"No doubt these external gills are absent also in a few of the admittedly primitive forms such as, e.g., (_Neo-_) _Ceratodus_. But I would ask that in this connection one should bear in mind one of the marked characteristics of external gills--their great regenerative power. This involves their being extremely liable to injury and consequently a source of danger to their possessor. Their absence, therefore, in certain cases may well have been due to natural selection. On the other hand, the _presence_ in so many lowly forms of these organs, the general close similarity in structure that runs through them in different forms, and the exact correspondence in their position and relations to the body can, it seems to me, _only_ be adequately explained by looking on them as being homologous structures inherited from a common ancestor and consequently of great antiquity in the vertebrate stem." As to the third theory, Professor Kerr suggests tentatively that the external gill may be the structure modified to form the paired limbs. Of the homology of fore and hind limbs and consequently of their like origin there can be no doubt. The general gill-structures have, according to Kerr, "the primary function of respiration. They are also, however, provided with an elaborate muscular apparatus comprising elevators, depressors, and adductors, and larvæ possessing them may be seen every now and then to give them a sharp backward twitch. They are thus _potentially_ motor organs. In such a Urodele as _Amblystoma_ their homologues on the mandibular arch are used as supporting structures against a solid substratum exactly as are the limbs of the young _Lepidosiren_. [Illustration: FIG. 62.--_Polypterus congicus_, a _Crossopterygian_ fish from the Congo River. Young, with external gills. (After Boulenger.)] "I have, therefore, to suggest that the more ancient Gnathostomata possessed a series of potentially motor, potentially supporting structures projecting from their visceral arches; it was inherently extremely probable that these should be made use of when actual supporting, and motor appendages had to be developed in connection with clambering about a solid substratum. If this had been so, we should look upon the limb as a modified external gill; the limb-girdle, with Gegenbaur, as a modified branchial arch.