the nests of the ants merely from hearsay. Later various, more or less erroneous, observations on the var. horti-deorum were made at Santa Fe, N. M., by Captain W. B. Fleeson (Edwards, 1873), Saunders (1875) and Loew (1874). The first to publish a trustworthy account of this, or in fact of any of our Myrmecocysti, was McCook (1882b). He discovered horti-deorum in the Garden of the Gods, near Manitou, Colo. (Fig. 215). The nests, which were found on the tops of stony ridges, are described in great detail. The large circular entrance, 2-2.5 cm. in diameter, is in the center of a cone-shaped crater of small pebbles 8-25 cm. in diameter at the base and 5-8 cm. high (Fig. 216). The entrance opens into a vertical or oblique gallery which at a depth of 9-15 cm. breaks up into several smaller galleries. These usually

run to one side of the entrance gallery. At a depth of 20-35 cm. the smaller galleries lead into chambers with smooth, flattened floors and rough, vaulted ceilings. These chambers vary from 12-15 cm. in length and 7-10 cm. in width, and may be 4 cm. high in the middle. McCook has described some very large formicaries, the galleries of one of which extended over an area of more than 2 m. and reached a depth in the soil of more than a meter. From the ceilings of the chambers the repletes hang, side by side, by means of their claws, and the vaulting is evidently left rough as an adaptation to this peculiar habit (Fig. 220). The repletes are capable of more movement than has usually been supposed, but if they fall from the ceiling they are unable to regain their pendent position without assistance. Large nests may contain as many as three hundred repletes distributed over several chambers.

McCook effectively dispelled the notion that the repletes manufacture the honey which they contain, a notion started by Wesmael and held by several writers, by showing how they obtain and store the liquid. M. horti-deorum is decidedly nocturnal, unlike the different subspecies and varieties of melliger, which are diurnal. Indeed, the etiolated appearance and pale yellow color of the northern forms of mexicanus at once suggest a fondness for darkness, just as the deeper tints of the typical form of the species suggest diurnal or crepuscular habits. During the day, therefore, the workers of M. horti-deorum are never seen outside of the nest, but frequently a guard of workers is stationed just within the large opening, apparently for the purpose of preventing other ants, spiders, etc., from entering. McCook found that during July the workers leave the nest in a file at about 7.30 P. M. and visit the shin oaks (Quercus undulata) which grow abundantly along the rocky ridges in the Garden of the Gods and the surrounding country. The twigs of these oaks are often covered with small woody galls about the size of a pea and of a more or less conical or spheroidal shape, the work of the Cynipid Holcaspis perniciosus. At night these galls exude minute droplets of a sweet, watery secretion which is eagerly imbibed by the ordinary workers, carried to the nest in their crops and fed to the repletes (Fig. 222).

Forel, in 1880, showed that the gaster of the replete horti-dcorum owes its size and rotundity exclusively to an enormous distension of the crop, or ingluvies and not of the stomach as Leidy (1852) and Blake (1873) had supposed, and that all the other structures found in the gaster of the ordinary worker are present in the replete, though they are necessarily flattened against the gastric wall. These observations were confirmed by McCook's careful dissections and figures of the gaster of ordinary workers, semirepletes ("semirotunds") and repletes. He inferred that "the process by which the rotundity of the honey-bearers has probably been produced, has its exact counterpart in the ordinary distension of the crop in overfed ants; that, at least the condition of the alimentary canal, in all the castes, is the same, differing only in degree, and therefore, the probability is very great that the honey-bearer is simply a worker with an overgrown abdomen." He found, moreover, that "a comparison of the workers with the honey-bearer shows that there is absolutely no difference between them except in the distended condition of the abdomen" and he therefore inferred "that the worker majors, for the most part, and sometimes the minors, are transformed by the gradual distension of the crop, and expansion of the abdomen, in the honey-bearers, and that the latter do not compose a distinct caste. It is probable, however,

that some of the majors have a special tendency to this change by reason of some peculiar structure or form of the intestine and abdominal walls." Although McCook gave these excellent reasons for believing that the replete must develop from a worker of the ordinary type, he did not actually witness the transformation. His account covers a great many other details in the behavior of M. horti-deorum, but as many of these are common to most other Camponotine ants, they need not be discussed in this connection.

