The fact that the Amphibia possess only external gills throws important light upon their phylogeny. Not only do the Apoda, Urodela, and Anura agree much more with each other than {44}would be the case if the Anura possessed both internal and external gills, but the Amphibia reveal themselves also in this point as connected with the Crossopterygii and the Dipnoi, some of which fishes also possess external gills. It is of course quite possible that the Amphibia have developed these organs independently, but we understand now that the latter are accessory, and not the primitive respiratory organs; they are developed in adaptation to embryonic conditions and to prolonged larval, occasionally perennibranchiate, aquatic life (cf. the chapter on Neoteny, p. 63). There is no valid reason for supposing that the Stegocephali had true internal gills. We know their branchial skeleton, and we can discern even gill-rakers on the arches. Such gill-rakers occur also, although but feebly developed, in Urodela. The whole branchial framework of the Urodela and Apoda undergoes simple reductions during metamorphosis (see p. 86), but in the Anura these arches are in early tadpole life transformed into a most complicated basket-work which acts as a straining apparatus or filter, to prevent any particle of food or other foreign matter from finding its way into the delicate gills, the current of water passing from the mouth through the filter, past the gills and out of the clefts. During metamorphosis this whole elaborate apparatus is again transformed, almost beyond recognition, into the hyoidean apparatus for the support of the generally very movable and much-specialised tongue. The fact that the hyoid apparatus of the Aglossa, especially that of _Xenopus_, is constructed upon the same lines, is a strong indication that these creatures have arrived at their tongueless condition through the loss of this organ, and this is intelligible in correlation with their absolutely aquatic life. The opercular folds assume great dimensions in all tadpoles. They cover the whole gill-region, thereby producing on either side an outer gill-chamber. The posterior margins of the folds gradually become continuous with the rest of the surface of the body. Each gill-chamber opens at first by one lateral canal, usually called the spiracle. This condition prevails in the tadpoles of the Aglossa. In the Discoglossidae the two canals gradually converge and combine into one median opening on the middle of the belly. In all the other Anura the right opening becomes closed, or rather its canal passes over to and joins that of the {45}left side, both opening by one short tube laterally on the left side, at a variable distance between the eye and the vent. Hence the elegant terms of Amphi-, Medio-, and Laevo-gyrinidae (γυρῖνος being the Greek for tadpole).