abortive attempts at penetration made by the ants, the resistance of the tissues there having caused them to abandon the effort, which, however, easily succeeds at the upper part of the internode, where, from an early age, the medullary tissue is less dense and the peripheral tissues less resistant. Perhaps the irritation produced by the ants may cause an increase in the diameter of the internodes and of their cavity. The ants doubtless render to the Kibaras services of various kinds; first, protection against plant eaters; then transportation of the cochineals to a location where they will be less injurious than upon the young and undeveloped parts; finally, fertilization of the flowers. The Kibaras are, in fact, monoecious, and their floral structure is such that their fertilization seems impossible without the intervention of insects. In exchange for these services they offer these ants a cavity for lodgment, and very likely an aliment indirectly furnished by the cochineals. The Cecropia adenopus is a plant of Brazil, belonging to the Araliaceæ, that in its native country bears the name of Amboiba or Imbauba. As long ago as 1648 Marcgrave said of it "Totus intus cavus a radice ad summum usque et cavitas illa per interstitia semi-digiti ubique distincta et transversali membrana, in cujus medio foramen rotundum magnitudine pisi. In hac cavitate reperiuntur semper formice rubræ ipsa coloris et hepatici." The medullary tissue is narrow at the base of the trunk, enlarging above, and is interrupted at each node by a ligneous disk. Two consecutive disks thus bound a closed cavity corresponding to an internode. In these cells the ants pursue the culture of cochineals. This is a myrmecophilous feature similar to that which Beccari pointed out in Kibara formicarum and hospitans. Belt and Fritz Müller have studied the relations of this plant to ants. According to the latter author there is a small cavity in the upper part of each internode, where the wall of the internodal chamber is thinner. At this point a pregnant female ant makes a hole in order to penetrate the chamber. These perforations can be plainly seen in herbarium speci mens. Their relative position is perfectly regular, and they can by no means be considered as accidental. The ants once installed in the chambers, perforate the nodal disks and may thus circulate, under shelter, throughout the entire length of the trunk (Belt). Three species of ants frequent the Cecropia, but if either of these species is present the others are not found upon the tree. Fritz Müller thinks, on the contrary, that perforation of the disks does not take place. These two opinions, apparently contradictory, may perhaps be easily explained. Possibly the various species of ants that frequent the Cecropia may not have identical habits. Some may leave the disks intact, others may perforate them. Each of these observers may have been dealing with a different species of insect. Perhaps, also, they did not observe the same species of Cecropia. The Cecropia adenopus may not be the only one that presents a hollow stem separated into chambers by nodal disks. The same features are found in most of the species of that genus, which are, without doubt, myrmecophilous. The utility of the ants for the Cecropia they inhabit may consist entirely in the protection they afford. This protection would be not only against plant-eating animals, but also against cochineals. We may suppose that the latter are transported from the surface of young buds, where they would be injurious to the normal development of the leaves, to the cavities of the stem, where their injurious action would be less. The ants here act like our gardeners who free hothouse plants from infesting scale insects. But the honey dew of cochineals being endowed with nutritive properties, the ants do not destroy them, but merely transport them to a part of the plant where their life is more compatible with the normal evolution of the vegetable. There would thus be established a consortium of three members-between the plant on one hand and the cochineals and the ants on the other. Upon the Cordia Gerascanthos we find enlargements of the branches that are terminated by axes of inflorescence. Into these enlargements the ants make openings, using, perhaps, the place where some little bud is implanted. The cavity of the enlargment at first contains a flocculent tissue that the ants remove so that they may arrange within the cavity disks like a sort of pasteboard. In this domicile the ants pursue the raising of cochineals. It should be noted that these enlargements do not appear to be constant in the species. In another species of the same genus, Cordia nodosa, the internodes, especially those bearing the inflorescence, are enlarged and hollowed near the insertion of the opposite leaves. The cavity communicates with the outside by an orifice situated, not laterally, as in the preceding species, but at its top. Both cavity and opening seem to be natural and not affected by the agency of ants. In cavities not yet visited by ants the internal surface is invested with stiff scattered ridges, some of which hang over the opening. In this species the lodging organ is formed hereditarily all ready for occupancy by the ants without any preliminary labor on their part. The myrmecophilous features merely outlined in the first species of Cordia would thus attain their perfection in the second and their origin be purely hereditary. This is an excellent example of the fixation of a character primitively accidental, and, so to speak, teratological. If the ants vary the place of penetrating the lodging cavity, the opening they make will not tend to become hereditary-that is to say, to reproduce itself independently of their This is the case, for example, in Acacia cornigera and the Endospermums. But if the ants always make their opening at the same point, the lesion tends to become a part of the morphologic plan of the vegetable. The point of lesion in the ancestor becomes a point of less resistance in the descendant-that is to say, a point where the ants can make an opening with the greatest facility, as the wall of the lodging organ would there be thin and easily perforable. In those action. types in which myrmecophilous evolution is most advanced the perforation will become hereditary and the cavity of the lodging organ communicate with the exterior independently of any action on the part of the ants (Clerodendon fistulosum). In all the cases we have reviewed the ants do not seem to have estab lished themselves in organs whose differentiation relates to the habitat of the host plant. A considerable number of host plants are epiphytes, subject to peculiar physical conditions. They have to especially struggle against drought, and a number of them possess in their organs true reservoirs of water. The best known of these myrmecophilous epiphytic plants are the Myrmecodias and the Hydnophytums. We will dwell especially on the former because of the interesting observations that have been made upon them. The Myrmecodias and Hydnophytums are epiphytes belonging to the Rubiaceae and attach themselves by means of adventitious roots to the branches of trees, often at a considerable height. These plants are almost wholly formed of large tubercles, globular or cylindric in form, surmounted by one or more leafy stems. These tubercles, which may be either smooth or prickly, enlarge so as to attain several decimeters in diameter. (Pl. XX.) Instead of forming a solid mass their internal tissue is traversed by a system of intercommunicating cavities and passages that open externally by one or more quite large openings and numerous narrow orifices scattered over the entire surface of the tubercle. * All those who have collected these strange plants in their natural habitat have found their tubercles inhabited by ants scattered in great numbers throughout the galleries and passages. It was Rumphius, the old explorer of the Malay Archipelago, who first called attention to this. According to him the ants not only inhabit the tubercles, but they produce the entire vegetable. "This is," he says, "a strange creation of nature springing up without father or mother, for it is known that these plants spring from the substance of the nests of ants where there can never have been any seeds, and yet each colony forms a separate plant." Rumphius then distinguishes two kinds of Nidus germinans, according to the species of ants found therein: Nidus germinans formicarum rubrarum—that is, a Myrmecodia, and Nidus germinans formicarum nigrarum—that is, a Hydnophytum. It was Beccari, the eminent explorer of Malasia, who made the first accurate observations upon the biology of these curious Rubiaceæ and their relations to ants. These observations led him to suppose that the presence of the insects was indespensable to the plant. He thought he had ascertained that at the time of germination the tigellus merely thickens a little at the base and takes on a conical form with the cotyledons opening at the summit (Fig. 2, Nos. 1-5, Pl. XX), thus remaining until a species of ant hollows a little cavity in its side at the most swollen part of the tigellus. If the tigellus is not attacked by the ant |