My own observations on M. horti-deorum were made during July and August, 1903, and 1906. At first I worked in the Garden of the Gods and located several nests on the ridges where McCook made his classical observations many years ago. But the region is now so overrun by tourists that it is no longer a favorable spot for the quiet study of ant colonies. I therefore sought new localities and was soon able to locate several fine nests south of the Fontaine-quiBouille along Bear Creek and Red Rock Cañons. A few nests were also found west of Manitou and south of the Ute Pass at a much greater distance from the Garden of the Gods. In all of these localities there are thickets of shin oaks (Quercus undulata and gambeli), and the nests are situated only on the summits of dry, stony ridges, just as they are in McCook's locality.

During the two summers I excavated fully a dozen fine colonies of horti-deorum and was able to confirm in nearly every detail McCook's interesting account of the nest architecture and the habits of the ordinary and replete workers. I was unable to make observations at night, but have no doubt that McCook's account of the foraging habits is perfectly trustworthy. I am inclined to believe, however, that the exudations of the Holcaspis galls may furnish only a portion of the food of the ants and that these insects obtain much, if not most, of their honey from the coccids and aphids on the oaks and other plants in the neighborhood. It would be strange, indeed, if these ants did not take advantage of a food supply so much more copious than that furnished by the galls.

I have been able to prove, what has been surmised by McCook and his predecessors, namely, that workers of the ordinary size and form

develop into repletes. In the nests which I excavated during July there were many callow workers, males and females. While keeping several colonies in artificial nests it occurred to me that the change from the ordinary to the replete worker must begin during the callow stage, while the integument of the gaster is still very soft and distensible. I accordingly isolated a number of young callows in two of my nests and fed them with maple syrup and cane sugar water. They partook of these substances greedily, and a few of the workers in each nest gradually began to assume the replete condition. During the course of four to six weeks several of them became what I have called semirepletes (McCook's semirotunds), and four, three in one nest and one in the other, actually attained the dimensions of the perfect replete. Most of the workers, however, showed no inclination to assume this form. In most cases, as McCook has shown, it is the major workers which most readily become repletes, but this is not an invariable rule. In the honey chambers of opulent colonies I have usually found also a few replete mediæ and minimæ hanging among their larger but no more turgid sisters. Thoroughly hardened workers of the ordinary form, according to my observations, are no longer able to become repletes. It is probable that McCook's failure to secure these from isolated major workers was due to his using old individuals in his experiment.

Why certain callows should aspire to become animated pots or tuns, while others prefer to be active foragers and providers, is an enigma. I do not believe, however, that this is due to differences in the "structure or form of the intestine and abdominal walls," as McCook suggests. It is more probably an unusual example of the division of labor, which is shown by careful study to exist in various forms and degrees among all ants with monomorphic workers. The individual worker performs different duties at different stages in its life, beginning in its callow stage as a mere nurse, then becoming a forager, warrior or guard, and in its old age sometimes encroaching on the function of the queen and becoming a parthenogenetic mother, or gynæcoid. It is not improbable that many worker ants acquire habits-using this word, for the moment, in its restricted and technical sense as employed in human psychology—and tend to perform throughout life the particular function which they happen to assume while in the callow stage. This may account for the development of the passive replete, not only in M. hortideorum, but also in all other honey ants. Those who, in anthropomorphic mood, are wont to extoll the fervid industry and extraordinary feats of muscular endurance in ants, should not overlook the beatific patience and self-sacrifice displayed by the replete Myrmecocystus as

it hangs from the rafters of its nest, month in, month out-for years, perhaps a reservoir of temperamental as well as liquid sweetness.

6. Cremastogaster inflata and difformis.-Frederick Smith (1857, 1858c) published an account of two Myrmicine ants, which has led to

FIG. 219.

(Original.)

their being regarded as honey ants. Both species range from Tenasserim to Borneo through Burma, Java and Sumatra. C. inflata occurs also in the Philippines and difformis in Celebes. Of the latter species