Cockerell: Scales of Ciiaracinid Fishes. 105 



scales accords witli tho shape of the fish, which is not nearly so 



deep-bodied as is usual in the genus. 

 Ilemigrammiis unilineatns Gill. Scales about one and two-thirds 



broad; radii usually four; nuclear sculpture as usual; radial area 



longitudinally striatulate. 

 Hemigrammus occUijer Steindachner. Scales about one and two-thirds 



mm. broad, very broad for their length; radii about four to six; 



nuclear sculpture distinct. 



Ilyphessobrycon Durbin. 



The scales are entirely of the Hemigrammus type. 

 Ilyphessobrycon gracilis (Reinhardt). Scales about one mm. broad; 



ladii usually four; only about three circuli basad of the nucleus, the 



circuli failing in the middle of the base, instead of becoming crowded 



as in most scales. 

 Ilyphessobrycon rosaceus Durbin (cotype). Scales hardly one mm. 



broad, but very much broader in jM'oportion to their length than 



those of //. gracilis; circuli very few and widely sjiaced; radii two, 



rarely four. 

 Hyphessobrycon cos Durbin (cotype). Scales one and one-half to 



one and two-thirds mm. broad, shape as in //. gracilis; radii six to 



nine. 

 Hyphessobrycon stictus Durbin (cotype). Scales about one mm. broad, 



shape nearly as in //. gracilis; radii usually two, sometimes more. 



In all these fishes, the deepening of the body is accompanied by a 



widening of the scales, rather than an increase in the number of rows. 



Pristella Eigenmann. 



Scales also of the Hemigrammus- Cheirodon type. 



Pristella riddlei (Meek). Scales very broad for their length; breadth 

 about one and three-fourths mm.; radii usually four; nuclear 

 sculpture weak. 



Pristella aubynei Eigenmann. Scales not nearly so broad in propor- 

 tion to length; radii usually five; nuclear area broadly leticulated, 

 but the network more or less broken. 



Moenkhausia Eigenmann. 

 In this genus the circuli appear to be basal and lateral only, as in 

 the Hemigrammus series, hnt the radial field is always minutely longi- 



