GROWTH AND MATURITY OF SALMON IN THE OCEAN 
59 
An examination of the tables and text figures given above shows clearly that 
in all of the age groups represented among the fish taken in the ocean a marked 
increase in length occurs during the fishing season, from May to September, in- 
clusive. Similar increases in length are shown in five of the age groups taken in- 
side the river, but in three of the age groups — the 2-year and the 5-year fish with 
ocean nuclei and the 6-year fish with stream nuclei — the trend is reversed and the 
fi.sh belonging to these age groups are smaller toward the end of the season than 
earlier. It is also noticeable that, even excluding these age groups in which the 
trend is reversed, the trends for the fish taken inside are more variable than for 
the fish taken outside. The average daily increment (b in Table 12) for the fish 
taken outside varies from 0.0888 to 0.1410 cm., while that for the groups taken 
inside, omitting those age groups in which the trend is downward, is much greater, 
ranging from 0.0671 to 0.1989. This is over two and one-half times that shown 
by the fish taken outside. Attention has also been called repeatedly to the fact 
that in each age group the deviations of the average lengths for each collection 
(shown as the points in the figures) from the trend (as determined from the series 
of collections taken over the whole season) is greater among the fish taken in the 
river than among those taken in the ocean. All of the evidence available indicates 
a decidedly greater variability among the fish taken inside. A few instances of this 
sort might be attributed to chance, but the uniformity of the evidence demands 
some other explanation. 
It seems very probable, if not, indeed, certain, that this greater variability in 
the rate of increase in length among the fish taken in the river is due to the fact 
(which has been mentioned above) that there are more profound changes in the 
racial constitution of the fish foiuid migrating up the river than in those found 
schooled outside, and that these different races vary sufficiently in their average 
size so that the effect of growth may be masked more or less completely by these 
racial changes. Evidence has been presented which shows that both of these 
factors are present in the migrating fish within the river. It seems hardly neces- 
sary to argue that there is a succession of races during the season among the mi- 
grating fish. Granting the existence of such races it follows that there must be a 
succession of them in the river during the season, so that at one time the run con- 
tains a conspicuously large percentage of fish from one or at most a few races, 
while at other times other races will predominate. This would be particularly 
true in the case of such a large river system as the Columbia with its numerous 
tributaries, which probably have, as Gilbert has shown for the sockeye salmon, 
each its individual race. It has been shown in one specific instance — that of the 
4-year fish with ocean nuclei taken during August and September, 1919 — that 
distinct races are to be found among the chinook salmon of the Columbia River, 
and, moreover, that these races may vary sufficiently in size so that the predominance 
of one of them at one time and another at another time would undoubtedly greatly 
modify the true effect of growth. 
The reversed, downward trend of the fish belonging to some of the age groups — 
the fish in their second and fifth years with ocean nuclei and those in their sixth 
year with stream nuclei — may be explained, then, by supposing that the races that 
predominate in these age groups early in the season are large and those running 
