282 16 



form a strip between the posterior end of the coracoidal plate and the ventrals, 

 and behind these running bacl^wards, below and parallel to the lateral line, to the 

 end of the tail. In the youngest stage figured (PI. VII, fig. 1) I have not been able 

 to detect these structures, but in a specimen of about 280 mm. they are discernible, 

 at all events anterior to the ventrals, and in larger ones they always seem to be 

 present"*. 



III. In Fistularia petimba a single median row of narrow longitudinally keeled 

 scales, quite superficially imbedded, is found along the back, running from about 

 the level of the ventrals to the base of the dorsal fin, and behind the latter to 

 some distance from the end of the tail, where the hindmost scales grow shorter 

 and finally take the shape of nodules; a similar row is found in the ventral median 

 line, beginning in front of the ventrals at some distance behind the coracoids and, inter- 

 rupted by the anus and anal fin, continuing on the tail, stopping at some distance from 

 the caudal fin. In all large and medium-sized specimens I find these scales present, but 

 in the smallest specimens (170 and 130 mm.) they seem not yet to be developed''. 



IV. Common to all 3 species are the dermal ossifications of the lateral line. 

 On the short anterior part of the trunk, which is protected by the endoskeletal 

 parts described below, these ossicles are tube-shaped, on the rest of the body they 

 gradually take the form of narrow longitudinal shields each provided with a more 

 or less prominent short keel or spine, most prominent on the tail where the shields 

 form a kind of serrature. On the caudal filament, on which the lateral line is 

 continued, the ossifications again take the shape of tubes. In the two species 

 where dermal asperities are found, the latter may be present in the thin dermal 

 layer covering the ossicles of the lateral line "'. 



Endoskeleton. 



The vertebral column does not show any important differences in the 

 two species examined (F. petimba and tabacaria). In both species I have found 

 the same number of vertebræ, 56 (4 + 52) abdominal and 31 caudal vertebræ". 



The 4 anterior vertebræ (PI. I, fig. 11, PL II, fig- 9) are modified in the same 

 way as in Aulostoma, being extremely elongated and immovably united by sutures, 

 forming one piece, which (in petimba) is between '/.=. and '/r, of the whole length 

 of the column. The spinous processes are united into one longitudinal crest, the 

 transverse processes into a similar, but broader, almost horizontal plate, the mem- 

 bers constituting which overlap each other towards the skull while those of the 

 vertical plate are mostly joined with their margins. 



As in Aulostoma the second and third vertebræ are by far the longest, their 

 anterior and posterior portion being about equally elongated, whereas of the first 

 vertebra only the posterior part, of the fourth only the anterior part is elongated 

 and modified. 



The first carries articular processes for connection with the skull, the fourth 

 has on the posterior, unmodified part a small separate transverse process, corre- 



