330 
PACIFIC SCIENCE, Vol. V, October, 1951 
TABLE 6 
Number of Nehu Larvae per 10-Minute Surface Haul with the 22 cm. Net (Surveys 1, 3, and 4) or 
THE 12.3 CM. Net (Survey 2) according to Survey, Station, and Direction (A and B), and Summary 
SURVEY 1 
SURVEY 2 
SURVEY 3 
SURVEY 4 
MEAN PER CENT 
STATION 
OF SURVEY 
A 
B 
A 
B 
A 
B 
A 
B 
TOTAL* 
1 
3 
1.51 
9 
_ 
_ 
_ 
_ 
_ 
_ 
_ 
3 
1 
1 
1 
_ 
_ 
1 
_ 
7.72 
4 •. . . 
4 
12 
_ 
_ 
8.08 
5 
3 
9 
1 
2 
_ 
1 
_ 
1 
15.84 
6 
8 
4 
2 
_ 
4 
2 
23.69 
9.21 
4.04 
7 
3 
2 
1 
_ 
_ 
8 
4 
4 
_ 
_ 
_ 
_ 
9 
8 
1 
- 
- 
- 
- 
1 
- 
6.63 
10 
1 
2.56 
11 
1 
_ 
_ 
— 
2.56 
12 
13 
14 
— 
— 
1 
- 
- 
“ 
3 
2 
12.98 
15 
_ 
_ 

1 
2.08 
16 
- 
2 
- 
- 
- 
- 
- 
1.01 
17 




— 
— 
— 
— 
— 
18 

_ 





19 
20 
21 
- 
- 
- 
- 
- 
- 
1 
- 
2.08 
22 
_ 

_ 
23 
- 
- 
- 
- 
- 
- 
- 
- 
- 
Totals .... 
31 
35 
7 
6 
- 
1 
11 
5 
99.99 
* Omitting Survey 3. 
variation, i.e., nets towed simultaneously 
seem to yield partially correlated data. 
Before presenting further data to assist in 
the interpretation of the significant SXH 
interaction, the relative efficiency of the nets 
will be considered. The (geometric) mean 
relative efficiency may be calculated from the 
logarithmic data. Thus, for Survey 1, the sum 
of the logarithms of the catches for the 100 
cm. net (A and B hauls) is 42.8000, and for 
the 22 cm. net it is 26.2514. The difference, 
16.5486, divided by the number of paired 
hauls, 18, is 0.9194. The catch ratio, 22 cm. 
net/100 cm. net, is the antilog of -0.9194 
(1.0806), or 0.1204. 
Modifying the formula presented by Win- 
sor and Clarke (1940), the standard error of 
the ratio involving two nets is given, in 
logarithms, by 
N 
where N is the number of paired hauls, and 
ki and k2 are the number of items associated 
with the respective variance components. In 
our data (except that of Survey 3), is 
considered to be 0, rather than a negative 
quantity. The variance components crSHN^ 
and k20-SN^ constitute the mean square for 
the SXN interaction. Thus for Survey 1, the 
standard error of the ratio, in logarithms, is 
- / 1 . 2(0.02157) =0.0490. 
18 
The confidence interval, mean ±2 s.e., may 
be calculated in logarithms and then con- 
verted, giving 0.0961 and 0.1509. 
The following data have been calculated 
for each survey: 
