(3) List the lobster buyers that had five or 

 more boats fishing for them as the second- 

 ary sampling units. From this list we 

 compiled a dealer-code which represents 

 the county and the number assigned to 

 each dealer in that county (Fig. 3). 



(4) Interview all of the lobstermen who de- 

 livered their catches on sample-days 

 during the period of maximum landings, 

 12:00 noon to 5:00 p.m. In addition, we 

 selected a random cluster of 10 lobsters 

 per boat. 



It was only possible to sample 10 days a 

 month due to commitments to entirely different 

 jobs still within this project. 



From the interviews and the cluster samples, 

 we compiled the following information by boat: 



(1) Total catch in pounds 



(2) Total catch in numbers 



(3) Total hours expended for catch 



(4) Total number of traps hauled for the day 



(5) Number of days set-over for the hauled 

 traps 



(6) Total number of traps-set-out 



(7) From the cluster of 10 lobsters, we made 

 individual determinations of: 



(a) carapace length (millimeters) 



(b) weight (grams) 



(c) sex 



(d) cull or normal 



(e) shedder or hard-shell. 



The estimates and variances were calculated 

 using these formulas: 



(1) Xj 



sjtyyy 





* N ij 





I,{Xj — X. 



) 2 



(2) »{*■) '- n(n-l) 



Probability sampling also enabled us to make 

 unbiased estimates in the prescribed categories 

 on a monthly and yearly basis for all of the 

 dealer-days in the survey. In this case, the 

 estimates and variances were calculated using 

 these formulas: 



(1) x hij = x hijk '■> 

 U 



(2) x hi 



(3) x h 



yt ^ x hij 



N 



n 



Sx/j; ; 



- W2 



(4) v h = N 



Sx : 



hi 



P x hi) 2 

 n 





n(r 



i-l) 



k = 1, 





, m 



J = 1, 





, M 



i = 1, 





, " 



h = 1, 



2 





This information was tabulated in a format 

 of five boats per sample sheet (Fig. 4), then 

 these sheets were summarized by day (Fig. 5), 

 month, and year with an additional calculation 

 of catch in numbers per trap-haul-set-over-day 

 after August 1967. The ratios of catch per 

 unit of effort and the variances were calculated 

 using these formulas: 



(1)R = y/x; 



a y/l-f 

 (2)SR=- 



'n x 



^RZyjXi + R 2 Ex, 1 

 n-1 



In addition to the summary compilations, 

 we calculated the average length and weight, 

 percentages of females, shedders, and culls. 



Methods for Improving Survey 

 Estimates 



Another important determination can be made 

 from the expanded estimates of the survey, 

 that is ways to improve the precision of these 

 estimates and, in fact, include the entire fishery 

 in the sampling. 



After completing the sampling and expanded 

 estimates for 1967 and 1968, we determined 

 by optimum allocation the number of days 

 that we should sample a month with a 15% 

 standard error of the estimate. We were not 

 optimistic about reducing the maximum of 10 

 days that we have to sample per month because 

 of the restriction of the total number of days 



