylleæ; and I have no doubt of its being Zygophyllum lanatum of Willdenow[101], by whom it is stated to be a native of Sierra Leone; I suppose, however, on insufficient authority, for the specimens in the Banksian Herbarium, from which I have made my observations, were found in South Africa, near Olifant’s River, by Francis Masson.

In all the species of Fagonia, and in the two species of Zygophyllum in Dr. Oudney’s collection, a character in the fructification still remains, which is not found in Diosmeæ or Rutaceæ, and which, were it general in Zygophylleæ, would satisfactorily distinguish this order from all the families it has usually been compared with. This character consists in the direction of the embryo with relation to the insertion of the funiculus, its radicle being seated at the opposite extremity of the seed, or to express, in the unimpregnated ovarium, the infallible indication of this position, the direction of the inner membrane and nucleus of the ovulum corresponds with that of its testa.

But this character, in general very uniform in natural families, and which, equally existing in Cistineæ, so well defines the limits of that order, as I have long since remarked[102], would seem to be of less importance in Zygophylleæ.

M. Adrien de Jussieu, who, in his memoir already cited, admits its existence in Fagonia, and in both our species of Zygophyllum, considers it as an exception to the general structure of the latter genus, in the definition of which he retains the character of “radicula hilo proxima.” I believe, however, that in all the species of Zygophyllum, except Fabago, which possesses, also, other distinguishing characters, this opposition of the radicle to the external hilum will be found; for in addition to the two species contained in the herbarium, in both of which it is very manifest, I have observed it in Z. coccineum, and in all the species of South Africa that I have had an opportunity of examining. In some of these species, indeed, it is much less obvious, partly from the greater breadth of the funiculus, and also from its being closely applied, or even slightly adhering, to the testa of the seed. But hence it is possible to reconcile the structure of these species with that of Fabago itself, in which the raphe seems to me to be external: and if this be really the case, Fabago differs from those Zygophylla of South Africa alluded to, merely in the more intimate union of the funiculus with the surface of the testa. Whether this observation might be extended to the other genera of the order, I have not yet attempted to ascertain.

Balanites Ægyptiaca, though not belonging to Zygophylleæ, may be here mentioned. The specimen is from Bornou, but like all the other plants of that country, has no particular place of growth indicated, nor is there any observation respecting it. For a very full and interesting history of this plant, I may refer to M. Delile’s Flore d’Egypte (p. 77. tab. 28).

Of Cistineæ, three species were observed between Tripoli and Mourzuk.

The Geraniaceæ of the collection consist of four species of Erodium, all of which were found on the same journey.

Of Malvaceæ, considered as a class, there are twelve species in the herbarium. Only two of these are particularly deserving of notice. The first, Adansonia digitata, found in Soudan, where the tree is called Kouka, is described by Captain Clapperton; the second, Melhania Denhamii, a new and remarkable species of the genus, differing from all the others in having its bracteæ regularly verticillated, and, at the same time, longer and much broader than the divisions of the calyx.

A single species of Vitis is in the collection, from Bornou.

Neurada prostrata, generally referred to Rosaceæ, was found in Wady Ghrurbi.

Tamariscineæ. A species of Tamarix, apparently not different from T. gallica, is the Attil, common in Fezzan, where, according to Dr. Oudney, it is the only shady tree.

Lorantheæ. A species of Loranthus, parasitical on the Acacia nilotica, was observed very commonly from Fezzan to Bornou.

Leguminosæ. Of this class the herbarium contains thirty-three species, among which there are hardly more than two undescribed, and these belonging to a well-established genus.

