these which confirm my theory. A glance at the illustration (Pl. III) shows that Andricus noduli has a very short egg-stalk with a relatively long ovipositor. We must remember that Andricus noduli lays its egg in August, in the cambium layer of the oak bark: but there can be no want of oxygen in a plant tissue in which metabolism is constantly going on, and it is not therefore necessary to inhale oxygen by means of the egg-stalk. The summer generations of many species lay their eggs under the same favourable circumstances as Andricus noduli, therefore their egg-stalks are short. But this holds good only of species which lay their eggs in leaves; and we must except those which prick the winter buds. As winter buds are properly speaking resting buds, the surrounding plant tissues can afford no nourishment, consequently the egg-stalk must be long enough to come in contact with the outer air. Against this explanation of the use of the eggstalk the fact may be advanced that in other hymenoptera this contrivance is wanting, but it would not be difficult to prove that in none of these cases is such an arrangement needed. Thus all the Ichneumonidae give over their eggs to the selected host, and in its body they obtain all the nourishment they require. Many saw-flies also sink their eggs into various parts of the plant, but they do so always at a time when active metabolism is going on. In the case of oak gall-flies on the contrary, the eggs of most of the winter generations are laid at a time when the plants themselves give no signs of life, and when metabolism is dormant.

Another apparent objection is that inquilines closely

allied to the Cynipidae have also stalked eggs. In this case we cannot ascribe to the stalk the important functions of a respiratory organ, as the egg does not need it; but nobody can doubt that their great conformity in outward habit and whole organization proves that these inquilines have been evolved from the Cynipidae, and the peculiarity of the stalked egg has remained with them as a survival. But in their case the egg-stalk is not required to act in the way described, and does not do so; because the peculiar conditions of embryonic development, which we find existing in Biorhiza aptera, are wanting in them.

In order to explain the function of the egg-stalk, I have spoken of embryonic development in its advanced stage; but the egg-stalk plays a part also at the moment when the egg is being laid. It has already been explained that the egg cavity communicates with the egg-stalk, consequently part of the contents of the egg can pass without impediment into the stalk. This happens regularly at the laying of each egg. If an egg which has been laid by a fly in a bud, is afterwards removed, the preparation will show the first part of the egg-stalk to be quite full of a finely granular emulsion which forms the egg contents. After a time changes take place in this emulsion, and there are formed in it highly refractive globules of various sizes; finally the whole contents of the egg-stalk clear, and a fine membrane appears, separating the egg-stalk from the egg cavity. These preparatory changes are always a sure sign that egg development is taking its proper course; a matter as to which we may sometimes be in doubt

when we remove a freshly laid egg from a bud, to be further examined in the damp chamber. Although I have frequently observed this occurrence in the eggstalk, I am unable to explain it more clearly. This much is certain, that it is a phenomenon of great importance, and it is only after it has taken place, that eggs kept in the damp chamber are observed to undergo the various stages of embryonic development. But I have never succeeded, no matter how careful the precautions adopted, or what the modifications under which I carried out the experiment, in bringing an egg to perfect development, after removal from the ovary of a parthenogenetic fly.

CHAPTER V.

COMPARATIVE CLASSIFICATION OF CORRESPONDING GENERATIONS OF CYNIPIDAE ACCORDING TO THEIR ORGANIZATION.

THE activity of the fly culminates in the act of oviposition; care for its progeny occupies the whole period of its individual existence. I began therefore by giving a description of the complicated apparatus by which egg-laying was carried out, and I now proceed to compare the whole organization of the two generations in the various stages of imago and larva. I have already described in detail the outward configuration of the gall-flies, and the differences in colouring, sculpture, and pubescence which they exhibit.

Outward characters are usually of little distinctive value in gall-flies; a uniformly dull colour prevails in almost all species; and there are many which cannot be distinguished from each other by an outward examination of the flies only. In comparing two alternating generations, colouring counts for little; their form, structure, and size are of much greater significance. In these respects there are very important differences in the two generations. If a Neuroterus and its corresponding Spathegaster form be compared with

K

each other, they could never be mistaken in spite of considerable similarity of colouring. The size may be nearly the same, but the form of the thorax, the cut of the wings, and the configuration of the abdomen are so different, that it would be impossible to confound the two insects. These outward differences are still further confirmed by the form and structure of the ovipositor. The delicate little ovipositor of Spathegaster takes up very small space, while the long spirally-coiled ovipositor of Neuroterus requires the whole abdominal cavity to contain it, and hence the difference in the contour of the abdomen. Again, the manner in which Spathegaster bores into the leaves requires a greater mobility of the abdomen, and so we find it is distinctly pedunculated; Neuroterus on the contrary is almost sessile. Lastly, Spathegaster selects only leaves of very tender consistence in which to lay its eggs, and must consequently be able to fly easily; we find Spathegaster therefore provided with longer and broader wings than Neuroterus, which can everywhere find buds in which to lay its eggs, and needs no particular power of flight. While in a manner we can thus construct the whole insect from the ovipositor, it is also possible to distinguish with certainty the various genera by its different methods of action. If two alternating generations live in perfectly distinct outward conditions, it is absolutely essential that the ovipositor should accommodate itself to them, and take that form which is most suitable for the proper insertion of the eggs. If one generation appears at a season when there are only buds to be found, it must be provided with an