Egg and Larva Production of Nehu — Tester 
35 
TABLE 1 
Analysis of Variance of (Transformed) Egg and Larva Count over a Two Year Period 
SOURCE OF VARIATION 
DEGREES OF 
FREEDOM 
SUM OF 
SQUARES 
MEAN 
SQUARE 
F 
Egg Count: 
Substations 
2, 
0.6248 
0.3124* 
3.2 
Days 
206 
416.5646 
2.0222** 
21.0 
Error 
412 
39.6087 
0.0961 
— 
Larva Count: 
Substations 
2 
0.0184 
0.0092 
0.2 
Days 
206 
65.9477 
0.3201** 
6.4 
Error 
412 
20.5733 
0.0499 
— 
* P about 0.05; ** P less than 0.01. 
over the 2 year period. In the basic adjusted 
data, the range (and hence the standard de- 
viation) of the substation determinations was 
obviously correlated with the mean, indicat- 
ing the need for a transformation of the data 
in order to apply tests of significance. A 
logarithmic transformation was used: y = 
log (at + 1) (Barnes, 1952: 65); this not only 
tended to decorrelate the standard deviation 
and the mean but also avoided the logarithm 
of zero. 
The transformed data were analyzed ac- 
cording to two criteria of classification, sub- 
stations and days, with the results included in 
Table 1. There are highly significant differ- 
ences between the three geometric means for 
substations: A-6.36, B-7.94, and C-8.80. Ap- 
parently on the average, Substation C was 
located closest to the focus of abundance of 
eggs in the southern sector of the bay (Tester, 
1951: 336). The large and highly significant 
variation between sampling days was anti- 
cipated, and reflects both erratic daily egg 
production and seasonal variation in spawning 
activity. 
Larva counts at Substations A, B, and C 
ranged respectively from 0 to 62, 0 to 51, and 
0 to 57, with grand arithmetic means of 2.66, 
2.18, and 2.44 larvae per 100 cubic meters. 
As shown in Table 1, there were no significant 
differences between the geometric means for 
substations (A-0.97, B-0.92, and C-0.98) 
but, as in the case of the eggs, there were 
highly significant differences between the 
means for days. 
The great decline in numbers between the 
egg and larva stage is worthy of note and will 
be referred to again. For the grand arithmetic 
mean the decline is from 67.8 to 2.4 (96.5 per 
cent) and for the grand geometric mean it is 
from 7.01 to 0.96 (86.3 per cent.) 
Variation with Time 
The logarithms of the adjusted egg counts 
were averaged for each sampling day and were 
plotted against time (the detailed graph is not 
reproduced). It was at once apparent that 
there was considerable variation in mean egg 
count between successive sampling days dur- 
ing certain months. For example, on June 5, 
8, and 12, 1951, the geometric mean counts 
per 100 cubic meters were respectively 10.2, 
607.0, and 21.2; on July 2, 5, and 9, 1951, 
they were 0.8, 537.4, and 288.0. It is possible 
that this large daily variation in egg count 
could be due to shifting of the focus of spawn- 
ing to and away from the vicinity of Station 4. 
However, it seems more likely that it is due 
mostly to large daily variation in actual egg 
production and thus in spawning activity. If 
so, it would be of interest to discover the 
cause or causes. 
Perusal of the data on temperature and 
salinity indicated that these factors per se were 
