tively small number of myrmecophytes is rather surprising considering the abundance and variety of tropical plant life and the many opportunities which ants must have had to become acquainted with it.

(b) The African myrmecophytes belong to a few taxonomic types, represented by 7 families and 12 genera, as follows:

Schotia, Cola, Scaphopetalum, Barteria, Epitaberna, and Cuviera are precinctive Ethiopian genera, while the others are either also represented in the Oriental and Indomalayan Regions (Macaranga, Sarcocephalus, Plectronia) or tropicopolitan (Vitex, Uncaria, Randia). The family Rubiaceae leads the list with the largest number of myrmecophilous species (26, belonging to 5 genera), which is true also in other tropical regions. For Barteria, Epitaberna, and Cuviera, myrmecophytism is to all appearances one of the generic peculiarities, probably being present in all the members.

It is a curious fact that in the Ethiopian Region and elsewhere some of the largest families of the vegetable kingdom, in which differentiation into species has been most active, show very few (Leguminosa3, Orchidaceae) or no cases of myrmecophily. As illustrations of the latter may be mentioned the Gramineae, Cyperacese, Liliaceae, Labiatse, and Compositse.

(c) True myrmecophytes are restricted to the sections of the earth situated between the tropics, a fact easily accounted for hj the uniform temperature which prevails there and permits ants to establish their perennial abodes within the rather thin walls of plant tissues. I already have shown (p. 371) that the so-called ant acacias of the dry East African plains and Clerodendron formicarum of the savannah country are by no means myrmecophytes. When these cases are eliminated, all African antplants known at present occur only in the permanently moist and evergreen Rain Forest of the western and equatorial parts of the continent. All the Oriental and Indomalayan and the vast majority of the Neotropical myrmecophytes grow similarly in the moist tropical forest areas.

The one notable exception is presented by the true ant acacias of Central and South America (p. 510), which do not grow in the forests, but only in the open country or savannahs and along road-sides, and in some cases even prefer semiarid regions.

(d) The African myrmecophytes are all perennials and of a woody texture, either bushes, low trees, or woody creepers. This also holds true for the ant-plants of southern Asia, Malasia, and tropical America, though the types there are somewhat more varied, including, for instance, typical epiphytes. It is essential to the prosperity of the ant colonies that their permanency be assured for many years, a condition which, of course, could not be provided by annual or biennal plants. In addition, the woody texture of the walls adds considerably to the solidity of the domatia and to the protection of the formicaries. In a number of cases (Cola, Scaphopetalum), but not all, the leaves and stems of plants inhabited by ants are abundantly covered with long, stiff hairs.

(e) The structures offered as myrmecodomatia by the ant-plants show but little diversity, are usually of a very simple type, and affect few organs of the plant. There is nothing here comparable with the intricacy and endless variety of adaptations presented by entomophilous flowers to pollinating insects. The following types of myrmecodomatia have been recognized in Africa.

1. —The stipules persist for some time and are much swollen, their recurved margins enclosing a pouch-like cavity: Macaranga saccifera. A more primitive condition of stipular myrmecodomatia is illustrated by the Uragoga described on p. 453.

2. —The leaves produce pouches at the base of the blade: species of Cola and Scaphopetalum.

Swollen stipules and leaf pouches may be regarded as myrmecodomatia of a very primitive type. They are not much sought by the ants, probably because they do not offer enough solidity and permanency as shelters for formicaries. In the few cases in which I observed ants using the swollen stipules of Macaranga saccifera and the foliar pouches of Cola Laurentii and Scaphopetalum Thonneri, the colonies were very small and the ants timid.

3. —The stems of the plant arc externally normal, but hollowed out practically their entire length: Vitex Staudtii and Barteria Dewevrei.

4. —The stems present fistulose swellings either in the middle of the internodes (Randia Lujx and R. myrmecophyla), in, above, or below the nodes (Uncaria, Sarcocephalus, Plectronia, and Cuviera), or at the base of certain branches (Barteria fistulosa).

