GILBERT AND STARRS— FISH F.S OF PANAMA HAV 23 



on the upper surface of tlie inner rays willi ;i iiuu'li thickened foM of tlic iiiULru- 

 inoiit, as in (}. ])l<i(ijii(Kjon and G. oujciunnnii'i. 



'i'lic spcrinicns taken a<;ree well with Ei,i;riimainr~ ili-ciiplii)ii of tiie types, 

 except in tlic size (if the [talatine patcii nf teeth, whieli is in a(hdts about as hirj^e as 

 in G. eigennutmii, and many times the size of the small vomerine patcii. It is probaljly 

 true of this species, as of G.fjUherti from Mazatlan (Jordan and (Jill)ert, 1S82 h, p. 47. 

 under Arms nssimUix), that the palatine hands vary "considerably in size and some- 

 what in form, the width ranjiing from \ diameter of eye to 1, beinj:; jrenerally larj^er 

 in adults." Eigenmann's type of G. jordnni was an immature specimen. In the 

 smaller of our specimens the palatine patch is much less developed than in the 

 adult, thoMi:;h somewhat larger and more ovate than in Ei<renmann's fii^nre. 



In both of the specimens taken the fontanel groove terminates a very short 

 distance in advance of the base of tlu- <icci|>ilal process, not reaching the base of 

 process as described by Eigenmann. 



The gill-rakers number 5 + 11 and 5+12, the most anterior being very minute, 

 if tlie one at the branchial angle be reckoned with the vertical series instead of 

 the horizontal, as above, one of the specimens will have the formula (5+10, as com- 

 pared with Eigenmann's type 6 + 9. 



Comparing the Panama material with the co-types of G. (jUberti from Mazatlan, 

 it is impossible to detect any appreciable differences, unless possildy in the color, 

 which is lighter in the jMazatlan specimens. G. f/i/berti is said to lack the i>ectoral 

 pore, but in reality possesses a minute rouml pore, as is evident in the co-types exam- 

 ined. The adidt G. jordnni from Panama has the pore likewise minute, while the 

 younger example has a considerably larger, slightly elongate opening. The width 

 of the mouth is the same in the Afazatlan and Panama specimens; measured exter- 

 nally, at the posterior labial angle, it equals the distance from the tip of the snout 

 to the hiniler margin of the pupil, and is contained 2| to 2| times in the length 

 of the head. 



The two would be united without question, were it not that the co-types of 

 G.gilberti (three in number) agree perfectly among themselves and differ from the 

 Panama specimens of G. jordani in having the occipital plate much wider, more evenly 

 roundeil in transverse section, and with a much lower keel. The occipital plate is, 

 in each of these specimens, wider than long l)y half the width of the basal plate of 

 the dorsal spine. In G. jordnni the width of the occipital plate is J or I its length. 

 This plate is also much more sharply keeled and more densely granular. On the 

 basis of these differences the two species are held provisionally distinct. 



In both species the upper lobe of the caudal is longer and more falcate than 

 in related species, reaching far beyond the lower lobe, and contained 3[ to S^ in the 

 length. Tlie head is finely and often sparsely granular, the granular area not con- 

 tinued forward in any of the Panama specimens as far as a line joining posterior 

 margins of orbits. The fontanel groove widens anteriorly, is rather deep, with sharply 

 defined margins, and terminates abruptly, not "merging into the broad, flat, smooth, 

 interorbilal area" as is described in the type of G. gilbtrti. 