Of the order or tribe Mimoseæ only three species occur, namely, Acacia nilotica, Mimosa Habbas, and Inga biglobosa, or a species very nearly related to it. Of this last named plant, I judge merely from ripe fruits adhering to the singular club-shaped receptacle, or axis of the spike. The specimens were collected in Soudan, and belong to a tree of considerable importance to the inhabitants of that country, by whom it is called Doura. According to Captain Clapperton, “The seeds are roasted as we roast coffee, then bruised, and allowed to ferment in water; when they begin to become putrid, they are well washed and pounded; the powder made into cakes, somewhat in the fashion of our chocolate; they form an excellent sauce for all kinds of food. The farinaceous matter surrounding the seeds is made into a pleasant drink, and they also make it into a sweetmeat.” The Doura of Captain Clapperton is probably not specifically different from the Nitta mentioned by Park, in his First Journey; nor from Inga biglobosa of the Flore d’Oware of M. de Beauvois, according to whom it is the Nety of Senegal; and he also well remarks, that Inga biglobosa, described by Jacquin as a native of Martinico, has probably been introduced into that island by the Negroes, as he himself found it to have been in St. Domingo.

Inga Senegalensis of M. De Candolle (Prodr. 2. p. 442) may also belong to the same species.

It is possible, however, that some of the plants here mentioned, though very nearly related to each other, and having all the same remarkable club-shaped spike, may be specifically distinct; for it appears from specimens collected at Sierra Leone by Professor Afzelius, that two plants having this form of spike are known in that colony; and two species, with similar inflorescence, probably distinct from those of Africa, are described in the manuscript Flora Indica of Dr. Roxburgh. All these plants possess characters fully sufficient to distinguish them from Inga, to which they have hitherto been referred. The new genus which they form, one of the most striking and beautiful in equinoxial Africa, I have named Parkia[103], as a tribute of respect to the memory of the celebrated traveller, by whom the fruit of this genus was observed in his first journey, and who, among other services rendered to botany, ascertained that the plant producing Gum Kino is a species of Pterocarpus[104]. I have formerly endeavoured to distinguish Mimoseæ from Cæsalpineæ, by the valvular æstivation of both its floral envelopes, and by the hypogynous insertion of its stamina. Instances of perigynous insertion of stamina have since been noticed by MM. Kunth and Auguste de St. Hilaire; but no exception has been yet pointed out to the valvular æstivation of their calyx and corolla. Parkia, however, differs from other Mimoseæ not only in its æstivation, which is imbricate, but in the very manifest irregularity of its calyx, and in the inequality of its petals, which, though less obvious, is still observable.

Erythrophleum, another genus indigenous to equinoxial Africa, which I have elsewhere[105] had occasion to notice, and then referred to Cæsalpineæ, more properly belongs to Mimoseæ, although its stamina are perigynous. In this genus, both calyx and corolla are perfectly regular, and their æstivation, if not strictly valvular, is at least not manifestly imbricate, though the flower-buds are neither acute nor angular. In Erythrophleum and Parkia, therefore, exceptions to all the assumed characters of Mimoseæ are found, and there is some approach in both genera to the habit of Cæsalpineæ. It is still possible, however, to distinguish, and it will certainly be expedient to preserve, these two tribes or orders. Abandoning divisions strictly natural, and so extensive as the tribes in question, merely because we may not be able to define them with precision, while it would imply, what is far from being the case, that our analysis of their structure is complete, would, at the same time, be fatal to many natural families of plants at present admitted, and among others to the universally received class to which these tribes belong. No clear character, at least, is pointed out in the late elaborate work of M. De Candolle[106], by which Leguminosæ may be distinguished from Terebintaceæ and Rosaceæ, the orders supposed to be the most nearly related to it. It is possible, however, that such characters, though hitherto overlooked, may really exist; and I shall endeavour to show that Leguminosæ, independent of the important but minute differences in the original structure and developement of its ovulum, may still be distinguished at least from Rosaceæ.

In the character of Polygaleæ, which I published in 1814[107], I marked the relation of the parts of the floral envelopes to the axis of the spike, or to the subtending bractea. I introduced this circumstance chiefly to contrast Polygaleæ with Leguminosæ, and to prove, as I conceived, that Securidaca, which had generally been referred to the latter family, really belonged to the former.

