560 



Fishery Bulletin 89(4), 1991 



and 



5.09 



kcal 



g dry food 



21.6g dry 

 lOOg wet food 



= 1.1 



kcal 



g wet food 



Growth in length 



Final mean length was significantly higher than the in- 

 itial mean length in all feeding treatments except the 

 low-ration, fast-speed treatment (£-test, P<0.05; Table 

 2). However, the slopes of the regressions describing 

 the apparent rates of growth (in length) were not sig- 

 nificantly different from zero, because of the variation 



not explained by growth or selective mortality. The fish 

 schooling in the flume may have sorted themselves by 

 size so that the dipnet frequently caught a sample con- 

 taining larger- or smaller-than-average fish. The coef- 

 ficient of variation of length in each sample ranged 

 from 5 to 7%. The low number of samples would make 

 this an important source of error. 



