Summary.The chief contents of this paper may be summarised as follows. Three occipital segments occur in the head ofAmblystoma. Of the three somites developed in these, the first forms no muscle and disappears early the second forms a myotome divided into dorsal and ventral portions, of which the former alone persists, fusing with the myotome next behind. The myotome of the third segment persists dorsally, that of the first trunk‐segment combines with it to form the temporal longitudinal muscle supplied in the adult by the first spinal nerve. The glosso‐pharyngeal nerve belongs to the first metaotic segment and has no ventral root. To the second segment belongs the vagus root, with which seems to be combined the ganglia of the next two nerves. No ventral root was found in the second segment; but a ventral root occurs in the Iast metaotic segment in early stages of development. The first spinal issues from between the skull and first neural arch as a ventral root only; it joins the second spinal to form the hypoglossal nerve supplying muscles derived from ventral outgrowths of the second. third, and fourth myotomes. The basilar plate of the skull is formed by a backward growth of the trabeculæ: meeting a forward growth of the base of the orcipital arches. It fuses with the auditory capsules, and with the praæoccipital arches developed in the septum between the first and second myotomes. The occipital arches arise in the septum between the second and third myotomes. There is no evidence of the disappearance of segments behind the vagus, and no valid objection to the view that the hind limit of the skull may shift backwards or forwards in the course of phylogeny. At the same time there is no reason to suppose that the ancestors of the Amphibia had more than three occipital segments, when the occipital condyle became clearly de