482 
BULLETIN OF THE BUREAU OF FISHERIES 
number of reds caught in Lynn Canal without some relaxation in the intensity of 
fishing in Icy Strait. 
Figure 16 shows the deviations of the catches from the trends in both districts. 
It is evident from this that there has been no marked correlation between the catches 
of red salmon in the two districts, if the entire period from 1902 to 1925 is considered, 
although this might have been expected since the catches are drawn from the same 
runs. There is, however, a definite positive correlation in the last few years of the 
period — from 1912 to 1925 — and a slight negative correlation during the period from 
1902 to 1911. The Pearsonian coefficients of correlation (r) have been calculated for 
these periods and are as follows: (1) For the period 1902-1911, —0.35 ±0.19; for the 
period 1912-1925, +0.68 ± 0.098; and for the entire period 1902-1925, +0.013 + 0.136. 
The positive correlation during the later period is unquestionably significant; the 
correlation during the earlier period is of doubtful significance; while the correlation 
for the entire period has certainly no significance due, obviously, to the combining 
of a positive with a negative correlation. 
The interpretation of these phenomena is difficult; but it is interesting to note 
that the negative correlations occurred during the period in which the catch in Icy 
Strait was increasing while that in Lynn Canal was decreasing and that, after the 
peak of production in Icy Strait had been reached and the catches in both districts 
were decreasing, the correlation became distinctly positive. This is, of course, just 
what might have been expected if the trends had not been eliminated, but since the 
correlations are based on deviations from the trends, the influence of the change in 
the trends from opposed to parallel should have had no effect. Why there should 
have been a negative correlation in the early part of this period is very doubtful and 
we have no explanation to offer. In the case of the more significant positive correla- 
