tains that A. fistulosa, which furnishes a gum of commercial value, is never eaten by cattle and he attributes this immunity to the protecting ants. Concerning the growth of the thorns he says: "Of the manner in which the young inflations arise I am unable to form any satisfactory conception. They are produced in great numbers at the beginning of the rainy season, when the vegetation awakens, and are then green and soft. I never saw a hole in one of them. They are completely closed on all sides, and it is not till later that they are opened by the ants. I have seen no injuries, wounds, nor anything that could indicate that the deformation is due to insects, and I cannot therefore regard the inflations of the thorns as gall-formations. With this conclusion also harmonizes the fact communicated to me by Schweinfurth, that acacias grown from seed in Cairo also developed the inflations. The only explanation I can suggest is that in this plant an originally abnormal growth has become perfectly normal, under the influence of natural selection, through adaptation to symbiosis with ants." Keller calls attention to the singular fact that only a small number of the thorns on a plant become inflated. This suggests that bacteria or other pathogenic organisms may be responsible for the deformation, which is then put to good use by the ants.

FIG. 180. Broken twig of sunflower (Helianthus annuus) showing ants (Myrmica brevinodis) caught and killed by the exuding sap. (Original.)

A few words may be added on certain South American plants of the genera Cordia, Humboldtia, Ficus (e. g., inæqualis), Tococa (Figs. 170 and 171) and Triplaris (Figs. 166, B and 167) and the East Indian Clerodendron fistulosum. All of these have preformed cavities either in the stems or in bursæ on the leaves and petioles. Clerodendron is said to have in the internodes preformed thin spots, or prostomata, which are selected as entrances by the ants (Colobopsis clerodendri). The leaves of this plant are furnished with innumerable nectaries along the midrib on the lower surface. In Cordia nodosa the flower-bearing stem is dilated above and contains a short, conical cavity which, according to Schimper, is not homologous with the medullary cavity of other plants, as its walls are formed by a fusion of a number of stems. The chamber, which is furnished with a small preformed opening above, is commonly tenanted by ants. Schumann (1888) and Metz (1890) have noted the remarkable fact that

this plant, when growing in the Antilles, fails to develop the hollow swellings in the stems. The hollow twigs of the Polygonaceous Triplaris, or "palo santo," of which some twenty species are known, are said to be invariably occupied by Pseudomyrma. Of these ants Forel (1904) says: "Through the investigations of Mr. Ule the fact becomes more and more firmly established that a definite group of Pseudomyrma species (arboris-sanctæ, dendroica and triplaridis) lives symbiotically in the natural medullary cavities of Tripiaris. In 1896 I myself

FIG. 181. Mound of Formica exsectoides .70 meters high and 2.46 meters in diameter, almost completely covered with a moss (Polytrichum commune) which eventually envelops the summit of the nest and extinguishes the colony. (Original.) observed in Colombia how P. arboris-sancta var. symbiotica fiercely attacked anyone who touched the tree. Their brood filled the whole living tree from the trunk to the smallest green branches. They seemed to have entered this secure and ramifying domicile through a small dead and broken branch on the lower part of the trunk."

No doubt the various cases cited in the preceding pages are of great interest, both to the botanist and myrmecologist, but it is equally certain that none of them has been studied with sufficient care to warrant the conclusions advocated by Belt, Schimper and others. The relationships under discussion are all compatible with the view that the ants have adapted themselves to the plants-plantas itaque norunt formice -but the converse of this proposition is in most, if not in all instances, open to doubt. Travelers and naturalists who observe for a short

time in the tropics, where all of these wonderful cases occur, are very apt to jump to conclusions, and carefully devised experiments, which alone can throw the necessary light on the subject, are still wanting.

The opinion here maintained is indirectly supported by what is known concerning some of the other relations that may obtain between plants and ants. These relations may be considered under the following heads:

1. Ants as Seed Distributors. In the preceding chapter Moggridge's observations on the distribution of seeds by Messor barbarus were mentioned, together with other facts which indicate that ants are important agents in scattering seeds. This habit is not confined to granivorous species. Lubbock (1894) saw Lasius niger carrying violet seeds into its nest. More recently Sernander (1903) and some other botanists have come to believe that the ants eat the caruncles and that these structures are developed as lures, like the extrafloral nectaries and food-bodies, to induce the ants to carry the seeds to a distance and thus increase the chances of their survival. Dr. E. B. Southwick tells me that he has seen the ants in Central Park, N. Y., carry away the seeds of the blood-root (Sanguinaria canadensis) and feed on their caruncles.

