Scale samples o - Beginning in 1941-42j scale samples have been sys- 

 tematically collected from ten fishes in each length sample by all of 

 the agencies engaged m sampling,. During three prior seasons similar 

 samples were less regularly obtained by the Fish and Wildlife Service 

 and in these and earlier seasons some scales and otoliths were gathered 

 by the California State Fisheries Laboratory in California and by the 

 Fisheries Research Board in Canadao The utilitj'- of these earlier collec- 

 tions is dubious but is now being surveyed by the Fish and Wildlife Service© 



Analysis 



Catch-per-boat-weeko - The processes described under this heading 

 have been devis'eTTor handling the California statistics for the seasons, 

 1932'!-33-1940-41o Modifications will be necessary for use with data for 

 the Pacific 'Worthwest and for other periods in California. The accepted 

 manner of compiiting the catch-per-boat eliminates two sources of variability 

 that are extraneous to fluctuations in abundance or availability,, namely; 

 variations in number and in the type of size of boats fishing. This is 

 accomplished by comparing a boat's performance during two separate periods. 

 The resulting pairs of relative catches are combined by averaging to repre- 

 sent fleet performance J, and combined by linking to fonn a time series. 

 Because the opportunity to fish is variable, depending on the duration 

 of the moonless portion of the night, the statistical series have been 

 divided into lunar periods and each lunar period into four lunar weeks. 

 Year-to-year ratios are then computed between pairs of homologous lunar 

 weeks. However, this statistic still contains several extraneous sources 

 of variability. One is the varying amount of idleness of boats, assumed 

 to be fishing, but actually in port on account of repairs to machinery 

 or gear and of miscellaneous causes. It has been assumed that this is 

 random in occurrence, and has no trend effecti and lacking any record, 

 no account of it is taken in the computations. Tno other sources of vari- 

 ability are treated below. 



Adjust for efficiency. - There may be changes in efficiency connected 

 with the skill of crews and modifications of gear. As yet no means has 

 been devised of dealing with such variations. 



Adjust for weather. - By computing regressions of average daily catch 

 per boat according to weather conditions a relation between the two is 

 derived and used as an adjustment factor. The effect of variation of 

 available fishing time each night is also incorporated in the regression 

 systemSo 



Extensive experimental analysis of wind movement as recorded by 

 anemometer at such meteorological stations as were available, has led to 

 the discarding of this source of weather data. Instead, the wind movement 

 is deduced from barometric gradients in a manner which gives a representa- 

 tion of conditions over an extensive sea area adjacent to each fishing 

 port. Direction as well as force is correlated with catch, and both enter 

 into the adjustment. Other elements such as fog and precipitation appear 

 to have no statistically consistent effects on the catch, probably because 

 conditions at the obserx'ation points differ too much from those on the 

 fishing grounds. 



8 



