Summary1. Development in echinoderms may be indirect, involving pelagic, bilaterally symmetrical, larval forms, or more or less direct, with the larval stage either reduced or omitted. Of the five living classes, only the Echinoidea are characterized by being predominantly of the type with indirect development.2. It is possible to regard the dipleurula stage of the classes Asteroidea, Ophiuroidea, Echinoidea and Holothuroidea as recapitulating a common ancestral Dipleurula. It is no longer possible to regard any of the other types of echinoderm larvae as anything but specialized forms without broad phylogenetic significance.3. Embryos and larvae of echinoderms are extremely plastic, often exhibiting convergence, divergence and adaptation, susceptible to evolutionary modifications of structure which may act quite independently of the adult stage. Ancestral structure cannot be deduced from such forms.4. In the Echinoidea larval evolution seems to have occurred subsequently to the separation of the main orders and families. Within relatively small groups larval evolution has followed similar trends, so that characteristic larvae occur in various sub‐groups, where the young stages may follow similar ontogenies; but such independent evolution tends to obscure the phylogenetic relationships between the class as a whole and the other classes.5. In the Asteroidea larval evolution has occurred along channels not so markedly correlated with the taxonomy of the adults. Phylogenetic speculations based on such larval stages prove incompatible with other evidence.6. In the Holothuroidea and Ophiuroidea larval evolution cannot at present be related with adult taxonomy, save in one or two cases too unimportant to have general significance.7. The egg of echinoderms is liable to undergo changes in volume. Increase of volume is directly related to increase in cytoplasm and its product, the yolk material. Such increases have led to direct development.8. Increase in cytoplasm and yolk has not greatly affected the cleavage process, which is almost always total. A distinction between micromeres and macromeres frequently results.9. With increasing cytoplasm, the wall of the blastula becomes thicker, and the blastocoel is in extreme cases reduced to a vestige in the animal hemisphere. The mesenchyme fails to separate as such, but projects as a solid mass into the blastocoel. Invagination is reduced to a solid inpushing of cells, and epiboly may ensue. The archenteron may become vestigial, in which case the definitive enteron is excavated in the solid endoderm by splitting. The enterocoele become reduced or lost, and the coelom and its adjuncts may arise by schizocoelous splitting in mesenchyme.10. In Ophiuroidea a succession of stages in reduction of the ophiopluteus may be seen, suggesting a recession backwards in time of the moment at which metamorphosis is initiated. In extreme cases the gastrula itself becomes radially symmetrical and the larva is completely lost.11. By convergent evolution among echinoderms with yolky eggs, a special vitel‐laria larva has arisen independently in Holothuroidea, Ophiuroidea and Crinoidea. The vitellaria is characterized by its barrel shape, and the transmutation of the ciliated band into annuli. In the Crinoidea this is the only larva as yet known.12. Viviparity does not seem to have been an important factor in causing direct development, though it may influence the physiology and morphology of the young stages.13. If larval stages of echinoderms are interpreted as recapitulating ancestral stages, the conclusions reached are seriously discordant with other evidence. Therefore it is not possible to base phylogenetic interpretations on larval stages alone.14. Echinoderm embryology cannot provide any valid support for the hypothesis that chordates arose from echinoderms.