414 



Fishery Bulletin 97(3), 1999 



B^^'s= the coefficients or "unstandardized 



fianction coefficients"; and 

 C = a constant. 



Initially we intended to follow the 1960s methods 

 of Wilk (1977) by assigning a spring-southern and 

 summer-northern a priori reference designation. We 

 were unable however, to collect a reference group of 

 spring- and summer-spawned yearling bluefish. 

 From the length frequencies of the yearling fish (211- 

 382 mm) that we collected, it appeared that all were 

 "spring-spawned" (Chiarella and Conover, 1990). As 

 a result we made no effort a priori to separate 

 "spring- spawned" from "summer-spawned" yearling 

 fish on the basis of scale back calculations as 

 Chiarella and Conover { 1990) had done but assumed 

 all were spring-spawned. A priori assignment of ref- 

 erence or learning groups is often used to determine 

 the discriminant function which is then used to clas- 

 sify the individuals of known origin to one or the other 

 reference groups. If, however, more than two stocks 

 are present, the individuals of the third stock will be 

 "force fitted" into one of the reference groups. Rather 

 than forcing a pnoW assignments of two groups, and 

 to identify possible additional stocks (from the south 

 IMcBride et al. 1993 1) or morphometrically distinct 



