226 



TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY 



in fact the leading specializations of the Fistularia skull together with others of its own, 

 notably the presence of a row of "antorbital plates" on the sides of the oral tube and the 

 downward bending of the basioccipital, in correlation with the upright posture of the body. 

 The skulls of Nerophis (PI. V, Figs. 9, 10) and of Siphonostoma (PI. V, Figs. 1-5) appear to 

 be less specialized and tend further to connect the syngnathid with the aulostomid stem. 





Fistulana sp. 

 Fistularia serrata 



Fig. 105. Fistularia sp. After Jungersen. 



Nevertheless it seems not improbable that part of the resemblances between the syn- 

 gnathid and the aulostomid skulls may be due to parallelism, in view of the fact that 

 Solenostomus (Fig. 107) tends strongly to connect the syngnathids with the centriscoids. 



The centriscoids, in spite of their peculiar specializations, probably indeed stand nearer 

 to the gasterosteoid stem than do the solenostomids. Centriscus (Fig. 108) could be de- 



FiG. 106. Phyllopteryx foliatuj. After Jungersen. 



rived from a form like Spinachia by the marked elongation of the snout and by the great 

 increase in' size of one of the dorsal spikes. The latter is so enormous as to require the 

 magnification and close appression of the first four vertebral spines, epineurals and centra. 

 Amphisile (Fig. 109) is obviously only a peculiarly specialized centriscid. 



