bud ontogeny there is no trace of anything which can be interpreted as ancestral structures, and that some most striking and recent changes, such as the loss of limbs in snakes or the reduction of the toes of the ostrich to two, are not recorded in embryology, i. e. the organ concerned shows from its inception the adult arrangement.

How then does the theory we have adopted account for the retention of ancestral characters by larvæ ?

So far as we can judge by comparative anatomy, the stimuli to evolution (in the sense of change of structure) have been two, viz. (1) change of environment and habits, and (2) increased or decreased demands on the working of certain organs. As we therefore pass along a series of genetically connected animals, we should find, pari passu with the environment and the functional demands of the organism, the structure changing. If these stimuli commenced to act from the beginning of free life, then each individual adult in the chain would show from the beginning the modified structure belonging to it; but if these stimuli were deferred in their operation till the animal had attained a certain size, then what was before a uniform life-history would become differentiated into two periods-a larval during which the ancestral habits were retained and the structures corresponding to them, and an adult in which new habits were assumed and structure correspondingly modified.

An illustration will make this clear: if young flat-fish when they emerge from the egg were at once to adopt the adult mode of life, then that most interesting larval stage, in which they are bilaterally symmetrical, would be missed out in their development.

Thus we see as a race of animals progressed from point to point in evolution, it would tend to develop a trail of larval stages, each grade of development surmounted being represented by a new larval stage intercalated in the ontogeny. This process, however, could not go on indefinitely; there would soon arise the tendency for the earlier larval stages to be passed over whilst still in the egg-membrane, and so a

is embryonic, but in every case larval development is preceded by a longer or shorter period of embryonic development.

The whole interest of the science of Embryology lies, of course, in the fact that features observed in both types of development seem inexplicable except on the assumption that they are reminiscences of structures possessed by the ancestors of the animals in whose development they appear. Such traces of the history of the race are to be found in the vast majority of larvæ; in embryos they are likewise to be found, though here they are less prominent, as is seen by comparing the development of two allied forms, in one of which the larval type prevails, and in the other the embryonic. Now Mr. Sedgwick's theory of the relation of the two types to one another is that that portion of embryonic development in which ancestral features are observable represents a larval stage passed over inside the uterus or egg-membrane and modified in consequence. Thus the chick during the first four or five days of its existence is to be regarded as an immensely modified larva.

If this view be true it follows that, however modified the record of ancestral history contained in the larval development may be, the embryonic record of the same history can never rise above it in value.

It was until lately customary to assume, explicitly or implicitly, that there was an inherent tendency for the ontogeny of the individual to be a summarised repetition of the phylogeny of the race. In proof of this statement we may adduce Balfour, who in his 'Text-book of Comparative Embryology' (vol. ii, p. 298), says, "Unless secondary changes intervened this record [of ancestral history] would be complete;" and Bateson, in his discussion of the ancestry of the Chordata, commits himself to a similar position. That there can be no such general tendency is, however, shown by the fact that in

"The Ancestry of the Chordata," W. Bateson, 'Quart. Journ. Micr. Sci., 1886. "Development within an egg-shell as involving a less complicated struggle with enviromental forces, is less subject to variation than that in the open sea, and consequently is more likely to preserve ancestral features."

bud ontogeny there is no trace of anything which can be interpreted as ancestral structures, and that some most striking and recent changes, such as the loss of limbs in snakes or the reduction of the toes of the ostrich to two, are not recorded in embryology, i. e. the organ concerned shows from its inception the adult arrangement.

How then does the theory we have adopted account for the retention of ancestral characters by larvæ?

So far as we can judge by comparative anatomy, the stimuli to evolution (in the sense of change of structure) have been two, viz. (1) change of environment and habits, and (2) increased or decreased demands on the working of certain organs. As we therefore pass along a series of genetically connected animals, we should find, pari passu with the environment and the functional demands of the organism, the structure changing. If these stimuli commenced to act from the beginning of free life, then each individual adult in the chain would show from the beginning the modified structure belonging to it; but if these stimuli were deferred in their operation till the animal had attained a certain size, then what was before a uniform life-history would become differentiated into two periods—a larval during which the ancestral habits were retained and the structures corresponding to them, and an adult in which new habits were assumed and structure correspondingly modified.

