302 36 



ossified, while in all the following it is cartilaginous. The epibranchial is some- 

 times incompletely ossified. The second gill-arch is composed of the typical 4 

 elements, the hypobranchial (hvn) joins its fellow at the posterior end of the se- 

 cond basibranchial; its cartilaginous hind end passes without any articulation di- 

 rectly into the cartilaginous part of the ceratohyal (cn); the latter is widely separated 

 from the epihyal (en) which joins the considerably longer pharyngobranchial (phn). 

 The third gill-arch is devoid of hypobranchial, its epibranchial separated from the ce- 

 ratohyal but articulated to the pharyngobranchial; the latter (phm) is joined to the 

 upper and inner side of the preceding pharyngobranchial. The fourth and fiflth 

 arches have only the ceratobranchial. Thus the branchial apparatus is still more 

 reduced than in Aulostoma and Fistularia, not only the second and third epibran- 

 chials but also the first being separated from their ceralohyals; the fourth pha- 

 ryngobranchial is lost, as well as the liypobranchials of the first and third arches; 

 but two basibranchials are present; no trace of teeth. As the interspaces between 

 the arches in the ventral median line are rather large, the whole gill-bearing appa- 

 ratus appears elongated and reaches farther back than is usual in fishes; ac- 

 cordingly the operculum is also elongated. The five gill-clefts on each side are 

 surrounded by gill-rakers, which are pointed and fairlj' long and contain an ossi- 

 fication (sometimes divided into two, the point being separately ossified); the gill- 

 rakers are never directly joined to the skeleton, but enclosed in the skin, the post- 

 erior or inner row being farthest awaj' from the skeleton; they do not go beyond 

 the boundaries of the ceratobranchials. 



The pectoral arch. This part seems to have been thoroughly examined by 

 only two or three authors, Parker (in his great work on the Shoulder-girdle (40) 

 1868), Smitt (54), and lately Goodrich (15). None of these authors have made out 

 the facts correctly; consequently their interpretation cannot be correct. 



It is well known that the dermal armature in all Syngnathidœ is composed 

 of large bony scutes, arranged regularly in series and forming rings round the 

 body. In Siphonostoma typhle each ring of the trunk anterior to the dorsal fin and 

 the anal opening is composed of 7 plates (PI. Ill fig. 2), 3 pairs — namely the sup- 

 erior (si), median (ml) and inferior (il) lateral plates — , and one unpaired, the 

 ventral plate. The interspaces between the large plates are filled out by small 

 and thin ones forming one median series along the back, two along each lateral 

 aspect (i) and one along each side of the belly. This arrangement is found imme- 

 diately behind the pectoral fin. Anterior to the pectoral the arrangement is some- 

 what different. Here the clavicle (cl) forms part of the exoskeleton; the superior 

 lateral plate is wanting, likewise the ventral plate; but in the closely related 

 Syngnalhiis the ventral is present. In the dorsal median line is found an unpaired, 

 rather thick and solid longitudinal plate (n), and anterior to this, behind the skull, a 

 similar one, but smaller (n'). These two "nuchal" plates I regard as belonging to 

 the endoskeleton, being in my opinion interspinous bones. Finally a plate is found, 

 which seems to correspond to the true or median lateral plate of the following 



