samples with 20 or more measurements were 

 used. 



As is clear from the table, out of 45 

 comparisons, A has 28 and B has 13 in which 

 the variance shows significant differences; A 

 has 9 cases out of 17 and B has 15 out of 32 in 

 which the difference in variance is not signifi- 

 cant but the difference between the means is 

 significant and A has 8 cases and B has 17 in 

 which neither the variance nor the mean differ 

 significantly. These facts show, for both group 

 A and group B, that the samples from individ- 

 ual schools have no common size composition, 

 and therefore in neither group can they all be 

 regarded, on the basis of probability, as 

 samples drawn from the same universe. 



Kawasaki (1952), discussing the 

 populations of skipjack migrating into the 

 Northeastern Sea Area, has reported that the 

 size compositions of individual schools sepa- 

 rated by age cannot be considered on the basis 

 of probability as samples from a single popu- 

 lation. This agrees with the above-described 

 results obtained with bigeye. 



Various reasons for the lack of uniformity 

 among the schools, such as the influence of 

 biological characteristics and environment, 

 sampling error, and so forth, can be thought 

 of, but account must also be taken of the possi- 

 bility that one reason may be a lack of uniform- 

 ity within the populationS-' . If this is the case, 

 it is believed that the lack of uniformity seen 

 among the schools will provide an important 

 clue for deducing the lack of uniformity of the 

 population. 



Relation of Nonuniformity of Composition 

 to Locality 



In table 2, for the size compositions of the 

 landings of individual vessels the average 

 lengths of groups A and B have been found and 

 their relation to longitude has been shown. A 

 classification by cruise has been used because 

 there were few data available which were re- 

 corded by separate schools and the separation 

 by cruise was thought to provide the best 

 approximation. Since the problem here is the 

 local character of the size composition, we 

 have used only data from vessels which oper- 

 ated within areas of 4 degrees of longitude and 



— In this case the "population" indicates 

 the schools migrating into the fishing grounds 

 considered separately by groups. 



2 degrees of latitude. The vessels which 

 operated over more than 2 degrees of longitude 

 were assigned to the area in which the majority 

 of their fish were taken. Furthermore, since 

 the average length of the fish in samples varies 

 widely depending on the amount of data available, 

 only those samples in which 20 or more fish in 

 each group were measured were used. 



Of course, the size composition of the 

 catch of an individual vessel is thought to rep- 

 resent the averaged values of the composition 

 of several schools, but it is assumed that in its 

 trend it will reflect the peculiarities of the in- 

 dividual schools in the area in question. It can 

 be clearly seen from the table that for both 

 group A and group B the average length of the 

 fish is smaller in the more western areas and 

 that it tends to increase to the eastward. .2/ By 

 comparing, in the lower parts of both tables, 

 the frequency by longitude and by 10-day 

 periods, it is clearly shown that the above- 

 mentioned slope downward from east to west 

 is not due to a difference in growth arising 

 from any lag in the time of capture. JI' 



In figure 3 the distribution of average 

 lengths has been indicated, as explained in the 

 legend, by plotting the positions of capture with 

 symbols corresponding to three categories. By 

 referring to the plot, the downward slope from 

 east to west indicated in table 2 can be noted at 

 different latitudes. 



Figure 4 presents the size composition of 

 the bigeye measured at the Makurazaki fish 

 market. 



Since very few data classified by individual 

 vessels were available, the data from a number 

 of vessels were combined for presentation. The 

 locations of the catches in both June and July 

 were west of 130°E., and when they are com- 

 pared with the fish measured at the Yaizu 

 market during the same period of time, it can 

 clearly be seen that the fish were captured in 

 nnore westerly areas (see fig. 3). As for the 

 size composition, all of the inodes appearing in 



— For group A r = 0.73, for group B 

 r = 0. 56. In both cases the probability that 

 these values could be obtained from a universe 

 in which r = is P < 0. 005. 



7/ 



— For group A r = 0.11, for group B 



r = -0. 13. In both cases the probability that 

 these values could be obtained from a universe 

 in which r = is 0. 50>P >0.25. 



40 



