Yet another clue to the meaning of these muscles is to be found in their innervation, which is very extraordinary and unexpected. Throughout the branchial region the striated muscles of each segment {311}are strictly supplied by the nerve of that segment, and, as already described, each segment is as carefully mapped out in its innervation as it is in any arthropod appendage. One exception occurs to this orderly, symmetrical arrangement: a nerve arises in connection with the facial nerve, and passes tailwards throughout the whole of the branchial region, giving off a branch to each segment as it passes. This nerve (_Br. prof._, Fig. 123) is known by the name of the _ramus branchialis profundus_ of the facial, and its extraordinary course has always aroused great curiosity in the minds of vertebrate anatomists. Miss Alcock, by the laborious method of following its course throughout a complete series of sections, finds that each of the segmental branches which is given off, passes into the tubular muscles of that segment (Fig. 124). The tubular muscles which belong to the velum, _i.e._ those belonging to the lower lip-segment and to the hyoid segments, receive their innervation from the velar or mandibular nerve, and belong, therefore, to the trigeminal, not to the facial, system. [Illustration: FIG. 123.--DIAGRAM SHOWING THE DISTRIBUTION OF THE FACIAL NERVE. Motor branches, _red_; sensory branches, blue.] The evidence presented by these muscles is as follows:-- In the ancestor of the vertebrate there must have existed a segmentally arranged set of dorso-ventral muscles of peculiar structure, concerned with respiration, and confined to the mesosomatic segments and to the last prosomatic segment, yet differing from the other dorso-ventral muscles of respiration in their innervation and their attachment. Interpreting these facts with the aid of my theory of the origin of vertebrates, and remembering that the homologue of the vertebrate ventral aorta in such a palæostracan as Limulus is the longitudinal {312}venous sinus, while the opercular and chilarial segments are respectively the foremost mesosomatic and the last prosomatic segments; they signify that the palæostracan ancestor must have possessed a separate set of segmental dorso-ventral muscles confined to the branchial, opercular and chilarial or metastomal segments, which, on the one hand, were respiratory in function, and on the other were attached to the longitudinal venous sinus. Further, these muscles must all have received a nerve-supply from the neuromeres belonging to the chilarial and opercular segments, an unsymmetrical arrangement of nerves, on the face of it, very unlikely to occur in an arthropod. [Illustration: FIG. 124.--DIAGRAM CONSTRUCTED FROM A SERIES OF TRANSVERSE SECTIONS THROUGH A BRANCHIAL SEGMENT, SHOWING THE ARRANGEMENT AND RELATIVE POSITIONS OF THE CARTILAGE, MUSCLES, NERVES, AND BLOOD-VESSELS.