In other tropical regions there are a number of additional types not yet recognized in Africa, such as stipular thorns (Acacia), swollen petioles (Tachigalia, Nepenthes bicalcarata), pitcher-shaped leaves (Dischidia), inflated leaf-sheaths (Korthalsia), hollowed pseudobulbs (Schomburgkia), and fistulose rhizomes (Polypodium sinuosum, Lecanopteris carnosa, Myrmecodia, etc.)

In the case of stipular or leaf pouches, the slit which leads into the cavity is a natural result of the production of the pouch. In all other African myrmecophytes, there is no preformed entrance to the domatia and the apertures are gnawed by the ants.

(/) A very small number of African ants have become adapted to nesting in the domatia of ant-plants. A distinction should be drawn here between Obligatory plant ants, that live exclusively in myrmecophytes, and species which are only occasionally or accidentally associated with these plants and may therefore be designated as Facultative (Wheeler, 1913, p. 115). Most of the African plant ants fall in the second group; they belong to such genera as Crematogaster, Tetraponera, Monomorium, Leptothorax, Tetramorium, Cataulacus, Technomyrmex, and Prenolepis, which are abundant in the forest, usually leading an arboreal or semiarboreal life; many of the species make no distinction between cavities of dead or living plants wherein to shelter their formicaries. Viticicola tessmanni and the two species of Pachysima (P. xthiops and P. latifrons) are the only African obligatory plant ants. They have never been found away from their hosts, Vitex Staudtii in the case of Viticicola and various species of Barteria and Epitaberna myrmcecia in the case of the Pachysimse. It is possible that certain African species of Engramma and Plagiolepis, which have been collected only in plant domatia, are also of the obligatory type, but their case calls for further investigation.

There are a number of doubtful cases of myrmecophily among African plants and also others that are based on erroneous or incomplete observations. Some of these have been dealt with in the present paper under their respective families or genera, but a few others must be briefly mentioned here for the sake of completeness.

Stereospermum dentatum Richard (Bignoniaceae), of Abyssinia and Kordofan. According to Penzig (1894) the pith in the upper part of a flowering branch is excavated for a space of one or two internodes and the cavity is inhabited by Tetraponera penzigi (Emery), its offspring, and also some coccids. The aperture is found at the tip of what appears to be an aborted limb in the bifurcation of the flowering branch. There are no swellings and the normal stems are filled with pith. Penzig believes that the ants trim the growing upper end of the branch in order to enter the pith and are thus responsible for the dichotomous inflorescence of this species. I am rather inclined to think that the galleries are bored by some insect larva and are only settled by ants after being left by their maker.

Annibale (1907a) mentions two other African Bignoniaceae, Kigelia africana (Lamarck) and Newbouldia bevis (P. de Beauvois), as "myrmecophilous " because he found nectaria on the under side of the leaves. In addition, herbarium specimens of Newbouldia bevis examined by him were hollow in the upper part of the flowering branches, the cavities having one or two apertures at the base. The author assumes that these hollows are natural formations of the plant and are settled by the ants, which pierce the exit holes. He does not state that these insects were actually found in the branches, and the explanation offered above for similar cavities in Stereospermum is probably also true here. ,

Grumilea venosa Hiern (Rubiacese). Belgian Congo. "Bush of about 2 m., always inhabited by numerous black ants" (Dewevre; see De Wildeman and Durand, 1901, p. 130).

Microdesmis puberulaJ. D. Hooker (Euphorbiaceae). Belgian Congo. "Em. Laurent regarded this plant as a myrmecophyte; indeed some of the branches on specimens collected at Bombaie and provided with witch-brooms, are excavated with galleries; but the myrmecophy tic character is not much pronounced." (De Wildeman, 1910, 'Etudes Flore Bas- et Moyen-Congo,' III, 2, p. 250.)

