EDWIN CHAPIN STARKS 11 



the snout and is contained in the head from 5 to 5 3^ times. The maxillary 

 has a bluntly rounded end which is even with the posterior end of the 

 mandible. 



23. Lycengraulis grossidens (Cuvier). 



A few specimens were seined in the harbor of Natal, the largest 6J/^ 

 inches in length. 



The head is contained from 4 to 4^ times in the length to the caudal 

 base; the depth from 4^^ to 4^ times. The eye is longer than the snout 

 and is contained from 2 times in the postorbital part of the head (in speci- 

 mens 3^2 inches long) to 2^^ times (in specimens from 5 to 6 inches long). 

 The number of teeth in the mandible is very variable, as might be expected 

 from their uneven size and position. The maxillary teeth are more even 

 in size and smaller. Those on the anterior part of the bone are directed 

 downward or slightly backward, while those on the posterior part are di- 

 rected slightly forward. The mandible curves upward and is sharp at the 

 tip. The maxillary is slender, slightly widened, lance-head-like, posteriorly, 

 and ends in a slender point just behind the articulation of the mandible. 

 The head is rather broadly rounded behind at the gill cover, and not 

 particularly elongated obliquely. 



The insertion of the dorsal is midway between the base of the caudal 

 and a point varying from the middle to the posterior edge of the eye. The 

 front of the anal is under the middle of the dorsal. In the large specimens 

 the pectorals reach to, or nearly to, the ventrals. 



The head and particularly an area just behind the upper half of the 

 shoulder girdle is richly supplied with ramifying canals. On the cheek 

 and on an area behind and above the eye the canals form a net-work, in- 

 closing small spots that appear, at first sight, like well separated, imbedded 

 scales. 



24. Lycengraulis batesii (Gunther). 



A single specimen, 8 inches long, was collected in the market at Para. 



In this specimen the teeth in the mandible are fewer and a little farther 

 apart than in Lycengraulis grossidens, though more were described for the 

 type. Their number is doubtless variable. This species differs in being 

 much more slender, in having the front of the anal only slightly behind 

 that of the dorsal, in having the head shorter as compared with the entire 

 length, and in having the gill-rakers shorter. The last are low on the side 

 of the arch and are about as long as the mandibular teeth, though they are 

 not "like tubercles," as originally described. 



