PACIFIC HERRING 
301 
x 
1 
Of 
E 
sion of sizes due to the growth of dominant year classes, as shown by our sampling, 
is reflected in the commercial catch. 
DOMINANT YEAR CLASSES SHOW RELATIONSHIP OF AREAS 
In nearly every case the age distributions of any two localities differ considerably 
in the proportions of fish of each year class. Thus the age frequencies of Elrington 
Passage show no similarity to those of samples taken farther to the westward, and 
the dominant year classes may well be different. This lack of agreement in most 
cases may be due to different physical conditions in each locality at the time of 
spawning, although the correlation of these dominant age groups with the physical 
conditions is far in the future. While this lack of agreement may possibly be 
wholly environmental in cause, it 
indicates the lack of migration 
between stocks of different locali- 
ties. Indeed, the age frequen- 
cies differ between all localities, 
for which ages have been read, 
which are shown to be racially dis- 
tinct (see Independence of Areas 
p. 272), thus corroborating the evi- 
dence of lack of migration between 
certain stocks set apart by the in- 
vestigation of their structural differ- 
ences. 
There is a slight possibility 
that this lack of agreement between 
the age frequencies in different local- 
ities might be due largely to chance. 
That the reverse is true is strongly 
suggested by a comparison of the 
1926 age frequencies of the sum- 
mer herring of Shuyak Strait and 
KachemakBay. (Fig. 40, Table 32.) 
The Shuyak Strait samples 
were caught on July 15 and the Kachemak Bay' samples from August 25 to 28, 
about six weeks later. Herring were caught in Shuyak Strait for two days following 
the opening of the season on July 15 and then disappeared, so that it is quite pos- 
sible that the Kachemak Bay samples came from the same school. 
Assuming that the sampling was truly random and the age readings correct, a 
mathematical means of testing the significance of the similarity exists, and a measure 
of judging the probability that two frequency distributions are samples of the same 
population may be obtained by the formula: 
EG H AMY BAY 
/926 
V 
— , _ r _, 
Figure 39. 
6 7 
AGE 
-Age histograms of Eshamy Bay, 1926, and McClure Bay, 
1927 
x 2 = N,N 2 S 
| (L-k\\ 
U .nJ \ 
{ jl +/2 I 
(Pearson, 1914) 
where j x is the frequency in a given class of one frequency distribution and f 2 is the 
frequency in the corresponding class of the other frequency distribution. N x and 
