526 PROCEEDINGS OF THE NATIONAL MUSEUM. vol. 41. 



straight, the angle sUghtly produced. Opercle with two stout spines. 

 Teeth bristle like, on vomer, in A -shaped band, on jaws and pala- 

 tines in narrow bands, that on upper jaw several times as broad as 

 that of lower jaw, which is in two or three series. 



Dorsal spines stiff, fifth contained If in head; the penultimate 3 in 

 head, last 2 ; longest dorsal ray 2 in head, fhst not extending beyond 

 base of last when supine; fin outline truncate, base of soft dorsal 1§ 

 in head. Soft anal base slightly longer than that of soft dorsal; 

 third anal spine 2§ in head; first soft ray extending to base of seventh. 

 Pectoral l| in head. Ventrals extending to anus, 1^ in head. 

 Caudal forked, rather deeply, inner rays slightly more than half 

 length of outer, measuring from last scales, but two-thirds from last 

 vertebra. 



Color in alcohol somewhat silvery below lateral line, much darker 

 above, frequently in irregular spots. Soft dorsal and anal narrowly 

 edged with white, black submarginally; spinous dorsal edged with 

 black; caudal dark, edged with darker. Peritoneum silvery, spotted 

 with brown. 



This species is nearest allied to KuJilia rupestris, another river 

 species from the same regions. It varies principally in a shorter 

 maxillary, shorter head, mandible, number of transverse series of 

 scales below the lateral line, form of soft dorsal, anal, of caudal, and 

 in more spotted peritoneum. In all these characters and in physiog- 

 nomy it approaches slightly to Boulengerina. 



Our Formosan specimen and the two from Mindoro, Philippine 

 Islands, correspond very closely, save for a somewhat shorter first anal 

 spine, longer soft dorsal base, and wider spread opercular spines in 

 the former. Since the number of specimens we have is so small, it 

 is very likely that these are individual variations. The Formosan 

 specimen has the first anal spine 0.055 of body length to last vertebra; 

 soft dorsal base 0.2; and spread of opercular spines 0.05. The 

 Samoan specimens have one more scale above and one more below 

 in transverse series, and gill-rakers 8 to 10 + 22 to 24 (instead of 8 + 15 

 to 17). 



The synonymy of this species, now current and given in part 

 above, is nearly worthless because of the great probability of varia- 

 tion in these fresh-water fishes from island to island. There are at 

 least two species included in Boulenger's Kulilia marginata, and it 

 is very probable that there are more. The Samoan specimens, at 

 least, are not identical with those here described. We have no 

 Japanese specimens of this fish. A memorandum of Doctor Jordan 

 refers to a specimen from Izu, in the Imperial Museum in Tokyo. 

 This is mentioned by Ishikawa as "Dules sp." 



(marginatus, edged.) 



