590 



40 SO 80 100 120 140 



ESCAPEMENT (THOUSANDS OF FISH) 



Figure 4. --Snake River, 1950-60 (see fig. 2). 



(1963);° and Ossiander, 1964,' 1965/ 1966 

 (footnote 1), 1967 (footnote 2). 



Birectilinear Regressions 



Regression diagranns were plotted for 10 of 

 the stream-lake systems tributary to Bristol 

 Bay, including all those with major sockeye 

 salmon runs (figs. 2-11). The distribution of 

 points on several of these figures suggests 

 conformity in sonie degree to the birectilinear 

 regression surface postulated above and il- 

 lustrated for the Naknek River (fig, 1). Such 

 regressions were therefore approximated by 

 two straight lines for each system, wherein 

 the ascending left-hand line was defined by 

 three points: the origin and two group means. 

 The line, fitted by inspection, was made to 

 pass through the origin and to have about equal 

 deviations from the two group means. The 

 right-hand level line was set at the mean of 

 the group of points to the right of those in- 

 cluded in the two group means (for the Kvichak 

 River, the single point I960 determined the 



Talt, Howard D. (editor). 1963, Forecast of Bristol 

 Bay red salmon run In 1963. Alaska Dep. Fish Game, 

 Juneau, Inform. Leafl. 23, 13 + [9] pp. (Processed,) 



'' Ossiander, Frank J, (editor). 1964, Bristol Bay red 

 salmon forecast of run for 1964, Alaska Dep. Fish Game, 

 Juneau, Inform. Leafl, 39, 43 pp. (Processed,) 



Ossiander, Frank J. (editor). 1965, Bristol Bay red 

 salmon forecast of run for 1965, Alaska Dep, Fish Game, 

 Juneau, Inform. Leafl. 59, 22 pp. (Processed.) 



500 



400 - 



a: 200 



100 200 300 400 500 

 ESCAPEMENT (THOUSANDS OF FISH) 



Figure 5, — Nuyakuk River, 1946-60 (see fig, 2), 



spoo 



500 1,000 1,500 2.000 



ESCAPEMENT ( THOUSANDS OF FISH ) 



Figure 6.— Wood River, 1950-60 (see fig, 2). 



2,500 



level of the right-hand line). Some of the points 

 selected for the right-hand and left-hand limbs 

 overlapped. In figures 2 to 11, points of sepa- 

 ration between group nneans are indicated by 

 vertical broken lines and overlaps by cross- 

 hatched areas. 



Although selection of the points for the group 

 means was somewhat subjective, lattempted to 

 reduce the subjectivity by setting up the fol- 

 lowing criteria: 



1 . No overlap of escapement ranges for the 

 two left-hand group means. 



