The length frequency tables are derived primarily from data collected 

 by the California Department of Fish and Game- The tonnages include all 

 records of sardine deliveries to shore plants, and during years when 

 deliveries were made to floaters off San Francisco these are included. 

 The catches of the San Francisco floater fleet were compiled by the U. S. 

 Fish and Wildlife Service. (Anonymous, 19U8=) 



The problem of sampling populations of migratory fish is not simple. 

 Sette (192U, p» 70) reported results of analysis of data on length compo- 

 sition of catches taken during the first season of the sardine investiga- 

 tion, the winter of 1919-20, on which the sampling program in subsequent 

 years largely is based. He summarizes (p. 90) "it may confidently be 

 assumed that the choice of a sample from the boat load of sardines is 

 truly random . . . the catch of sardines made by the cast of the net into 

 or around a school of sardines is truly random except that schools of 

 fish composed mainly of individuals under eighteen centimeters in length 

 are avoided; but it may not be assumed that the assemblage of sardines in 

 a school is truly random . „ , With the sardines carrying on their own 

 selection into schools, it cannot be hoped that validity of sampling may 

 be determined on the basis of established statistical methods based on 

 the random selection of individuals. This need not, however, preclude 

 the investigation of the sardine by representative sampling." 



Sette concluded (p. 96) "That to secure a representation sufficient 

 to show the location of a distinct mode to within 10 percent of the 

 probable movement of that mode within one year, a sampling of twice a 

 week is necessary." He also concluded that this frequency of sampling 

 was necessary "to determine the relative importance of fairly well 

 defined and large size-groups within 10 percent error." 



Thompson (1926) further emphasized errors of sampling and of selec- 

 tion, and forecast a future need for weighting the samples to the catch. 

 He concluded (p. 188), "It is, from the point of view of the future 

 program of great interest to define the most adequate and representative 

 system of sampling. We have been unable to find any radical need for 

 change from the present method except . . . the possibility of weighting 

 each sample by the amount of the commercial catch it represents." 



In the light of subsequent years of study of the fishery, it appears 

 that the behavior of sardines, as reflected in their availability to the 

 fishermen, and the variable contribution of different components of the 

 stock„ may magnify the difficulties of representative sampling. 



The California Department of Fish and Game has carried on a program 

 of sampling the commercial catch since 1919 ° The intensity of sampling 

 has varied from year to year and from port to port. When sampling was 

 first started at Monterey and San Pedro (in 1919), samples were taken 

 daily from five boats. -Mater, semiweekly samples were obtained from five 

 boats. A sample consisted of either U0 or 3>0 fish. In later years, 

 samples have been taken from five boats daily, semiweekly, orweekly, all 

 samples consisting of $0 fish. 



