Cockerell — Scales of Some Malacopterygian Fislics. 113 



Nowhere are the circuli at all moniliform. From the nucleus a radius 

 extends on each side laterally, and from it spring three or four 

 upwardly directed radii, following the lines of the apical circuli; two 

 of these apical radii may start almost from the nucleus. There are 

 three or four strong basal radii. 



It thus appears that although the scales of Pantodon and Phracto- 

 Ixnius may both be described, on superficial characters, as Alestiform, 

 they are extremely diflferent in detail, the whole system of circuli being 

 diverse. The scale of Phractoliemus is incipiently ctenoid. 



Boulenger states tliat the Phractolaemidee are an isolated group, 

 nearest apparently to the Osteoglossidae. 



Kneriidse. 



Knerin cameronensis Boulenger. ]\Iinute subquadrate scales of a most 

 peculiar type. At the sides are widely spaced longitudinal circuli, 

 about four in number, and basally and apically are widely spaced 

 radii ; between the basal radii are numerous very irregular but 

 mainly transverse broken lines, which seem to represent the circuli. 

 In the middle of the scale, over a large area, the radii and other 

 markings become very strongly zigzag, producing a very singular pat- 

 tern. I do not know anything like this, unless it is the Rhodeine 

 Cyprinidfe, in which the radii become more or less zigzag. There is 

 a certain suggestion of Umbra, but on comparison it seems to be 

 fallacious. 



This family was formerly placed among the Haplomi, but in Bou- 

 lenger' s work on African fishes follows the Phractolpemida?. The 

 fishes have a Cobitoid apiiearance, but the scales do not resemble 

 those of the Cobitids. 



Chirocentridse. 



Chirocentrus dorah (Forsk.). Cavite, Philippine Islands. Scales alto- 

 gether Clupeoid, differing however from all Clupeidte seen in having 

 the apical field with five transverse circuli, much less dense than 

 those of the basal field, the latter sometimes evanescent. Transverse 

 radii as in the Clupeids. One scale is elongate, the long basal field 

 free from circuli, and witli about twelve wholly longitudinal radii, 

 which do not run to a nucleus, but connect with the apical radii. 

 This must be abnormal. 



