262 
BULLETIN OF THE BUREAU OF FISHERIES 
0.19 and 0.23, respectively. Although these actual differences are less than the 
maximum actual difference, 0.27, found between samples in the same locality, yet 
there is good reason for considering them to be valid. The Shuyak Strait mean is 
derived from nine samples taken over four years and representing a number of year 
classes, so that it should not be expected to vary as widely as the mean of a single 
sample. The Old Harbor and Shearwater Bay samples, from practically the same 
locality, containing 115 and 165 specimens, respectively, differ only by 0.04, although 
taken two years apart, in 1926 and 1928. This would attest to the reliability 
of their means. Hence we have provisionally considered Shuyak Strait and Old 
Harbor-Shearwater Bay to be racially distinct. 
In western Alaska (Table 7) the Shumagin Islands, Unalaska, and Golovin Bay 
have already been shown to differ by amounts well in excess of any differences found 
in the same locality. The mean of the Chignik sample (Table 12) is lower than that 
of the Shumagin Islands, only 90 miles distant, by 1.34 ± 0.060, or 22 probable errors. 
This also is considerably in excess of any variations found in the same locality, so 
Chignik may be considered racially distinct. 
SUMMARY OF VERTEBRAL COUNT FINDINGS 
1. Following the general trend of the coast northward and westward from San 
Diego the means of the vertebral counts increase with the distance, the general 
trend being practically linear and widely departed from only by the herring of the 
Shumagin Islands and Golovin Bay. 
2. Statistically significant differences are found between the vertebral count of 
different age classes from the same locality. 
3. Successive samples from the same locality show differences larger than any 
assignable to chance. 
4. In practically every case distant localities show differences greatly in excess 
of any that can be assigned to variations within the same population, as determined 
by the maximum variation between the means of samples from the same locality. 
5. Many closely adjacent localities also show differences in excess of any assign- 
able to variation in the same locality. 
6. The analysis of the means of the vertebral counts indicates the distinctness 
of the populations in the following areas from the stocks of other areas sampled: 
California, southern British Columbia, Craig, Chatham Strait, Stephens Passage, 
Cook Inlet-Shuyak Strait-Prince William Sound, Shearwater Bay-Old Harbor, 
Chignik, Shumagin Islands, Unalaska, and Golovin Bay. 
DORSAL RAYS 
We have for comparison 3,864 dorsal-ray counts distributed from Puget Sound 
to Unalaska (Table 19). To avoid all errors in counting due to preservation, we 
have only used counts made on fresh specimens. Counts are available from other 
investigators but we do not care to use them in comparison with our own due to the 
chances for personal error in counting. 
