RACES OF HERRING, SOUTHEASTERN ALASKA 
129 
Table 4. — Frequency of variances of 162 individual samples of the vertebral count from various 
localities in southeastern Alaska 
Variance 
Number 
Variance 
Number 
Variance 
Number 
Variance 
Number 
0.1900-0.2300 .. . 
1 
0. 5600-0. 6300 ... 
19 
0. 9600-1 . 0300 
1.3600-1. 4300 
l 2 
0. 2400-0. 3100 
16 
0. 6400-0. 7100 
15 
1. 0100-1. 1100 
1. 4400-1.5100 
i 1 
0 3200-0. 3900 
25 
0. 7200-0. 7900 
9 
1. 1200-1. 1900 
1. 5200-1. 6000 
0. 4000-0. 4700 
38 
0 8000-0. 8700 
3 
1.2000-1. 2700 
1 1 
0. 4800-0. 5500 
30 
0. 8800-0. 9500 
2 
l. 2800-1. 3500 
i These samples omitted in analysis of variance. 
HOMOGENEITY OF MATERIAL FROM INDIVIDUAL LOCALITIES 
Evidence tending to prove a lack of admixture in the samples from the individual 
localities is shown in table 5, in which a test of the homogeneity of samples of the 
vertebral count of the 1926 year class has been made for each of the 7 localities from 
which sufficient data were available for such a test. In contrast to the results ob- 
tained when the southeastern Alaska material was considered as a whole, none of 
the 7 localities have an observed z exceeding the value of the calculated z at a prob- 
ability of 0.05, indicating the homogeneity of the population sampled in each 
locality. 
However, the observed z for the Warren Island samples, 0.2768, is only slightly 
less than that of the calculated, 0.3448, for a probability of 0.05. Therefore the 
Warren Island samples must be regarded with some suspicion, especially as the 
proportion the observed 2 forms of the calculated z is larger in the Warren Island 
samples than in those from the other localities. 
Table 5. — Analysis of variance of vertebral count samples of the 1926 year class from various localities 
in southeastern Alaska taken from 1929 to 1931, inclusive 
Number of 
samples 
Mean square 
Calculated 
z for prob 
ability of 
0.05 ‘ 
Locality 
Mean 
Number 
Between 
arrays 
Within 
arrays 
Observed z 
52. 423 
2, 143 
411 
60 
0. 6465 
0. 5247 
0. 1044 
0. 1414 
52. 421 
19 
.4215 
.4862 
.0714 
.3279 
52. 457 
449 
12 
.3828 
.4467 
.0785 
.4409 
52. 357 
367 
8 
.8397 
.4829 
.2768 
.3448 
62. 471 
189 
6 
.6426 
.4474 
. 1810 
.4052 
52. 188 
186 
9 
.5262 
.4756 
.0507 
.3322 
52. 378 
389 
n 
.6977 
.5304 
.1370 
.3054 
i The observed value of z is less than the calculated value at a probability of 0.05 in every case, thus showing that the popula- 
tion of each locality is homogeneous. 
SEGREGATION OF RACES 
Since the above evidence supports the hypothesis that the population of each 
locality is homogeneous, the vertebral counts from each locality have been compared 
to those of adjacent localities. (See fig. 3 and tables 3 and 6.) Only counts of fish 
of the same year class have been compared, necessitating the limiting of samples to 
those containing 50 or more counts. Smaller samples were not used, owing to the 
probability that occasional errors enter into our age determinations. 
