Fig. 88. Vitex Staudtii Guerke: a, external view of portion of stalk inhabited by the ant Viticicola tessmanni (Stitz); b, longitudinal section of an older stalk: e, entrances to cavities at nodes; i, lateral galleries excavated by ants through the xylem and ending blindly beneath the bark; le, accessory exit where one of the lateral galleries was gnawed through the bark. The lines A—A and B~B indicate the levels from which cross sections are figured in Pl. XXX, fig. 1 (A-A) and fig. 2 (B-B). Drawn from life at Avakubi, January, 1914; natural size.

has been suggested (Wheeler, 1913, p. 136) that ants may be able, through their, extremely delicate tactile (or rather chordonotal) senseorgans, to select the thinnest spot in the wall of a cavity for perforation. Their sense of smell may also warn them against gnawing parts of the stem containing certain distasteful substances.

A longitudinal section of the stalks (Fig. 886) discloses many features of further interest. In the first place, adult plants occupied by an ant colony are hollowed out nearly from top to bottom, all the internodes and various branches freely communicating with one another. The entire plant shelters one ant community, containing, in addition to one or more dealated queens, a number of fertile, ergatoid, wingless females. The formicaries of Viticicola tessmanni in the stems of Vitex Staudtii are thus splendid examples of polygynous insect societies. As in the case of Pachysima colonies in Barteria, they probably originate through secondary pleometrosis, or subsequent fusion of several isolated colonies, each started by a fecundated queen in the various limbs. A young specimen of Vitex Staudtii, scarcely 1 m. high, growing near the village of Nduye, between Penge and Irumu, was particularly instructive in this connection. Each of the lower internodes on the side-branches was occupied by a fertile, dealated female of Viticicola tessmanni, together with brood at various stages of development; no workers were present.

The ants clean out most of the medullary tissue nearly the entire length of the internodes, leaving only a peripheral layer of it for a short distance a little above the node. This remaining pith partly constricts the cavity and is probably left to keep the brood of the ants from dropping below the node, thus helping to distribute it regularly over the various internodes of the vertical stems and also preventing it from obstructing the apertures at the nodes. On a level with this inner circle of tissue the walls of the stem are also slightly thicker than in the other parts of the internode. At Kilo, in June, 1914,1 saw a very young Vitex Staudtii composed of an unbranched, leaved, erect, thin stem about two feet high and unoccupied by ants. The central cylinder of the whole plant was filled with soft medullary tissue. It is possible that this substance dries up by itself, causing the stems to be hollow without the intervention of ants. In nature, however, this must be rarely the case, for, in adult plants housing a colony of Viticicola, pith is found only in the topmost internode of very young branches which are still green and soft; the ants steadily work upward through the nodes and excavate the interior before it has begun to dry.

The inner walls of the hollow stalks also show a peculiar series of depressions or narrow channels, the like of which is not known for any other myrmecophyte. These lateral cavities perforate the xylem and end blindly just under the cambium; they are arranged at irregular intervals, one above the other, in two longitudinal rows. The rows are opposite each other and their position shifts at every node, so that they always run on the sides corresponding with the upper pair of apertures of every internode. The number of channels in a row varies with the length of the internode, in some cases there are fifty or more, but often fewer. It occasionally happens that one of these lateral galleries perforates the bark, and this supplementary exit hole then produces the same projecting ring of sclerenchyma which surrounds the normal, crater-like apertures at the nodes. Since no trace of lateral cavities is found in young internodes where the pith has not yet been removed by the ants, we must conclude that they are excavated by the workers of Viticicola. They are not used by the inmates for their eggs or very young larvae. .Coccids are not found in these channels and, furthermore, are absent from the hollow stalks of Vitex Staudtii. It was at first believed that the channels assist in the aeration of the hollow interior, but this is disproved by Prof. Bailey (see Part V, p. 586). He found that the bark outside the depressions presents no lenticels or patches of aerenchyma for the exchange of gases. On the contrary, the overlying tissues are compact and, in old stems, there are disks of impervious sclerenchyma located just opposite the blind ends of the cavities. Moreover, Prof. Bailey discovered that the channels are not natural gaps in the woody portion of the wall, but are excavated by the ants in peculiar cores of delicate, unlignified cells, that are symmetrically distributed in certain radii of the stem and surrounded by abnormal tissues similar to those' presented by heteroplasmatic zoocecidia. The arrangement of the galleries in two rows below the apertures of the upper node results from the fact that in Vitex Staudtii the principal water-conducting passageways in each internode are largely confined to those sides of the stele which pass out to the leaves at the next (higher) node. The lateral cavities are excavated in the sides of the stele poorly supplied with vessels and, furthermore, located in those portions of the xylem which are devoid even of a narrow fringe of small primary trachea;.

