4 6 



ESCAPEMENT 



8 10 12 



( MILLIONS OF FISH ) 



Figure 7.— Kvichak River, 1952-60 (see fig. 2). 



.5 1.0 

 ESCAPEMENT 



1.5 2.0 



(MILLIONS OF FISH) 



2.5 



Figure 8.— Naknek River, 1952-60 (see fig. 2). 



2. Distribution of points for group means 

 should approximate nornnality in horizontal 

 and vertical directions. In some diagrams the 

 small number of points made this approxi- 

 mation rather sketchy, as with means based on 

 two or three points. 



3. The group of points for the right-hand 

 level line must include those for all escape- 

 ments above the selected mininnum. 



A log-log regression of right-hand limb 

 levels on lake area (fig. 12, r = 0.86, P<0.01) 

 revealed a strong relation, which indicates 

 that lake area provides a ceiling to return 

 size, as suggested earlier. 



200 400 600 800 1,000 1,200 



ESCAPEMENT (THOUSANDS OF FISH) 



Figure 9.--Branch or Alagnak River, 1955-60 (see fig. 2). 

 Computation of Returns 



Once the birectilinear regressions had been 

 fitted, the remaining computations were sim- 

 ple. Total returns that corresponded to es- 

 capements for each brood year were read 

 directly from the graphs (figs. 2-11) for the 

 birectilinear curves. These total returns were 

 then allocated among the several return years 

 according to the average age at return for 

 each stream system. Addition of appropriately 

 lagged returns fronn 3 or 4 brood years conn- 

 pleted the hindcast for each year and stream- 

 lake system. 



Individual system hindcasts were summa- 

 rized (table 2) by addition for each year. No 

 regressions were calculated for two groups of 