M. de Jussieu, who soon after published a character of Polygaleæ, entirely omitted this consideration, and continued to refer Securidaca to Leguminosæ. M. De Candolle, however, in the first volume of his Prodromus, has adopted both the character and limits of Polygaleæ, which I had proposed, though apparently not altogether satisfied with the description he himself has given of the divisions of the calyx and corolla.

The disposition of the parts of the floral envelopes, with reference to the axis of the spike, in Polygaleæ, namely, the fifth segment of the calyx being posterior or superior and the fifth petal anterior or inferior, is the usual relation in families the division of whose flower is quinary. This relation is in some cases inverted; one example of which I have formerly pointed out in Lobeliaceæ[108], as I proposed to limit it, and a similar inversion exists in Leguminosæ. But this class also deviates from the more general arrangement of the parts of the flower with regard to each other. That arrangement consists, as I have long since remarked[109], in the regular alternation of the divisions of the proximate organs of the complete flower. To this arrangement, indeed, many exceptions are well known; and M. De Candolle has given a table of all the possible deviations, but without stating how many of these have actually been observed[110].

In Leguminosæ the deviation from the assumed regular arrangement consists in the single pistillum being placed opposite to the lower or anterior segment of the calyx.

In these two characters, namely, the relation of the calyx and corolla both to the simple pistillum and to the axis of the spike or to the bractea, Leguminosæ differ from Rosaceæ, in which the more usual arrangements are found.

But in those Rosaceæ, in which the pistillum is solitary and placed within the anterior petal, its relation to the axis of the spike is the same as that of Leguminosæ, in which it is within the anterior division of the calyx. And in all families, whether dicotyledonous or monocotyledonous, this, I believe, is uniformly the position of the simple solitary pistillum with regard to the spike or bractea.

The frequent reduction of Pistilla, in plants having the other parts of the flower complete in number, must have been generally remarked. But the order in which these abstractions of pistilla take place, or the relations of the reduced series to the other parts of the flower, have, as far as I know, never yet been particularly attended to. It will probably appear singular, that the observation of these relations in the reduced series of pistilla should have suggested the opinion, that in a complete flower, whose parts are definite, the number of stamina and also of pistilla is equal to that of the divisions of the calyx and corolla united in Dicotyledones, and of both series of the perianthium in Monocotyledones.

This assumed complete number of stamina is actually the prevailing number in Monocotyledones; and though in Dicotyledones less frequent than what may be termed the symmetrical number, or that in which all the series are equal, is still found in decandrous and octandrous genera, and in the greater part of Leguminosæ. The tendency to the production of the complete number, where the symmetrical really exists, is manifested in genera belonging or related to those pentandrous families in which the stamina are opposite to the divisions of the corolla, as by Samolus related to Primulaceæ, and by Bæobotrys, having an analogous relation to Myrsineæ; for in both these genera, five additional imperfect stamina are found alternating with the fertile, and consequently occupying the place of the only stamina existing in most pentandrous families. Indications of this number may also be said to exist in the divisions of the hypogynous disk of many pentandrous orders.

With respect to the Pistilla, the complete number is equally rare in both the primary divisions of phænogamous plants. In Monocotyledones, the symmetrical number is very general, while it is much less frequent in Dicotyledones, in which there is commonly a still farther reduction.

Where the number of Pistilla in Dicotyledones is reduced to two, in a flower in which both calyx and corolla are present and their division quinary, one of these pistilla is placed within a division of the calyx, the other opposite to a petal or segment of the corolla. In other words, the addition to the solitary pistillum, (which is constantly anterior or exterior), is posterior or interior. This is the general position of the component parts of a bilocular ovarium, or an ovarium having two parietal placentæ; and in flowers whose division is quinary, I can recollect no other exceptions to it, than in some genera of Dilleniaceæ.

