of slender and narrow bracteoles of about 2.5 cm. The calyx is pale green, deeply cut into 3 to 5 elongate-lanceolate lobes, extending far over the corolla, and 1.5 to 7.5 cm. long. The corolla is short salvershaped (hypocrateriform), fleshy coriaceous, mostly deep green; the tube is short, bright carmine red, which color extends as a median acuminate line or triangular spot over the upper side of each of the five lobes. These five corollar lobes are lanceolate or oval-acuminate, rigid, and spread into a star when in full blossom. The tube of the corolla bears inside a ring of silvery, shiny hairs directed downward. The five stamens are exserted, placed in the sinuses of the corollar lobes, around the ring of hairs; their filaments are slightly flattened, fleshy, carmine red; their anthers are yellow. The style is thick, columnar, carmine red, densely white pilose, and ends in a large, obtuse, cap-shaped, pale green stigma. Frequently the fruit is oblique or even curved and falcate; but this is due to the aborting of one or more of the ovules; when the fruit is normally developed it is an obovate, dirty yellow drupe, about 23 to 32 mm. long and 18 to 25 mm. thick; its surface is deeply furrowed, there being 5 heavily developed ribs with less prominent ones between them; the ripe fruit is crowned with the remains of the withered calyx. It is noteworthy that very few of the flowers produce fruit.

Cuviera angolensis was recognized as a myrmecophyte by K. Schumann (1890, p. 121), who found unidentified ants in the domatia of Welwitsch's herbarium specimens. The first field-notes on this plant were made by Dewevre in 1896 (De Wildeman and Durand, 1900, p. 124) and these were completed by £m. Laurent (De Wildeman, 1916, pp. 296-299) and H. Kohl (1909, pp. 163-166). Their accounts agree in almost every detail with my own observations on Congo Cuvierse as reported below. The following ants are known from C. angolensis.

Crematogaster africana (Mayr). Rom6e (H. Kohl, 1909, p. 164; Forel, 19096, p. 69).

C. africana subspecies laurenti (Forel). Romee (H. Kohl, 1909, p. 164; Forel, 19096, p. 69). In that locality the coccid Stictococcus formicarius Newstead was living inside swellings of C. angolensis also occupied by this ant (Newstead, 1910, p. 19).

C. africana subspecies winkleri (Forel). Eala (fim. Laurent; see Forel, 19096, p. 69).

Cuviera angolensis Hiern was the only member of the genus recorded from the Belgian Congo, where it is far from rare. I had opportunity to examine in several localities a number of Cuvierx, all of which at the time I regarded as belonging to this species, since they agreed with its description. While studying the anatomy cf Cumera, Prof. Bailey discovered certain histological dissimilarities between specimens collected at different places, yet it is possible that these discrepancies are due either to the difference in the age of the various branches or to their mode of preservation. In view of the fact that the number of African species has been so increased recently, the Cumerse of the Congo Basin will need considerably more field study before their identity can be safely discussed. Meanwhile, my notes are presented separately for each of the specimens I examined.

1.—At Avakubi (January 6, 1914; Coll. No. 1796) a Cumera was found growing on the banks of the Ituri River. It was a low bush (4 to 5 meters high), well answering the general description given above, but without flowers or fruit; flower buds were, however, noticed a few days later on another specimen in the same locality. The trunk was cylindrical, and neither swollen nor hollow. Most of the branches showed at each node a spindle-shaped swelling which extended over the lower two-thirds to three-quarters of the internode, and was about 8 to 10 mm. thick and 6 to 7 cm. long. Notwithstanding the fact that expanded portions were almost completely hollow, their solid, woody walls made them very resistant to pressure. They were present even on young limbs and early became hollow, through the drying of the medullary tissue, before being attacked by ants.

