306 40 



to the hæmal arches of the two foremost caudal vertebiæ; the fin-muscles originate 

 only from the first. The anterior interneural is somewhat stouter than the others 

 and with well-developed expansions for connection with the foremost pair of lower 

 lateral plates, which bound the anal space; on the following interneurals the ex- 

 pansions are only indicated. 



H. Inngirostris seems only to possess two anal interneurals. 



The number of fin rays in the dorsal fin is 17 — 18, in the anal 4; their struc- 

 ture is as in the preceding genera '**. 



The other Syngnathids examined show essentially the same structure of the 

 skull as Siphonostoma. Syngnathus (aciis and rostellatus) only differs in minutiæ in 

 the outline or form of some of the bones. In Neropliis (ceqiioreiis and ophidion) (PI. 

 V, figs. 10, 9) the most conspicuous differences are the following. The entopterygoid 

 is longer and joins the upper branch of the symplectic. The posterior antorbital 

 (ao) is considerably more elongated, in N. œquoreus reaching past the suture between 

 the mesethmoid and vomer, while the anterior is much smaller and not in con- 

 tact with the cranial beak (very often it is almost rudimentary, not reaching behind 

 to the posterior preorbital); along the latter, therefore, the entopterygoid (ept) is 

 uncovered for a long way; the front margin of ao' almost reaches the suture between 

 the quadrate and the ectoplerygoid. The bones of the skull, the suspensorial parts 

 are stouter, the maxilla broader and stouter etc. than in Siphonostoma. In Nerophis 

 ophidion the snout is short and stout, the mesethmoid and vomer therefore parti- 

 cularly powerful. In both species these two bones are about of equal length. The 

 nuchal tendons, fastened to the epiotics, are ossified in Ner. ophidion (fig. 9 t), as in 

 Aiilostoma, Fistularia and Solenostomus, but unossified in N. œquoreus. 



In Hippocampus (PI. V, fig. 6) the differences seem more obvious; they are 

 partly due to the large spines on the frontals, the smaller ones on the mesethmoid 

 and supraoccipital, but besides, the braincase proper is relatively larger in pro- 

 portion to the snout, the latter being not only shorter but also broader than in 

 the preceding genera. The nuchal face is oblique in the opposite direction, from 

 below upwards and backwards; the supraoccipital and its crest is much larger, 

 forcing the epiotics (ep) down on the sides of the skull and giving attachment to 

 the tendons (at *), which in the other genera are fastened to the epiotics. The 

 posttemporal (pt) sends out from its lower margin a process (p) to the lateral wing 

 of the exoccipital. The hyomandibular (fig. 7 Ii}') is much longer, and somewhat 

 more sloping forwards, the ascending part of the preoperculum (pro), therefore, 

 longer. As in Sipljonostoma the entopterygoid does not reach the symplectic. The 

 operculum is higher than long, the suboperculum extremelj' reduced, never visible 

 from without (and very easily overlooked); the interoperculum short but high. 

 There are three antorbitals (ao, ao', ao"). The same is found in the other 

 genera of the Hippocampine group, e. g. Solenognathus (hardwickii), Gasterotokeus 

 and Phyllopteryx ifoliatus). The posterior antorbital (ao) is small, especially in 

 Phyllopteryx (PI. V, fig. 8) ; in the latter the elongated middle antorbital (ao') is 



