472 
PACIFIC SCIENCE, Vol. IX, October, 1955 
first place the several small samples available 
from Midway vary considerably in average 
count. What this means is not clear, but it 
has prevented me from placing too much 
faith in the results of single larger samples 
from elsewhere. If one were to exclude the 
samples from the low Hawaiian Islands (Mid- 
way, Lisianski, and French Frigate) one would 
obtain a correlation between increase in pec- 
toral count and decrease in water temperature 
for the species complex. On the other hand, 
if one considered the samples of Kuhlia sand - 
vicensis alone, one would obtain a correlation 
in the reverse direction. An attempt to explain 
these contradictory trends will be made after 
consideration of the gill-raker data. Mean- 
while, it may be noted that the Johnston 
fishes appear to be of almost pure Central 
Pacific stock. 
Before proceeding with the gill-raker in- 
formation, three features should be noted. In 
the first place, as Schultz (1943: 99) has ob- 
served, the young of Kuhlia have fewer gill 
rakers than the adult. By plotting the number 
of gill rakers against standard length in certain 
large Hawaiian samples, it was determined 
that K sandvicensis obtains approximately its 
full gill-raker complement by about 40 milli- 
meters in standard length. Consequently no 
fish smaller than 40 millimeters were used in 
the data which follow. The second point 
regards the method of making counts. In 
Kuhlia the most anterior one of two rakers 
frequently taper to almost nothing, and it 
seems preferable to count only the developed 
gill rakers. Here, then, only the pectinate 
rakers are counted; the shorter, nob-topped 
rakers are omitted. Even this restriction leaves 
some specimens in which the count remains 
somewhat questionable. To check the con- 
sistency in my own counting, the gill rakers 
in a sample of 37 fish were recounted at the 
end of a year. The original count gave an 
average of 24.41, the later count, 24.57. These 
and all other counts used here are only those 
below the angle (excluding the raker at the 
angle). 
TABLE 5 
Total Pectoral Ray Counts in Specimens of the 
Kuhlia sandvicensis-marginata Complex 
SPECIES AND LOCALITY 
NUMBER OF 
SPECIMENS 
AVERAGE 
COUNT 
K. sandvicensis 
Hawaiian Islands 
Midway 
(Mar., 1949) .... 
7 
14.57 
Midway 
(July, 1949) 
12 
14.91 
Midway 
(June, 1950) 
6 
14.33 
Midway 
(May, 1951) 
3 
15.00 
Midway 
(all specimens) . . 
28 
14.72 
Lisianski 
3 
14.33 
French Frigate 
5 
14.60 
Kauai 
10 
14.80 
Oahu 
25 
14.76 
Oahu (Schultz) .... 
17 
15.00 
Hawaii 
22 
15.00 
K. marginata 
Johnston 
9 
14.00 
Line Is.: Palmyra .... 
2 
14.00 
Wake 
20 
14.00 
Marshalls (Schultz) . . 
10 
13.71 
Pectinate gill rakers on the lower limb of 
the first arch range from 22 to 28 in my 
samples of K. sandvicensis , from 24 to 29 in the 
smaller numbers of K marginata examined. It 
seems well to discuss the K sandvicensis sam- 
ples first, as these are both the largest and 
the most numerous. In the first place it seems 
as if those samples containing small fish have 
somewhat lower average counts than the 
samples with large fish (Table 6). Perhaps the 
gill raker number does continue to increase 
above 40 mm. 
With this in mind, the pairs of samples 
from the same island may be compared. The 
two Oahu samples were taken in different 
years from exactly the same tide pool. The 
difference between the means of these two 
samples is 0.41. The two samples from Ha- 
waii are of an entirely different sort. One was 
taken from fresh water, the other from the 
sea. The difference between the means of 
these samples is 1.50 gill rakers, though part 
