entirely on the tree they inhabit, the food being usually sweet stuff secreted by glands of the plant. It is thought that the ants in return are of considerable benefit to the plant by defending it from various small enemies, and this kind of symbiosis has received much attention from naturalists. Α very curious condition exists in the epiphytic plants of the genera Myrmecodia and Hydnophytum; these plants form large bulb-like (Fig. 59) excrescences which, when cut into, are found to be divided into chambers quite similar to those frequently made by ants. Though these structures are usually actually inhabited by ants, it appears that they are really produced by the plant independent of the Insects.

Variability and Polymorphism of Ants.-Throughout the Hymenoptera there are scattered cases in which one of the sexes appears in dimorphic form. In the social kinds of bees and wasps the female sex exists in two conditions, a reproductive one called queen, and an infertile one called worker, the limits between the two forms seeming in some cases (honey-bee) to be absolute as regards certain structures. This sharp distinction in structure is rare; while as regards fertility intermediate conditions are numerous, and may indeed be induced by changing the social state of a community. In ants the phenomena of the kind we are alluding to are very much more complex. There are no solitary ants; associations are the rule (we shall see there are one or two cases in which the association is with individuals of other species). In correlation with great proclivity to socialism we find an extraordinary increase in the variety of the forms of which species are made up. In addition to the male and female individuals of which the species of Insects usually consist, there are in ants workers of various kinds, and soldiers, all of which are modified infertile females. But in addition to the existence of these castes of infertile females, we find also numerous cases of variability or of dimorphism of the sexual individuals; and this in both sexes, though more usually in the female. Thus there exists in ants an extraordinary variety in the polymorphism of forms, as shown by the table on p. 141, where several very peculiar conditions are recorded.

The complex nature of these phenomena has only recently

1 The parthenogenetic young produced by worker females are invariably of the male sex.

The

become known, and as yet has been but little inquired into. difference between the thoracic structure in the case of the winged and wingless females of certain species (Fig. 55, and in vol. v. fig. 339) is enormous, but in other species this difference appears to be much less. The ordinary distinctions between the queenfemale and worker-females appear to be of two kinds; firstly, that the former is winged, the latter wingless; and secondly, that the former possesses a receptaculum seminis, the latter does not. In a few cases it would seem that the dimorphism of winged and wingless forms is not complete, but that variability exists. Intermediate conditions between the winged and wingless forms are necessarily rare; nevertheless a certain number have already been detected, and specimens of Lasius alienus have been found with short wings. In rather numerous species some or all of the fertile females depart from the usual state and have no wings; (a similar condition is seen, it will be recollected, in Mutillides and Thynnides of the neighbouring family Scoliidae). A dimorphism as regards wings also exists in the male sex, though it is only extremely rarely in ants that the males are wingless. Neverless a few species exist of which only wingless males have been found, and a few others in which both winged and wingless individuals of this sex are known to occur. The wingless males of course approach the ordinary workers (= infertile wingless females) in appearance, but there is not at present any reason for supposing that they show any diminution in their male sexual characters. The distinction between workers and females as based on the existence or non-existence of a receptaculum seminis has only recently become known, and its importance cannot yet be estimated. The adult, sexually capable, though wingless forms, are called ergatoid, because they are similar to workers ('Epyarns, a worker).

1 The student must recollect that the winged female ants cast their wings previously to assuming the social life. The winglessness of these females is a totally different phenomenon from that we here allude to.

2 See Forel, Verh. Ges. deutsch. Naturf. lxvi. 1894, 2, pp. 142-147; and Emery Biol. Centralbl. xiv. 1894, p. 53. The term ergatoid applies to both sexes; a species with worker-like female is ergatogynous; with a worker-like male ergatandrous.

Table of the Chief Forms of Polymorphism in Ants.

In addition to the above there are apparently cases of females with post-metamorphic growth in Dorylides, but these have not yet been the subject of investigation.

Much has been written about the mode in which the variety of forms of a single species of ant is produced. As to this there exists but little actual observation or experiment, and the subject has been much complicated by the anxiety of the writers to display the facts in a manner that will support some general theory. Dewitz was of opinion that workers and queens of ants were produced from different kinds of eggs. This view finds but little support among recent writers. Hart in recording the results of his observations on the parasol ant (of the genus Atta) -one of the species in which polymorphism is greatest-says that these observations prove that "ants can manufacture at will, male, female, soldier, worker or nurse," but he has not determined the method of production, and he doubts it being the character of the food." There is, however, a considerable body of evidence suggesting that the quality or quantity of the food, or both combined, are important factors in the treatment by which the differences are produced. The fact that the social Insects in which the phenomena of caste or polymorphism occur, though belonging to very diverse groups, all feed their young, is of itself very suggestive. When we add to this the fact that in ants, where the phenomena of polymorphism reach their highest complexity, the food is elaborated in their own organs by the feeders that administer it, it appears probable that the means of producing the diversity may be found herein. Wasmann has pointed out that the ants'-nest beetle, Lomechusa, takes much food from the ants, and itself destroys their young, and that in nests where Lomechusa is abundant a large percentage of ergatogynous forms of the ants are produced. He attributes this to the fact that the destruction of the larvae of the ant by the beetle brings into play the instinct of the ants, which seek to atone for the destruction by endeavouring to produce an increased number of fertile forms; many ergatogynous individuals being the result. This may or may not be the case, but it is clear that the ants' instinct cannot operate without some material means, and his observation adds to the probability that this means is the food supply, modified either qualitatively or quantitatively.

The existence of these polymorphic forms led Herbert Spencer to argue that the form of an animal is not absolutely

1 Nature li. 1894, p. 125.

determined by those "Anlagen" or rudiments that Weismann and his school consider to be all important in determining the nature or form of the individual, for if this were the case, how can it be, he asked, that one egg may produce either a worker, nurse, soldier or female ant? To this Wasmann (who continued the discussion) replied by postulating the existence of double, triple or numerous rudiments in each egg, the treatment the egg receives merely determining which of these rudiments shall undergo development.1 Forel seems to have adopted this explanation as being the most simple. The probability of Weismann's hypothesis being correct is much diminished by the fact that the limit between the castes is by no means absolute. In many species intermediate forms are common, and even in those in which the castes are believed to be quite distinct, intermediate forms occur as very rare exceptions.2 Emery accounts for the polymorphism, without the assistance of the Weismannian hypothetical compound rudiments, by another set of assumptions; viz. that the phenomenon has been gradually acquired by numerous species, and that we see it in various stages of development; also that variation in nutrition does not affect all the parts of the body equally, but may be such as to carry on the development of certain portions of the organisation while that of other parts is arrested. Speaking broadly we may accept this view as consistent with what we know to be the case in other Insects, and with the phenomena of post-embryonic development in the class. But it must be admitted that our knowledge is at present quite inadequate to justify the formulation of any final conclusions.

The geological record of Formicidae is not quite what we should have expected. They are amongst the earliest Hymenoptera; remains referred to the family have been found in the Lias of Switzerland and in the English Purbecks. In Tertiary times Formicidae appear to have been about the most abundant of all Insects. At Florissant they occur in thousands and form in individuals about one-fourth of all the Insects found there. They have also been met with numerously in the European Tertiaries, and Mayr studied no less than 1500 specimens found

1 Biol. Centralbl. xv. 1895, p. 640.

2 Prof. Forel has favoured the writer by informing him of several cases of these rare intermediate forms he has himself detected.

3 Biol. Centralbl. xiv. 1894, p. 53.