It is particularly deserving of notice, that the common position of the cells of the bilocular pericarpium with relation to the axis of the spike was well known to Cæsalpinus, who expressly distinguished Cruciferæ from all other bilocular families by their peculiarity in this respect, the loculi in that family being placed right and left, instead of being anterior and posterior[111].

On the subject of the position of the Pistilla in the other degrees of reduction from the symmetrical number, I shall not at present enter. But in reference to Leguminosæ, I may remark, that it would be of importance to ascertain the position of the Pistilla in the pentagynous Mimosea, stated to have been found in Brazil by M. Auguste de St. Hilaire[112]. Are these Pistilla placed opposite to the divisions of the calyx, as might probably be inferred from the position of the solitary Legumen in this class? Or are we to expect to find them opposite to the petals, which is the more usual relation, and their actual place in Cnestis, though the single ovarium of Connarus, a genus belonging to the same family, is seated within the anterior division of the calyx?

In the very few Leguminosæ in which the division of the flower is quaternary, namely, in certain species of Mimosa, the ovarium is still placed within one of the divisions of the calyx.

As to Moringa, which was originally referred to this class from a mistaken notion of its absolutely belonging to Guilandina, it is surely sufficiently different from all Leguminosæ, not only in its compound unilocular ovarium with three parietal placentæ, but also in its simple unilocular antheræ; and it appears to me to be an insulated genus, or family (Moringeæ), whose place in the natural series has not yet been determined.

Cæsalpineæ. Of this tribe, four species only occur in the collection. One of these is Bauhinia rufescens of Lamarck (Illustr. 329, f. 2.); another is Cassia (Senna) obovata, which, according to Dr. Oudney, grows wild in small quantities in Wady Ghrurbi.

Papilionaceæ. Twenty-six species of this tribe are contained in the herbarium, none of which form new genera, and the only two species that appear to be unpublished belong to Indigofera.

Alhagi Maurorum, or Agoul, is abundant in Fezzan, where it forms excellent food for camels.

Compositæ. Of this class, thirty-six species exist in the collection. The far greater part of these were found in the vicinity of Tripoli and in the Desert. All of them appear to belong to established genera, and very few species are undescribed.

Rubiaceæ. The herbarium contains only six species of this family, five of which, belonging to Spermacoce and Hedyotis, were found in Bornou and Soudan; the sixth, a species of Galium, near Tripoli.

Of Asclepiadeæ only three plants occur. One of these is a new species of Oxystelma, exactly resembling in its flowers O. esculentum of India, from which it differs in the form of its leaves, and in that of its fruit[113]. A species of Dœmia was found in the Desert; but the specimens are too imperfect to be ascertained.

Of Apocineæ, strictly so called, there is no plant whatever in the collection; and of Gentianeæ, a single species only of Erythræa.

Sesameæ. An imperfect specimen of Sesamum pterospermum, of the catalogue of Mr. Salt’s Abyssinian plants[114], is in the collection from Bornou.

Sapoteæ. The only plant of this family in the herbarium is the Micadania, or Butter Tree of Soudan, particularly noticed by Captain Clapperton. The specimen, however, is very imperfect, consisting of detached leaves, an incomplete fruit, and a single ripe seed. On comparing these leaves with the specimen of Park’s Shea Tree[115], in the Banksian Herbarium, I have little doubt that they both belong to one and the same species. Whether this plant is really a Bassia, is not equally certain; and the seed at least agrees better with Vitellaria paradoxa of the younger Gærtner, (Carpol. tab. 205.) than with that of Bassia, figured by his father, (de Fruct. et Sem. Pl. tab. 104.)

That the woody shell in the nuts of all Sapoteæ is really formed of the testa or outer membrane of the seed, as I have elsewhere stated[116], and not of a portion of the substance of the pericarpium, according to the late M. Richard and the younger Gærtner, is proved not only by the aperture or micropyle being still visible on its surface, as M. Turpin has already shown in one case, (Ann. du Mus. d’Hist. Nat. 7, tab. 11, f. 3.); but also by the course and termination of the raphe, as exhibited in the younger Gærtner’s figures of Calvaria and Sideroxylum, (Carpol. tabb. 200, 201, et 202.) and by the origin and ramification of the internal vessels.

