and of two species of Holcomyrmex (H. lameerei and chobauti) peculiar to the sandy and extremely barren portions of the desert. H. chobauti resembles M. caviceps in having a pronounced cavity on the under side of the head. Of the former species he says: "I saw the long files of workers carrying seeds of the 'drîn' (Aristida pungens) to their nests. The seed of this plant has the form of a slender spindle surmounted by a long, trifid and plumose spine. The ant rides. this grain as a witch rides her broom; she carries it beneath her, holding it firmly by the small end in her mandibles with the end of the grain fitting into the notch under her head. This interesting character, which this ant shares with M. caviceps, may be regarded as an adapta

tion to the method of carrying the drin seed. It should be noted that I found M. caviceps in a region of the Eolian desert where the drin is almost the only plant that can subsist. On the other hand there is no drin in the region of Hamada where I first found H. lameerei, which has the under side of the head but little excavated."

Still another interesting ant of the North African desert has been recently discovered in Tunis by Santschi. This is a small black species, Oxyopomyrmex santschii. Its habits are described as follows by its discoverer in a letter published by Forel (1904a): The nests are So characteristic that when one has once seen one of them, nothing is

easier than to find others. I am surprised to learn that they have not attracted the attention of other observers. Especially remarkable is the tiny crater, which has the form of a cone [Fig. 151] hardly more than 4-5 cm. in diameter and 2.5-3 cm. high. The circumference of its funnel-shaped top is 3-4 cm. across and its margin is always perfectly circular and entire, except in nests in process of construction,

when it is at first semilunar like the very small nests of Messor arenarius. At the bottom of the funnel the small entrance is found at a depth of 2-3 cm. It is horizontal, attaining a length of 5 cm., a breadth of 1 cm. and a height of 5 cm. In this first chamber the pupæ are kept for the purpose of enjoying the warmth, and here I have found a number of workers and winged females. Thence the gallery continues to descend to a depth of 15-20 cm. and finally opens into two or three chambers of the same dimensions as the first. These contain pupæ and an ample provision of very small seeds. This ant is therefore granivorous. I surprised a few of the workers entering the nest with seeds in their mandibles, but they go out foraging singly and not in files like Messor and other genera. They are very slow in their move

ments and are very apt to stop motionless at the least alarm. Day or night one or two of the workers may be seen on the outer surface of the crater, scarcely moving unless molested, but when disturbed they hurriedly retreat into the nest to spread the alarm. Their habits are rather nocturnal. If a light is brought near the nest when a worker is on the point of leaving it with a grain of sand, she hurriedly backs into the entrance and there stops, closing it perfectly with her burden. If the observer remains very quiet, she eventually comes forth and deposits her load on the slope of the crater. There are scarcely more than thirty individuals in a nest. I have found this species only in a very circumscribed area, south of Kairouan, on compact, sandy soil in

FIG. 157. Male, virgin female, and worker of the Texan harvester, Pogonomyrmex molefaciens, nearly twice natural size. (Original.)

which the chambers are easily excavated." I quote this description at length and reproduce Santschi's figure on account of the remarkable resemblance of the Oxyopomyrmex nest to those of the fungus-growing Trachymyrmex to be described in the next chapter. The genus Oryopomyrmex is represented by several species, some of which have been placed in a subgenus, Goniomma. One of these, G. hispanica of southwestern Europe, is also a harvesting ant, according to Ern. André (1881e).

To the old observations of Sykes and Jerdon on harvesting ants in India, Wroughton (1892) has added accounts of the habits of Holcomyrmex scabriceps (Fig. 150) and Pheidologeton ocellifer. Of the former he says: "In a community of this genus there are workers of all sizes. Holcomyrmex is, as a rule, a most industrious harvester, and

sets about her work in a most methodical way. The workers never forage individually for grain, but all take the same road and all return by the same road; the result being that every nest is the starting point of one, or often of several, well-beaten tracks, cleared of vegetation and obstacles, and extending sometimes 100 feet and more in length. How these tracks are engineered I have never discovered, but am pretty certain that they are made gradually; a commencement at hazard is made, and then, as the country immediately adjoining the road is exploited, the road itself is carried forward. Where one of these roads crosses a sheet of bare rock, it is there marked in white; I can only presume that this is the result of some chemical action, set up by the formic acid exuding from the ants; this acid, though too small in quantity in a single ant to cause any appreciable effect, might easily become sufficient when thousands of ants are continually passing, backwards and forwards, all day long. Holcomyrmer brings home the grain unthreshed, and, in this form, it is taken into the nest, from whence the chaff is brought out and deposited around the entrance, or, where the force of a prevalent wind is felt, on a heap to leeward." Wroughton does not believe that H. scabriceps, which Rothney regards as the harvester par excellence of India, compares with Pheidole as a harvester.

The following note on Pheidologeton ocellifer, an ant with highly polymorphic workers, was communicated to Wroughton by Aitken: The entrance (of the nest) which is strewn with chaff, is large, but the passage soon splits up, and I failed to follow it. I turned up a lot of pupa, however, close to the surface. The community is enormous and industrious, collecting large seeds of trees or plants, which it takes a dozen to carry; these are taken in and the husks are thrown out afterwards. If P. ocellifer meets a white ant or any other insect, she collects it in the same way. The smaller soldiers often laid a jaw to a burden, but the giants appear to do nothing." Wroughton confirms this observation on the carnivorous tastes of Pheidologeton. He found also that the huge soldiers neither dig nor defend the nest and that they are less pugnacious than the smaller workers. It is probable that they function as seed crushers like the soldiers of the allied genus Pheidole.

Armit (1878), Roth (1885) and Tryon (1900) have published a few observations on harvesting ants in another arid or semiarid region, Australia. These are Pheidole longiceps, Merano plus dimidiatus and M. diversus. Tryon calls attention to the seed-distributing habits of the Pheidole in the following passages: “That this Brisbane harvesting ant, also, is an important agent in the local dispersion of plants-especially weeds and is connected with their sudden appearance on heaps of soil excavated from a depth, is sufficiently demonstrated in the fol

lowing observations: The ants of one nest were noticed to be harvesting the seeds of Portulacca oleracea Linn. and of Amaranthus viridis Linn. both common weeds-and growing at a comparative distance from the nest. These seeds had remained stored up in their nest for some time, when rain suddenly came on, and under its influence the seeds-especially those of the latter plant-commenced to germinate. Of those which had already thrown out a radical, this was bitten off and brought to the surface; some of these seeds were also gnawn into, and the ruptured black perisperm-containing more or less food substance-in like manner rejected. Other seeds, which had swollen in response to moisture, were carried up for the purpose of being dried and re-stored. In the midst of these operations, however, rain came on again, and the ants retired, leaving seeds on the surface. These immediately germinated, and a small patch of Amaranthus grew up, making the site of what was before a nest of harvesting ants, quite isolated among plants of different character. On a second occasion

a nest, in which much seed of Eleusine indica was known to have been harvested some months since, was dug up. Some of the grass seed selected from the nest was afterwards sown; also some of the earth from the nest which was known to contain both seeds of this plant and of another species of Amaranthus. In both cases the sowings were made in situations remote from such places in which any of these plants were already growing, and, as a result, in the course of time, numerous