for all sampling days at all ports the ordinates for each size-category 

 which most nearly represent homologous levels of availability© The method 

 i^akes its form from the idea that each size^category is present in varying 

 degrees of availability to the fishery from time to time and from place 

 to place throughout the extent of the fishing grounds,, If then, the weighted 

 sample-day freatiencies correctly portray the availability of each sizs^ 

 category;, then^ for each size-category a frequency distribution of ordinate 

 heights will describe the number of times this size=category occurs at 

 various levels of avai lability o 



The problem then is one of selecting from these ordinate-height fre- 

 (fancies the ordinate classes representing homologous levels of availa- 

 bilityo Theoretically, it would be desirable to have a selection that 

 would represent maximum ( or 100 percent) availability of each size-cate- 

 goryo This would require selection of the maximum ordinate among the 

 whole collection of ordinates for a given size-categoryo But the ordinate 

 heights contain two important sources of variability apart from availability g 

 errors introduced by weighting factors and those introduced by the random 

 departures of mumbers in each size class from the true number in the popu= 

 lation from which drawno Therefore the maximum ordinate in many instances 

 wouldj by chance, be above rather than at the maximum level of true avails 

 abilityo To avoid this it has been decTded tentatively to take the ordinate 

 class at the ninth decile positiono In other words, we would derive a 

 frequency curve which represents the 90 percent level of apparent avail- 

 ability of each sizeo For the reasons just stated, this leered of apparent 

 availability may be very close to the 100 percent level of real availabilityo 



The first plan requires that availability of each size-category have 

 the same pattern from year to year, the second requires merely that at 

 one time or another during the season the levels of availability of the 

 several size classes approach their maximao It appears that the latter 

 involves a more conservative approach. It is to be expected that knowledge 

 of aggregation habits, leading to the classification of school-groups, 

 will provide important modifications in, or appraisals of, the proposed 

 methods o 



Determine length frequencies of year-classes » - Assuming that one 

 plan or another has produced a properly weighted combination of samples 

 to represent the stock in the sea, there remains the problem of converting 

 each season's length-frequency, into an age-frequency, dis tributiono 

 Its solution, for the years prior to the beginning of routine scale sampl- 

 ing in 1941, involves the use of certain statistical constants derived 

 from scale readings, 



i 



By this is meant determination of the paremeters of the length-fre= 

 quencies of single age-groups or year classes that will define the shape 

 of the frequency curves in sufficiently general terms to be applicable 

 to years other than that in which the scales are collectedo So far the 

 only parameter we have worked with is the standard deviation, as a fanc" 

 tion of age, assigning each age group to have a normal distribution of 

 lengths. While apparently reliable results are obtained, there is e'.7ldence 



12 