On some of the branches the swellings contained a beetle larva feeding on the remains of dried pith, but there was always an orifice by which the insect had entered the stem. Some of these beetle larvae were accompanied by coccids, though no ants were associated with them in the cavity. This is of great interest because it shows that the coccids enter the domatia of their own accord as soon as an aperture is pierced.1

The majority of the swellings of older hmbs were inhabited by ants of different species, the most common being an unidentified Crematogaster. I further collected in other domatia of the same plant Cataulacus pilosus Santschi and Technomyrmex hypoclinoides Santschi. All of them had established in the cavities regular formicaries with larva; and pupae. In the case of the swellings tenanted by Crematogaster, each sheltered a separate colony, with its own queen, a number of workers, and abundant brood. Furthermore, the younger swollen internodes on the upper end of the branches were often occupied by a solitary queen, some

1Kohl (1909. p. 165) also mentions the presence of an insect larva, together with scale insects, in some of the swellings of Cumera angolensis.

Fig. 99. CuvieraangolensisHiem: a, upper portion of branch with one of thecaulinary myrmecodcmatia above the node; 6, longitudinal section of one of the swellings; e, aperture gnawed by the ants and leading into the cavity; d, pits often occupied by coccids. Drawn from alcoholic specimens collected at the Tshopo River, near Stanleyville; natural size.

times in company with a few coccids, the exit hole being partly closed by callus growth. Working down the branch, one frequently met with all stages in the development of the colony, ending with the appearance of the winged, sexual phases. It thus appears that the various colonies in a single Cuviera do not fuse into one great community as is the case with the Pachysimse of Barteria and the Viticicolse of Vitex Staudtii, yet they manage to live peacefully side by side.

The three species of Formicidae found" in this Cuviera were small and timid, and showed no aggressiveness, remaining inside the domatia when the plant was disturbed. They could not have been of much value as guards against phytophagous insects or other enemies. Indeed, numerous leaves of a specimen densely populated by ants were noticeably eaten by caterpillars.

2. —A Cuviera found at Penge (February 13,1914; Coll. No. 2461), along the Ituri River, agreed in every particular with the specimens from Avakubi described above. It was in full bloom and all of its swellings were occupied by an unidentified Crematogaster.

3. —Another Cuviera collected near the village ofMasaki, between Masisi and Walikale (December 31, 1914; Coll. No. 6429), also agreed entirely with the plants from Avakubi. Its swellings were occupied by two different ants, Engramma denticulatum Wheeler and Tetramorium meressei Forel, each in domatia of its own.

4. —The above remarks further apply to a Cuviera collected in fruit near Sitaweza, between Walikale and Lubutu (January 13, 1915; preserved in my herbarium without Coll. No.). In this case the inhabitants were Crematogaster excisa subspecies andrei (Forel).

5. —Along the Tshopo River, near Stanleyville, Mr. H. Lang and I collected, March 8,1915, much material of Cuviera which was abundantly settled by the ant Crematogaster africana subspecies laurenti variety zeta (Forel). Figure 99, drawn from alcoholic specimens, shows the outer and inner structure of the domatia, which were in every respect similar to those of the plants observed at Avakubi, Penge, Masaki, and Sitaweza. Prof. Bailey states that, compared with the swellings of the Kunga specimens, those of the plants from the Tshopo are "shorter, slimmer, and of a deep olive green color"; in addition, the cortex and bast are relatively free from "amber-colored substance" and the pith cells which contain this substance are diffused, with a peripheral row scattered along the inner margin of the stele. These Tshopo examples are referred to as "Cuviera angolensis" in Prof. Bailey's anatomical studies (Part V, p. 593).

6. —At Kunga, north of Malela, Mr. H. Lang and I found a Cuviera (July 11, 1915; Coll. No. 7983) inhabited by numerous ants, Crematogaster impressiceps variety frontalis Santschi. The myrmecodomatia (Fig. 100) are longer and broader than in the specimens from the Tshopo River and of a reddish green color. The histological structure of the stem is also somewhat different; the "amber-colored substance" is concentrated in the subepidermal and other cortical cells, whereas the

Fig. 100. Cuviera species? a, portion of branch giving external view of one of the domatia and the longitudinal section of another; b, longitudinal section of one of the domatia; e, aperture gnawed by ants; d, pits often occupied by coccids. Drawn from herbarium specimens obtained at Kunga, near Malela; natural size.