=SUBORDER Hemibranchii.=—Still another transitional group, the _Hemibranchii_, is composed of spiny-rayed fishes with abdominal ventrals. In this suborder there are other points of divergence, though none of high importance. In these fishes the bones of the shoulder-girdle are somewhat distorted, the supraclavicle reduced or wanting, and the gill structures somewhat degenerate. The presence of bones called interclavicles or infraclavicles, below and behind the clavicle, has been supposed to characterize the order of _Hemibranchii_. But this character has very slight importance. In two families, _Macrorhamphosidæ_ and _Centriscidæ_, the interclavicles are absent altogether. In the _Fistulariidæ_ they are very large. According to the studies of Mr. Edwin C. Starks, the bone in question is not a true infraclavicle. It is not identical with the infraclavicle of the Ganoids, but it is only a backward extension of the hypocoracoid, there being no suture between the two bones. In those species which have bony plates instead of scales, this bone has a deposit of bony substance or ganoid enamel at the surface. This gives it an apparent prominence as compared with other bones of the skeleton, but it has no great taxonomic importance. Dr. Hay unites the suborders _Hemibranchii_, _Lophobranchii_, and _Hypostomides_ to form the order _Phthinobranchii_ (φθινάς, waning; βράγχος, gill), characterized by the reduction of the gill-arches. These forms are really nearly related, but their affinities with the _Percesoces_ are so close that it may not be necessary to form a distinct order of the combined group. Boulenger unites the _Hemibranchii_ with _Lampris_ to form a group, _Catosteomi_, characterized by the development of infraclavicles; but we cannot see that _Lampris_ bears any affinity to the sticklebacks, or that the presence of infraclavicle has any high significance, nor is it the supposed infraclavicle of _Lampris_ homologous with that of the _Hemibranchii_. The dorsal fin in the _Hemibranchii_ has more or less developed spines; spines are also present in the ventral fins. The lower pharyngeals are separated; there is no air-duct. The mouth is small and the bones of the snout are often much produced. The preopercle and symplectic are distinct. The group is doubtless derived from some transitional spiny-rayed type allied to the _Percesoces_. The Lophobranchs, another supposed order, represent simply a still further phase of degradation of gills and ventral fins. Dr. Gill separates these two groups as distinct orders and places them, as aberrant offshoots, near the end of his series of bony fishes. We prefer to leave them with the other transitional forms, not regarding their traits of divergence as of any great importance in the systematic arrangement of families.