90 CALIFORNIA ACADEMY OF SCIENCES 



diflference between them in this respect. The appendages to the air-bladder vary 

 widely in length, from a trifle less than the diameter of the eye to twice the diameter, 

 in our Panama specimens. In the few individuals we have been able to examine 

 from the Atlantic (Havana and Jamaica), the appendages vary from I5 to 1^ times 

 the orbital diameter. It is highly improbable, therefore, that this character possesses 

 any significance. 



161. Centropomus nigrescens Oilnther. 



Only a few small specimens seen. 



162. Centropomus pedimacula Poey. 



Very abundant at Panama. 



We have had for comparison a single specimen from the Atlantic, which 

 shows a slightly longer anal spine and a slightly wider interorbital space than any 

 of our Pacific examples. In all other characters there seems to be perfect agreement 

 with our Panama specimens. We believe that the slight differences found are well 

 within the range of variation of the species, and shall therefore not recognize Centro- 

 pomus medius 1[the Pacific form) as distinct in any degree. A full series from the 

 Atlantic may eventually prove, however, that these characters have a higher value 

 than we have assigned to them, and necessitate specific separation of the two forms. 



C. grancloculatus is certainly not separable from C. medius {^pedimacula). 

 The scale-counts of different authors vary widely in this group, as some have included 

 a number of the scales overlying the base of the caudal fin, and others cease the 

 enumeration at a point corresponding to the base of the median rays. In this paper 

 we have adopted the latter course, and find very uniformly 47 or 48 tubes in speci- 

 mens from Mazatlan, Panama and Havana, including those from the San Juan 

 Lagoon, Mexico, upon which is based the description of grandoculatus given by 

 Jordan and Evermann. The first dorsal fin contains uniformly 8 sj^ines, as in all other 

 species of the genus. The describers of grandoculatus seem to have neglected the 

 first spine, which is very short and often concealed by the scales. Other characters 

 assigned to grandoculatus, such as the form of the body and the size of the eye, have 

 no significance. The amount of black on the ventral fins is also subject to wide 

 variation, and has no distinctive value. In young specimens, the black is usually 

 very distinct, but in adults it grows fainter and often disappears. 



163. Centropomus unionensis Bocourt. 



Plate XIII, Fig. 27. 



Abundant in Panama Bay. 



The species is closely related to armatus and rohalito, but has shorter anal 

 spines and weaker ridges and spines on head than in either species. It agrees with 

 rohalito in the slender, flexible dorsal spines, the second one of which is very short, in 

 the anterior position of the vent andin the pale lateral line; with armatus in the small 

 number of gill-rakers, and the increased number of scales in a vertical series above 

 the lateral line. The second anal spine scarcely extends beyond the tip of the longest 