Scrophularinæ. Only six species of this family occur, none of which are unpublished.

Orobanche compacta of Viviani was observed between Fezzan and Bornou.

Of Convolvulaceæ there are five species, four of which belong to Bornou; the fifth is an aquatic Ipomœa, found creeping on the borders of a small lake near Tintuma. Possibly this plant may be Ipomœa aquatica of Forskal, and consequently Convolvulus repens of Vahl, (symb. 1, p. 17.) It is not, however, the plant so called by Linnæus, which proves, as I have elsewhere stated, (Prodr. Fl. Nov. Holl. 1, p. 483.) to be Calystegia sepium; nor does it belong to either of his synonymes. Our plant differs also from Vahl’s description of his Convolvulus repens, in having constantly single-flowered peduncles, and leaves whose posterior lobes are rather acute than obtuse, and are quite entire. It is probably, therefore, distinct; and I have named it Ipomœa Clappertoni[117].

Among the few Labiatæ, there is a species of Lavandula, possibly distinct from but very nearly related to L. multifida. It was found on the mountains of Tarhona.

Of Boragineæ, the herbarium includes eleven species, the greater part of which were collected near Tripoli, and all of them belong to well established genera.

Primulaceæ. Of this family two species of Anagallis occur in the collection, and of these A. cærulea was observed both near Tripoli and in Bornou.

Samolus Valerandi was also found near Tripoli, in Wady Sardalis in Fezzan, and in Bornou.

Of Dicotyledonous, or even of all phænogamous plants, S. valerandi is perhaps the most widely diffused. It is a very general plant in Europe, has been found in several parts of North Africa, in Dr. Oudney’s herbarium it is from Bornou, I have myself observed it at the Cape of Good Hope and in New South Wales, and it is also indigenous to North America.

The geographical distribution of the genus Samolus is equally remarkable. At present eight species are known, of which S. Valerandi is the only one indigenous to Europe, or which, indeed, has been found in the northern hemisphere, except the nearly related S. ebracteatus of Cuba. All the other species belong to the southern hemisphere, where S. Valerandi has also a very extensive range.

Of Plumbagineæ, there are three species of Statice Taxanthema; for the latter name may be preserved as belonging to a section, though hardly as that of a genus, so far at least as depends on inflorescence, which in both subdivisions of Statice is essentially similar; that of Statice Armeria being only more condensed. Of the three species in the herbarium, one appears to be unpublished.

Among the plants of the Apetalous orders in the collection, there are very few remarkable, and hardly any new species.

Gymnocarpus decandrum was observed by Dr. Oudney very commonly in gravelly deserts, on the route from Tripoli to Fezzan; and Cornulaca monacantha of M. Delile is said to be widely extended from Tripoli to Bornou, and to be excellent food for camels.

Monocotyledones. The number of species belonging to this primary division contained in the herbarium is altogether seventy. But Gramineæ and Cyperaceæ being excluded, thirteen only remain, namely, three species of Juncus, a single Commelina, three Melanthaceæ, three Asphodeleæ, one species of Iris, and two Aroideæ, of which Pistia Stratiotes is one.

Of these thirteen plants, two appear to be unpublished, both of them belonging to Melanthaceae. The first, a congener of Melanthium punctatum, which is also in the collection, was found in Fezzan.

The second is a species of Colchicum, very different from any hitherto described; and which yet, by Mr. Ritchie, who first observed it, is said to be common in the desert near Tripoli, where it was also found by Dr. Oudney.