In addition to the ants indicated in the general account of African myrmecophytes which follows, Father Kohl collected at Stanleyville and in nearby localities a number of species "in myrmecophilous plants" which have not been identified thus far in the literature. I subjoin a list of these insects, compiled from Forel's recent paper (1916) on the ants collected in the Belgian Congo by Kohl:

Crematogaster ruspolii variety atriscapis (Forel).

C. sjdstedti subspecies kohliella (Forel).

C. nigeriensis variety wilniger (Forel).

C. kasaiensis (Forel).

C.kohli (Forel).

C. solenopsides subspecies flavida variety convexiclypea (Forel).
Monomorium oscaris subspecies springvalense variety paternum Forel.
M. exiguum subspecies flavesrens Forel.
Leptothorax evelyna. Forel.

Tetramorium simiUimum subspecies isipingense variety dumezi Forel.

Engramma laurenti variety congolense Forel.
Prenolepis grisoni Forel.

Leguminous

Though this is one of the four or five largest families of plants and contains many of the more common bushes and trees of the tropics, only very few of its members are known to be myrmecophytes. After the elimination of the East African so-called "ant acacias," which, as I have shown elsewhere, do not possess true myrmecodomatia, there remains in Africa only one genus that possibly presents bioccenotic associations with ants.

Schotia Jacquin

Schotia Jacquin, 1786, 'Collectanea Austriaca ad Botanicam Chemiam et Historian! Naturalem Spectantia,'I, p. 93. Oliver, 1871, 'Flora of Tropical Africa,' II, p. 309. Harms, in Engler and Prantl, 1897, 'Die Naturl. Pflanzenfam.,' Nachtrage zu III, pt. 3, p. 196.

Theodora Medikus, 1786, 'Theodora speciosa, ein neues Pflanzengeschlecht,' p. 16. Taubert, 1894, in Engler and Prantl, 'Die Naturl. Pflanzenfam.,' Ill, pt. 3, p. 138.

"Unarmed trees or shrubs. Leaves abruptly pinnate, with coriaceous often small leaflets; stipules small. Flowers red or purple, clustered in short often dense panicles, heads or racemes. Bracts and bracteoles caducous or subpersistent. Calyx-tube turbinate, campanulate or narrowh' infundibuliform; segments 4, much imbricate. Petals 5, slightly unequal, clawed or subsessile, longer or shorter than the calyx, imbricate. Stamens 10, free or shortly coherent below; anthers uniform, dehiscing longitudinally. Ovary stipitate with elongate style and small terminal stigma; ovules 4 to 8 or 10, or more. Legume oblong, often falcate, compressed, coriaceous, dehiscent or subdehiscent. Seeds exalbuminous" (Oliver, 1871).

This genus belongs to the subfamily Caesalpinioideffi, in which the flowers are not of the papilionaceous type usual in the family, but possess a rather spreading, zygomorphous corolla; in the bud the upper sepals and petals are covered hy the lower. Schotia is restricted to tropical and southern Africa and contains twelve species, one of which is supposed to be myrmecophytic.

Schotia africana (Baillon)

Humboldlia africana Baillon, 1870, 'Histoire des Plantes,'II, p. 99, footnote (Tropical West Africa).

Schotia humboldtioides Oliver, 1871, 'Flora of Tropical Africa.' II, p. 310. Harms, 1915, in Engler, 'Die Pflanzenwelt Afrikas,' III, pt. 1, p. 454, fig. 249.

Theodora africana (Baillon) Taubert, in Engler and Prantl, 1894, 'Die Naturl. Pflanzenfam.,' Ill, pt. 3, p. 138.

"A glabrous tree of 25 to 30 feet; extremities (in our specimens) tumid immediately under each node, narrowing gradually nearly to the middle of the internode. Leaves % to 1 ft. long, 2- to 4-jugate, glabrous; leaflets thinly coriaceous, the