Prof. Wheeler has given a detailed description of Viticicola tessmanni (Stitz), the obligatory guest of Vitex Staudtii, in its various adult phases and larval stages. My observations in the field furnish no clue as to the possible food of these insects, but the ants are evidently adapted to their life within the cavities. Owing to the fact that the plant grows in swampy places, I am inclined to believe that the ants seldom, or never, leave their host. Wheeler and Bailey's examination of foodpellets dissected from the infrabuccal pockets of the adults and the trophothylaces of the larvae failed to reveal traces of food from an outside source. The insect substances in the pellets of the larvae resembled the yolk of ants' eggs and the fat-body of the larvse themselves, suggesting that some of the brood had been used as nourishment for the more vigorous progeny. In one instance pieces of the skin of a Viticicola, larva could be clearly recognized. There were also spores and bits of hypha; in many cases and particles that seemed to be pith and callus tissue (Wheeler and Bailey, 1920, p. 261). Bailey thinks that the principal food of Viticicola tessmanni is provided by the medullary tissue of young twigs and the "nutritive layer" which is produced in the lateral galleries of the domatia (see Part V, p. 606).

Viticicola tessmanni is exceedingly vicious and alert. When its host plant is ever so slightly disturbed, the workers rush out of the hollow stalks in large numbers and actively explore the plant. Their sting is extremely painful and sometimes produces vesicles on the skin. It is certain that they constitute a very efficient body-guard of theii host. Yet, on one occasion, I observed galls on the leaves of a Vitex occupied by the ants.

Rubiaceae

In Africa, as elsewhere, this family is the richest in myrmecophytes, and without doubt the fist of its species which form ccenobiotic associations with ants will be considerably increased by future investigation. Unfortunately many of the genera contain a large number of closelj' allied forms and even the generic distinctions are often unsatisfactory. It is, therefore, urgent that field-observations on these plants be accompanied by complete and abundant herbarium specimens for later identification by botanical experts.

So little is known about the two following cases that I have not treated them in the same detail as true myrmecophytes.

drum ilea venosa Hiern Grumilea venosa. Hiern, in Oliver, 1877, 'Flora of Tropical Africa,' III, p. 217. De Wildeman And Durand, 1901, 'Rcliquia; Devvevreanae,' p. 130. Th. And H. Durand, 1909, 'Sylloge Flor. Congol.,' p. 280. H. Kohl. 1909, Natur u. Offen1 arunfr, LV, p. 167.

Specimens collected by Dewevre in the Belgian Congo (Leopoldville; Bokakata) bear the following note: "Arbuste de 2 m. environ, toujours habit6 par de norabreuses fourmis noires." (De Wildeman and Durand, 1901. p. 130).

Uragoga species? In the forest bordering one of the affluents of the Congo near Leopoldville, I came across a semiherbaceous, low bush, which I provisionally refer to the genus Uragoga (May 18, 1915; Coll. No. 7656). The flowers are white, with greenish spots on the teeth of the corolla; the fruit is a red berry. At each node, between the points of attachment of the leaves, there are two curious, persistent stipules, occupying the entire width of the stem (Fig. 89). They are convexly swollen to the upper side and the free margin is recurved downward, the whole forming an inverted cup or pouch broadly open below. Coccids were usually found inside this cavity and the ants, Crematogaster striatula variety obstinata (Santschi), had built a tent of vegetable material over the inferior opening of the

stipules. I did not find eggs, larvae, or pupae of these ants inside the stipules, which I therefore regard not as myrmecodomatia but merely as "kraals" to shelter the scale insects. Yet this case suggests useful comparison with the stipular pouches of Macaranga saccifera and other more typical ant structures of plants.

Uncabia Schreber

Vncaria Schkeber, 1789, in Linnams, 'Gen. Plant.,' Ed. 8,1, p. 125. Hiern, 1877,

in Oliver, 'Flora of Tropical Africa.' Ill, p. 41. Haviland, 1897, Journ. Linn.

Soc. London, Botany, XXXIII, p. 73. Ovrovparia Auhlet, 1775, 'Histoire des Plantes de la Guiane fra115aise,' I, p. 177,

Pi. Lxviii. K. Schumann, 1891, in Engler and Prantl, 'Die Naturl. Pflanzen

fam.,' IV, pt. 4, p. 57. Agylophora Necker, 1790, 'Elementa Botanica,' 1, p. 145.

Climbing shrubs with opposite, interpetiolar, fugacious stipules; the lower part of the terminal branches with axillary recurved hooks, often spirally rolled up and placed opposite each other; in some cases these hooks still bear a few aborted, opposite leaves. On older branches the recurved hooks are often replaced by heavy, woody thorns. Leaves usually leathery, rarely herbaceous; the stipules entire or bifid. Flowers pedicelled or sessile, crowded into loose, globose heads, without intervening bracteoles. Flower heads placed in the axils of the upper leaves, either singly or in decussate panicles. Calyx salver-, or bell-, or funnel-shaped; the calyx-tubes