294 
BULLETIN OF THE BUREAU OF FISHERIES 
Sette (1926) made a study of this sampling problem with the California sardine 
(. Sardina cserulea ) at Monterey. He obtained a very complete series of samples 
throughout the season and made the assumption that these represented the commer- 
cial catch. He then endeavored to discover by mathematical means how frequently 
samples need to be taken in order to represent the series with reasonable accuracy. 
He found that by taking samples at least semiweeldy the desired result could be 
obtained. 
No such series of samples is available for the herring. The herring “runs” are 
more erratic than those of the sardine, because in many localities the herring can be 
sampled for only two or three months out of the year, and the total catch of the year 
may be made on as few as 10 different day^s. Indeed, in Red Fox Bay, Shuyak Strait, 
in 1926, the whole catch of about 13,000 barrels was taken in two nights’ fishing. 
Sampling twice a week might be theoretically correct, but under such conditions it is 
impossible from a practical standpoint, except perhaps for the duration of a “run.” 
With the herring it is highly probable that in each locality a much smaller popula- 
tion is sampled than with the sardine. So far but very little evidence has been found 
to show the existence of local races in the California sardine (Hubbs, 1925, p. 12), 
while studies have shown the Pacific herring to be divided into many local races. 
The catch landed in a given port may come from several of these, making the sampling 
of the herring much more complex. 
These facts tend to show that in the case of the Pacific herring sampling can not 
be made as exactly as in that of the California sardine, so that in order to prove the 
validity of our sampling a more complete dependence must be placed upon the 
repeated occurrence of consistent variations (Thompson, 1926a). The inability to 
obtain samples over a long period during each season will make it less likely that our 
samples will represent the same portion of the population year after year, thus decreas- 
ing our chances of proving the consistency in occurrence of any variations and making 
it more difficult to foretell in advance what fluctuations in abundance may" be expected. 
Even if it were true that the commercial catch is taken each year from the same 
section of the population, it might be difficult to demonstrate clearly the exact 
manner in which dominant year classes would come into evidence. Thus in the 
sardine of California the dominant year classes affect the commercial take unequally 
at different sizes and ages, and prophecy of the catch is thereby limited. (Thompson, 
1926a.) 
EVIDENCE OF DOMINANT YEAR CLASSES 
In order to prove the existence or nonexistence of dominant year classes, samples 
are needed over a period of years from one locality. (Table 31.) The longest period 
for which consecutive samples are available extends from 1924 to 1927. These were 
taken in, or in the immediate vicinity of, Elrington Passage, one of four channels 
connecting the western side of Prince William Sound with the open ocean. All of the 
data for 1924 and a portion of those for 1925 consist only of length measurements, 
which were obtained from Clarence L. Anderson, a former technologist of the United 
States Bureau of Fisheries. Ages are available for a portion of the 1925 data, and 
for most of those of 1926 and 1927. 