An illustration will make this clear: if young flat-fish when they emerge from the egg were at once to adopt the adult mode of life, then that most interesting larval stage, in which they are bilaterally symmetrical, would be missed out in their development.

Thus we see as a race of animals progressed from point to point in evolution, it would tend to develop a trail of larval stages, each grade of development surmounted being represented by a new larval stage intercalated in the ontogeny. This process, however, could not go on indefinitely; there would soon arise the tendency for the earlier larval stages to be passed over whilst still in the egg-membrane, and so a

portion of the development would become embryonic, and so subjected to the various modifying influences which are connected with this type of ontogeny. Therefore it follows, as the first important deduction from Sedgwick's theory, that in seeking to obtain a basis for phylogeny, most importance must be attached to animals which show long larval histories.

Balfour with his usual sagacity has, so to speak, instinctively anticipated this conclusion. Although he points out that "the favourable variations which may occur in the free larva are much less limited than those which can occur in the fœtus," he says that there is "a powerful counterbalancing influence tending toward the preservation of ancestral characters, in that larvæ are compelled at all stages of their growth to retain in a functional state such systems of organs at any rate as are essential for a free and independent existence" (Comp. Emb.,' vol. ii, p. 299).

The objection, alluded to in Balfour's statement, that larvæ as well as adults have been subjected to the modifying influences of their environment, will readily occur to most minds. Let us consider whether it is possible to approximately estimate the nature and amount of such influences; and, first of all, let us consider what is meant by secondary larvæ.

Balfour imagined that secondary larval forms might be produced by a diminution in the food yolk, and consequent earlier commencement of free existence (loc. cit., p. 300). There is no evidence to suggest that such a change has ever taken place; all the facts point in a contrary direction. We shall see that food yolk produces the most diverse distortions of development; the developmental processes of free larvæ are, on the other hand, remarkably uniform. Secondary larvæ must be regarded as having arisen owing to the young adults having taken to a new mode of life; the best instances of this are perhaps the aquatic larvæ of the may-flies and dragon-flies. We have the strongest reason for believing that the immediate ancestors of insects were terrestrial animals, and the aquatic larvæ mentioned show their secondary character by the fact that their respiratory organs are modified from organs adapted

to air breathing. Their "tracheal" gills, instead of like all other gills bringing the blood into close proximity to the water, bring their blood first into contact with air contained in a system of closed tubes, and then this air into contact with the

water.

In the case of ordinary larvæ, the probability of modifica tions due to adaptation to the environment cannot be denied. If, however, Sedgwick's hypothesis is correct that "larval history is constructed out of ancestral stages," or, in other words, that the larva retains ancestral characters because it retains the ancestral mode of life, then the environment has remained to a large extent constant (at any rate in the commonest case, that of pelagic larvæ), and the changes they are likely to have undergone, instead of being, as Balfour supposed, unlimited, will be comparatively few in number.

Of these changes reduction in size is the most important. The passage to the adult state is often accompanied by the loss of larval organs, and great changes in those which are retained, necessitating in some cases the complete destruction of their constituent cells, and their reconstruction from rudiments which have retained the embryonic condition (histolysis). It is, therefore, clearly to the advantage of the larva to grow no larger than necessary before it undergoes metamorphosis. Correlated with this loss of size is the frequent disappearance of all traces of segmentation, since this is probably to be regarded as essentially the same phenomenon as vegetative reproduction, only held in check by the individuality of the whole. Metameric series of organs are represented only by those members which are absolutely necessary. Another change which larvæ are prone to undergo, is the acquisition of transparency. What results this carries in its train will be mentioned below. Finally, the occurrence of long spines is a widespread phenomenon, though what their precise use is it would be rash to surmise. Possibly they are of a protective nature. Let us now apply these principles in a concrete case, for example the larvæ of the Crustacea.

The characteristic larva of the Entomostraca is the well