2. Ant-gardens.-This name is given by Ule (1902) to certain. sponge-like ant-nests (Fig. 179) which he found built on the branches. of trees in the forests of the Amazon. These nests consist of soil carried up by the ants (Azteca olithrix, ulei and traili and Camponotus femoratus) and held together by the roots of numerous epiphytes, which grow out of it on all sides, making it resemble the head of a Medusa. The ants not only perforate the soil with their galleries but, according to Ule, actually plant the epiphytes. This he infers from seeing the insects in the act of carrying the seeds. Perhaps these are brought into the nest for the sake of their caruncles and then germinate in the rich soil, but it is quite as probable that they are sown by the wind.

3. Plants Injurious to Ants.-If it be true that some plants deserve to be called "myrmecophilous," because they are helpful to ants in the struggle for existence, it is equally true that there are other plants that might with even greater justice be called "myrmecophobic," or "myrmecechthric," because they are injurious or even deadly to these insects. Such are, for example, certain moulds and bacteria. Queen ants while founding their colonies in damp cavities in soil or decaying wood often succumb to the incursions of these organisms, which under certain conditions may even exterminate the brood of larger colonies. Miss Fielde (1901b) says: "Penicillium crustaceum grows to ripeness, in

either darkness or light, upon eggs, larvæ or pupæ, if left for a few days unattended in the humid atmosphere required by the ants, and its sprouting spores may be seen on their surfaces under a magnification of about five hundred diameters. If the spores are left undisturbed they cover the young with a delicate dense white coat that becomes sage-green with the ripening of the new spores. . . . This delicate mould does not grow upon the bodies of dead ants, but is there replaced by Rhyzopus nigricans, with long and spreading hyphæ, and in this

may lie the cause for the carrying off and casting away of all ants that die or are killed in the nest."

Botanists have described several peculiar arrangements in higher plants, such as excessive hairness, slipperiness or stickiness of the stems, or special palisades of hairs about the floral nectaries (nectarostegia ) as means of preventing ants and other desultory arthropods from plundering the secretions intended for bees and other cross-fertilizing agents. But these arrangements, if really developed for this purpose, are often inefficacious. Vosseler (1906) has recently described an African ant which manages to get around the woolly hairs protecting the nectaries

in Cobea scandens and gains access to these organs by biting a hole through the base of the petal, a habit which has also been observed in bees that are confronted with flowers whose nectaries they are unable to reach in any other way. But there are graver maladjustments between plants and ants, maladjustments that may lead to the death of the insects in great numbers or even to the extinction of their colonies. I have described a number of such cases in a recent paper (1906). The abundant and sticky juices of Silene, Lactuca and Helianthus exuding onto the stems or petioles often entrap and kill numbers of ants (Fig. 180). We owe to a similar property of the resiniferous conifers of the Tertiary the preservation of the ants in the Baltic and Sicilian ambers. Our North American pitcher plants (Sarracenia) also entrap, kill and digest enormous numbers of ants in the liquid at the bottoms of their ascidia. The ants most frequently found in these modified leaves are Cremastogaster pilosa, a species which will sometimes even nest in the dead pitchers of a plant whose active green leaves are busy killing them off in great numbers-a singular commentary on the "intelligence" of these insects, especially when we stop to consider that C. pilosa is one of the ants that constructs such beautiful sheds over aphids and coccids.

Another hostile relationship between plants and ants has been described in detail by Holmgren (1904). He observed that the mound. nests of Formica exsecta in the bogs of Lapland are gradually invaded and eventually so completely covered with a dense carpet of moss (Polytrichum strictum) that the ants are either driven away or destroyed. This moss is in turn replaced by a carpet of Sphagnum, in which many plants eventually take root, so that the ants are instrumental in forming the hummocks of moss and hence facilitate the growth of peat-forming vegetation. In the bogs of Prussia, according to Kuhlgatz (1902), the Myrmica nests are invaded in a similar manner by P. strictum, and I have been able to observe various stages in the extinction of colonies of our North American F. exsectoides by an allied moss, P. commune (1906). This moss starts in the form of a narrow zone around the base of the huge mound nests (Fig. 181) and gradually grows upward till it completely envelopes their summits. with a dense mat and either smothers the colony outright or compels it to emigrate (Fig. 182).