PACIFIC HERRING 
265 
The maximum variations found between the means of any two samples taken 
different years in the same locality are shown in Table 22. In two cases the differ- 
ences are probably of statistical significance, being 0.25 ±0.062 or 4.0 at Craig and 
0.38 ±0.099 or 3.8 probable errors for Shuyak Strait. As the maximum difference 
found between any two samples from the same locality, 0.38, is of probable statistical 
significance we can not claim as racial any lesser difference. This invalidates, from a 
racial standpoint, four more of the differences that are significant from a statistical 
standpoint (Table 20), leaving only seven differences that can be regarded as having 
possible racial significance. 
Table 22. — Maximum variations between the means of the dorsal-ray counts of any two samples taken 
different years in the same locality 
[Asterisk indicates those of probable statistical significance] 
Locality 
Years 
sampled 
Differ- 
ence 
be- 
tween 
means 
Prob- 
able 
error of 
differ- 
ence 
Differ- 
ence 
divided 
by 
prob- 
able 
error of 
differ- 
ence 
Locality 
Years 
sampled 
Differ- 
ence 
be- 
tween 
means 
Prob- 
able 
error of 
differ- 
ence 
Differ- 
ence 
divided 
by 
prob- 
able 
error of 
differ- 
ence 
1925, 1928 
0. 25 
0. 062 
*4.0 
Kachemak Bay 
1926, 1927 
1925, 1926 
0. 24 
0. 088 
2. 7 
Macleod Harbor 
1927, 1928 
.22 
.091 
2.4 
Shuyak Strait 
.38 
.099 
*3.8 
Returning to Table 20, those differences that are statistically significant and 
exceed the maximum variability found between samples of the same locality are 
shown by footnote 1 reference. It is noted that all of these differences are between 
Yakutat and other localities, Yakutat differing significantly from all except Dogfish 
Bay. None of these differences are excessive, and we are inclined to doubt seriously 
their racial validity, for the Yakutat sample, besides being small in numbers, was 
composed entirely of very small fish (64 to 85 millimeters in length) which might 
introduce two sources of error, one being due to counting the rays of such exceedingly 
small fish in a different manner than those of larger specimens (a low power of the 
microscope was used), and another to the fact that they were all of one year class, 
which as we have seen in the case of the vertebral count, might easily account for 
large differences between the means. 
In conclusion it may be said that the dorsal-ray count has not shown the dis- 
tinctness of any populations. 
ANAL RAYS 
The anal-ray data contains 1,175 counts of fresh specimens (Table 23). In 
making the anal-ray count, doubt often arises owing to the diminutive size of the 
first, and often the second, unbranched ray. For this reason we have not compared 
our counts with those made by Thompson (1917) or Hubbs (1925), as there is no 
means of correctly evaluating any personal error that may have arisen as a result of 
this difficulty. 
