HERRING IN PRINCE WILLIAM SOUND, ALASKA 
271 
As the herring in Prince William Sound spawn and the eggs develop during the 
months from March to June, inclusive, the inference to be drawn is obvious. The 
differences in the vertebral count found between different year classes are probably 
due entirely to environment. Also any significant differences found between herring 
of the same year class taken in different localities in Prince William Sound may be, 
and probably are, simply an expression of environmental differences on the spawning 
grounds. 
From the foregoing it is clear that comparisons between the vertebral counts of 
different localities are valid for showing population differences (in the absence of 
exact knowledge as to the conditions on the spawning grounds) only when the com- 
Figure 5. — Average mean annual air temperatures for the combined months of March, April, May, 
and June for Seward, Cordova, and Latouche. Solid line is from “Climatological Data” of the 
U. S. Weather Bureau. Dotted line (see text) from unpublished data taken by the U. S. Coast and 
Geodetic Survey at Seward only. Circles indicate Seward temperatures; dots, Cordova tempera- 
tures; triangles, Latouche temperatures taken by Weather Bureau 
parisons are between fish of the same year class. It is also clear that the absence of 
significant differences in the vertebral count between samples of the same year class 
does not necessarily indicate that the populations of any two localities are identical. 
Similarity of conditions on the spawning grounds may cause the lack of a significant 
difference between fish from two localities. 
Comparisons of vertebral count distributions of fish of the same year class from 
neighboring localities are given in Table 6. Any two means are compared by divid- 
ing their difference by the standard error estimated by the formula 
cr= j S(x — x) 2 + S(x' — x'y / 1 1 \ 
\ + 77.2 \7li + 1 %2 d" 1 / 
if Xi, x 2 , x ni +i and x' u x' 2 . . . x' n2+ i be two samples, and 
1 o , N 1 
