68 
PACIFIC SCIENCE, Vol. VI, January, 1952 
expected, as Ala Wai Canal is a narrow inlet 
with a small opening to the sea and with a 
relatively large stream entering near its head, 
whereas Kaneohe Bay is a large body of 
water with a wide exposure to the sea and 
with a relatively small volume of fresh water 
entering it from a few streams. As the average 
difference in temperature between localities 
(0.7°C.) is much less than the average dif- 
ference between summer and winter tempera- 
tures within localities (about 3° C.), it would 
appear that temperature alone could not 
account for the large difference in mean count 
between Ala Wai Canal (41.539) and Kaneohe 
Bay (42.639), although it is in the direction 
which might be expected. 
Differences in mean vertebral count which 
are apparently related to differences in salin- 
ity have been reported by Hubbs (1925) for 
the Pacific Coast anchovy {Engraulis mordax), 
as noted above, and also by Blackburn (1950) 
for the Australian anchovy {Engraulis aus- 
tralis). In the former, the fish in the brackish 
waters of San Francisco Bay had a much 
lower mean count (43.80) than those near 
the entrance of the bay and seaward from the 
entrance (45.73). Similarly, in the latter, the 
fish in the brackish waters of the rivers of 
Gippsland lakes also had a lower mean count 
(44.56) than those near the entrance to the 
sea (45.20). Thus, the strikingly different 
mean count for the Ala Wai Canal fish is 
probably related in large part to the brackish- 
water habitat. 
Inspection of Table 2 shows that not only 
is the mean for Ala Wai Canal (41.539) out- 
standingly low, but also that the means for 
other localities on the leeward shore of Oahu 
— Pearl Harbor (42.372) and Honolulu Har- 
bor (42.548) — are lower than that of Kaneohe 
Bay (42.639) on the windward shore. One 
might be tempted to the conclusion that there 
are also distinct populations in these other 
localities. Consideration of their relative 
sizes, extent of exposure to the open sea, 
fresh-water supply, etc., would suggest that 
Oahu localities range in the following order 
of increasing average salinity: Ala Wai Canal, 
Pearl Harbor, Honolulu Harbor, Kaneohe 
Bay. This order conforms with progressively 
increasing mean count: 41.539, 42.372, 
42.548, 42.639. However, the fiducial inter- 
vals of the means for all localities other than 
Ala Wai Canal overlap considerably. There- 
fore, they could have arisen in random samp- 
ling from one complex population. 
The mean vertebral counts for localities on 
the other islands also could, with some specu- 
lation, be fitted into the above gradation. 
Certainly, in the absence of more precise 
information, Kawaihae Bay, with the highest 
mean count (42.743), would be placed at the 
opposite end of the series to Ala Wai Canal, 
as it is open to the sea and has only one or 
two very small streams flowing into it which 
are dry during part of the year. Again, how- 
ever, there is no statistical evidence for the 
segregation of the mean counts of these other 
localities. 
The lack of significant differences between 
the means for localities other than Ala Wai 
Canal is more clearly demonstrated in the 
analysis of variance which follows. In this, 
the Ala Wai Canal data have been omitted. 
It will be observed that there is no longer 
significant heterogeneity between localities: 
SOURCE OF 
VARIATION 
DEGREES 
OF 
FREEDOM 
SUM OF 
SQUARES 
MEAN 
SQUARE 
F 
Islands 
2 
5.9321 
2.9661 
Localities 
3 
15.0686 
5.0229 
1.50 
Samples 
8 
26.8484 
3.3560 
9.78** 
Individuals 
2600 
891.7377 
0.3430 
** Highly significant; P is 0.01 or less. 
In the foregoing analysis, an additional 
category — islands — has been incorporated. 
It will be observed that there are no signi- 
ficant differences between the means for 
islands (Table 2). In fact, the mean square for 
islands is considerably less than might have 
been anticipated. This may be due to the 
small number of localities which were 