This species, which I have named Colchicum Ritchii, is easily distinguished from all its congeners by having two cristæ or membranous processes which are generally fimbriated, at the base of each segment of the perianthium, parallel to each other, and to the intermediate filament. But this character, though excellent as a specific difference, is neither of generic importance, nor sufficient to authorise the formation of a separate section[118].

Bulbocodium and Merendera, however, which, following Mr. Ker[119], I consider as belonging to Colchicum, appear to me decidedly to form subgenera or sections; and in this opinion I am confirmed by having found a fourth section of the same genus. This fourth subgenus is established on Hypoxis fascicularis, a plant which has been seen by very few botanists, and which Linnæus introduced into his Species Plantarum, and referred to Hypoxis, solely on the authority of the figure published in Dr. Russell’s History of Aleppo. In the Banksian Herbarium I have examined part of the original specimen of this species, found by Dr. Alexander Russell, and figured by Ehret in the work referred to, as well as more perfect specimens collected by Dr. Patrick Russell; and am satisfied that its ovarium is not in any degree adherent to the tube of the perianthium. I find, also, that Hypoxis fascicularis differs from Colchicum merely in having a simple unilocular ovarium with a single parietal placenta and an undivided style, instead of the compound trilocular ovarium with distinct or partially united styles, common to all the other sections of that genus.

A reduction, as in this case, to the solitary simple pistillum[120], though existing in all Gramineæ and in certain genera of several other families of Monocotyledones, is yet comparatively rare in that primary division of phænogamous plants, and in the great class Liliaceæ, the present species of Colchicum offers, I believe, the only known example. Yet this remarkable character is here so little influential, if I may so speak, that Hypoxis fascicularis very closely resembles some states of Colchicum Ritchii, and in the Banksian herbarium has actually been confounded with another species of the first or trigynous section of the genus.

To the first section, which includes Colchicum Ritchii, the subgeneric name Hermodactylum may, perhaps, be applied; while that established on Hypoxis fascicularis may be called Monocaryum.

The position of the pistillum in Colchicum (Monocaryum) fasciculare is not easily determined. I believe it to be placed within the anterior segment of the outer series of the perianthium, but, from the great length of the tube, it is difficult to ascertain such a point in dried specimens. This, however, is the position in which I should expect it, both in reference to the usual relation of the solitary simple pistillum to the axis of the spike, or to the subtending bractea in all phænogamous plants; and also with regard to the constant relation of the parts of the compound pistillum to the divisions of the perianthium in Monocotyledones: for it is worthy of remark, that a difference in this relation may be said to exist in the two primary divisions of phænogamous plants—the pistilla when distinct, or their component parts when united, being in Dicotyledones usually placed opposite to the petals, when these are of equal number; while in Monocotyledones the cells of the trilocular ovarium are, I believe, uniformly opposite to the divisions of the outer series of the perianthium.

Cyperaceæ. Of twelve species of this family existing in the herbarium, six are referrible to Cyperus, three to Fimbristylis, and three to Scirpus. Among these there is no remarkable, nor, I believe, any undescribed species. Of C. Papyrus, which, according to Captain Clapperton, grows in the Shary, there is no specimen in the collection.

Gramineæ. Of this extensive family, with which Dr. Oudney was more conversant than with any other, and to which, therefore, during the expedition, he probably paid greater attention, the herbarium contains forty-five species: and in dividing the order into two great tribes, as I have formerly proposed[121], thirty of these species belong to Poaceæ, and fifteen to Paniceæ. This relative proportion of these two tribes is considerably different from what might have been expected, in the climates in which the collection was formed: it seems, however, to be connected with the nature of the surface, for in the Great Desert the reduction of Paniceæ is still more remarkable; this tribe being to Poaceæ, in that region, in the proportion of only five to eighteen.

Dr. Oudney remarks with respect to the grasses of the desert, that he observed no species with creeping roots; for a species of Arundo related to Phragmites, which he notices as the only exception, is not properly a desert plant.

