occidentalis). I sowed the seeds of the acacia in my garden, and reared some young plants. Ants of many kinds were numerous; but none of them took to the thorns for shelter, nor the glands and fruit-like bodies for food; for, as I have already mentioned, the species that attend on the thorns are not found in the forest. The leaf-cutting ants attacked the young plants, and defoliated them, but I have never seen any of the trees out on the savannahs that are guarded by the Pseudomyrma touched by them, and have no doubt the acacia is protected from them by its little warriors. The thorns, when they are first developed, are soft, and filled with a sweetish, pulpy substance; so that the ant, when it makes an entrance into them, finds its new house full of food. It hollows this out, leaving only the hardened shell of the thorn. Strange to say, this treatment seems to favour the development of the thorn, as it increases in size, bulging out towards the base; whilst in my plants that were not touched by the ants, the thorns turned yellow and dried up into dead but persistent prickles. I am not sure, however, that this may not have been due to the habitat of the plant not suiting it.

These ants seem at first sight to lead the happiest of existences. Protected by their stings, they fear no foe. Habitations full of food are provided for them to commence housekeeping with, and cups of nectar and luscious fruits await them every day. But there is a reverse to the picture. In the dry season on the plains, the acacias cease to grow. No young leaves are produced, and the old glands do not secrete honey. Then want and hunger overtake the ants that have revelled

in luxury all the wet season; many of the thorns are depopulated, and only a few ants live through the season of scarcity. As soon, however, as the first rains set in, the trees throw out numerous vigorous shoots, and the ants multiply again with astonishing rapidity.

Both in Brazil and Nicaragua I paid much attention to the relation between the presence of honey-secreting glands on plants, and the protection the latter secured by the attendance of ants attracted by the honey. I found many plants so protected; the glands being specially developed on the young leaves, and on the sepals of the flowers. Besides the bull's-horn acacias, I, however, only met with two other genera of plants that furnished the ants with houses, namely the Cecropia and some of the Melastomæ. I have no doubt that there are many others. The stem of the Cecropia, or trumpet-tree, is hollow, and divided into cells by partitions that extend across the interior of the hollow trunk. The ants gain access by making a hole from the outside, and then burrow through the partitions, thus getting the run of the whole stem. They do not obtain their food directly from the tree, but keep brown scale-insects (Coccida) in the cells, which suck the juices from the tree, and secrete a honey-like fluid that exudes from a pore on the back, and is lapped up by the ants. In one cell eggs will be found, in another grubs, and in a third pupæ, all lying loosely. In another cell, by itself, a queen ant will be found, surrounded by walls made of a brown waxy-looking substance, along with about a dozen Coccide to supply her with food. I suppose the eggs are removed as soon as laid,

for I never found any along with the queen-ant. If the tree be shaken, the ants rush out in myriads, and search about for the molester. This case is not like the last one, where the tree has provided food and shelter for the ants, but rather one where the ant has taken possession of the tree, and brought with it the Coccide; but I believe that its presence must be beneficial. I have cut into some dozens of the Cecropia trees,

LEAF OF MELASTOMA.

し

and never could find one that was not tenanted by ants. I noticed three different species, all, as far as I know, confined to the Cecropic, and all farming scale-insects. As in the bull's-horn thorn, there is never more than one species of ant on the same tree.

In some species of Melastoma there is a direct provision of houses for the ants. In each leaf, at the base of the lamina, the petiole, or stalk, is furnished with a

couple of pouches, divided from each other by the midrib, as shown in the figure. Into each of these pouches there is an entrance from the lower side of the leaf. I noticed them first in Northern Brazil, in the province of Maranham; and afterwards at Pará. Every pouch was occupied by a nest of small black ants, and if the leaf was shaken ever so little, they would rush out and scour all over it in search of the aggressor. I must have tested some hundreds of leaves, and never shook one without the ants coming out, excepting on one sickly-looking plant at Pará. In many of the pouches I noticed the eggs and young ants, and in some I saw a few darkcoloured Coccidæ or aphides; but my attention had not been at that time directed to the latter as supplying the ants with food, and I did not examine a sufficient number of pouches to determine whether they were constant occupants of the nests or not. My subsequent experience with the Cecropia trees would lead me to expect that they were. If so, we have an instance of two insects and a plant living together, and all benefiting by the companionship. The leaves of the plant are guarded by the ants, the ants are provided with houses by the plant, and food by the Coccide or aphides, and the latter are effectually protected by the ants in their common habitation.

Amongst the numerous plants that do not provide houses, but attract ants to their leaves and flower-buds by means of glands secreting a honey-like liquid, are many epiphytal orchids, and I think all the species of Passiflora. I had the common red passion-flower growing over the front of my verandah, where it was continually under my notice. It had honey-secreting glands

on its young leaves and on the sepals of the flower-buds. For two years I noticed that the glands were constantly attended by a small ant (Pheidole), and, night and day, every young leaf and every flower-bud had a few on them. They did not sting, but attacked and bit my finger when I touched the plant. I have no doubt that the primary object of these honey-glands is to attract the ants, and keep them about the most tender and vulnerable parts of the plant, to prevent them being injured; and I further believe that one of the principal enemies that they serve to guard against in tropical America is the leaf-cutting ant, as I have observed that the latter are very much afraid of the small black ants.

On the third year after I had noticed the attendance of the ants on my passion-flower, I found that the glands were not so well looked after as before, and soon discovered that a number of scale-insects had established themselves on the stems, and that the ants had in a great measure transferred their attentions to them. An ant would stand over a scale-insect and stroke it alternately on each side with its antennæ, whereupon every now and then a clear drop of honey would exude from a pore on the back of the latter and be imbibed by the ant. Here it was clear that the scale-insect was competing successfully with the leaves and sepals for the attendance and protection of the ants, and was successful either through the fluid it furnished being more attractive or more abundant.* I have, from these facts, been led to the conclusion that the use of honey-secreting glands in plants is to attract insects that will protect

* I have since observed ants attending scale-insects on a large plant of Passiflora macrocarpa in the palm-house at Kew.

P