66 
PACIFIC SCIENCE, VoL VI, January, 1952 
length groups but which includes all the 
data rather than only some of them, was 
undertaken with the following results: 
SOURCE OF 
VARIATION 
DEGREES 
OF 
FREEDOM 
SUM OF 
SQUARES 
MEAN 
SQUARE 
F 
Localities 
6 
520.2088 
86.7015 
31.97** 
Samples 
11 
29.8275 
2.7116 
7.61** 
Individuals 
3099 
1103.7526 
0.3562 
** Highly significant; P is 0.01 or less. 
As in the previous analysis, it is apparent 
that there are significant differences between 
the means for samples from the same locality 
and between the means for localities. The 
various means are tabulated in Table 2. 
Heterogeneity between the means of 
samples from the same locality arises chiefly 
from the material taken at Honolulu Harbor 
and Kaneohe Bay. In the former, the mean of 
one sample (42.806) differs significantly from 
those of the other two (42.352 and 42.449). 
In the latter, the mean of one sample (42.568) 
also differs significantly from those of the 
other two (42.754 and 42.707). Differences 
such as these may represent: (1) the presence 
of two or more genetically separate stocks 
within a locality; (2) the mixture of stocks 
between localities; or (3) seasonal variation 
in environmental conditions within a locality, 
the stock remaining discrete. 
Regarding (1), although the possibility of 
genetically separate stocks occurring within 
a locality cannot be ruled out, it seems un- 
likely that these would occur in fish, such as 
the nehu, which have pelagic eggs, and which 
spawn in partially enclosed waters with good 
circulation and mixing. Regarding (2), the 
possibility of schools migrating from one 
baiting area to another cannot be denied. 
However, as will be brought out in the dis- 
cussion, there is little evidence that nehu 
schools occur commonly in offshore waters. 
Regarding (3), there is a difference of about 
3°C. between summer and winter water tem- 
peratures within a locality. This, or seasonal 
differences in salinity, or a combination of 
both, may be sufficient to induce the dif- 
ferences in mean count between samples 
within localities. 
The analysis-of-variance tables show that 
there is highly significant heterogeneity be- 
tween localities, allowing for the variation 
between samples from the same locality. In- 
spection of the data (Table 2) indicates that 
the heterogeneity stems mostly from material 
for the island of Oahu. The mean for Ala Wai 
Canal (41.539) is outstandingly low as com- 
pared with the means for all other localities. 
To determine whether it differs significantly 
from those of other Oahu localities, the mean 
square for "samples” was accepted as a 
common variance (Snedecor, 1948: 10.4), 
thus leading to a calculation of the fiducial 
interval (P = 0.05) of a locality mean, which 
amounts to ± 0.279* The upper "limit” for 
Ala Wai Canal (41.818) is much less than the 
lower "limit” for Pearl Harbor (42.093), the 
locality with the closest mean count. It seems 
certain that the Ala Wai Canal fish constitute 
a distinct statistical group in respect to verte- 
bral number, and that they mix little, if at all, 
with the fish of other localities. 
The striking difference between the mean 
vertebral number of Ala Wai Canal nehu and 
those of other localities may be due to genetic 
factors, environmental factors, or both. The 
probability that the difference is of genetic 
origin could be investigated only by con- 
ducting rearing experiments on Ala Wai 
Canal fish and those of other areas under 
controlled conditions. The possibility that 
the difference is of environmental origin may 
be investigated with such data as are on hand. 
Surface temperatures and chlorinities over the 
period February to June, 1949, are available 
for three stations in both Ala Wai Canal and 
Kaneohe Bay. Average determinations, made 
within 1 or 2 days of each other, are shown in 
Table 3* During this period, Ala Wai Canal 
had a temperature which averaged about 
0.7°C. higher and a chlorinity which averaged 
about 2.4 ppm lower than Kaneohe Bay. 
These, or even greater differences, are to be 