Among the very few Gramineæ deserving particular notice, the first is Avena Forskalii of Vahl. The specimens in the herbarium which were collected in the desert of Tintuma in some respects differ from all the others that I have seen of this variable species. In the Banksian herbarium there is an authentic specimen from Forskal; I have received from M. Delile specimens both of his A. Forskalii and arundinacea, described and figured in his Flore d’Egypte; and am also in possession of others in somewhat different states, collected in Egypt by M. Nectoux and Dr. Sieber. From a comparison of all these specimens, I am led to believe, that A. Forskalii and arundinacea are not specifically distinct; and it is at least evident, that arundinacea more nearly approaches to the plant of Forskal than that to which M. Delile has applied the name Forskalii.

This grass, which does not belong to Avena, is referrible to Danthonia, from the structure of the outer valve of its perianthium. But Danthonia requires subdivision into several sections, of which, perhaps, our plant may be considered as forming one.

The character of the section established on Danthonia Forskalii would chiefly consist in the very remarkable obliquity of the joints of the locusta, which is, indeed, so great, that after their separation each flower seems to have at the base an almost vertically descendent spur; and as the inferior extremity of the upper joint is produced beyond the lower, a short calcar actually exists before separation, and this calcar is equally manifest in the terminal rudiment of the locusta. The present, therefore, is a case of more remarkably oblique articulation in grasses than even that existing in Holcus acicularis (Andropogon acicularis, Retz) which led to the formation of Centrophorum; a genus still admitted by Professor Sprengel[122], and respecting the structure of which a very singular explanation has been lately offered by M. Raspail[123]. In one respect, the two cases differ. In Danthonia (Centropodia) Forskalii, the articulations being in the axis of the locusta or spicula, each flower appears to have this spur-like process; while in Holcus (Rhaphis) acicularis, the joint being in the peduncle or branch of the racemus, the spur is common to three locustæ.

Dr. Fischer, in whose herbarium the specimen was observed which led to the formation of Centrophorum, will probably recollect the communication made to him on the subject of that plant, of which Dr. Trinius himself has since corrected the characters. He retains it, however, as a distinct genus, for which he has adopted the name Rhaphis, given to it by Loureiro, by whom it was originally proposed, on other, but not more satisfactory grounds.

Triraphis Pumilio is the second plant of this family to be noticed. It is undescribed, and belongs to a genus of which the only two published species were found in the intratropical part of New Holland[124]. In several points of structure the African plant is very different from T. pungens, the first of these species; in some respects it approaches to mollis, the second species, especially in the inequality of its setæ or aristæ, but it differs from both in habit, and in having only one perfect flower in each locusta[125].

Of Pennisetum dichotomum, (Delile, Flore d’Egypte, p. 15, tab. 8, f. 1.), which, in several different states, is in the collection, it is remarked by Dr. Oudney, that “it is a great annoyance to man and beast, from the prickly calyx (involucrum);” and by Major Denham, that from Aghedem to Woodie “it covered the surface of the country, and annoyed the travellers to misery;” he observes, also, that the seed is called Kasheia, and is eaten.

Panicum turgidum, (Forsk. Arab. p. 18; Delile, Flore d’Egypte, p. 19, tab. 19, f. 2.), is also one of the most common grasses from Tripoli to Bornou.

Of Acotyledones, the only plant in the collection is Acrostichum velleum, found on the Tarhona mountains. Mr. Ritchie’s herbarium contains, also, a single plant of the same family, namely Grammitis Ceterach.

The foregoing observations have extended much beyond the limits which the number and importance of the plants they relate to may seem to require. I still regret, however, that I cannot add a few remarks on such species as, although not in the herbarium, were observed, either indigenous or cultivated, in the countries visited by the mission, and for information respecting which I am indebted to Major Denham and Captain Clapperton. But it being determined no longer to delay the publication of the very interesting Narrative, to which the observations already made will form an Appendix, I am unable at present to enter on this part of my subject.

FOOTNOTES: