198 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



Table 10. — Summary of test to determine whether the magnitudes of catches of eggs and larvae con- 

 formed to the distribution expected from sampling a normal frequency surface 



' The catches were divided into four classes, leaving this class vacant. 

 ' Lower limit for cruises VIII and IX where 2-meter nets were used. 

 3 The catches were divided into 3 classes, leaving this class vacant. 

 * The catches were divided into 2 classes, leaving this class vacant. 



This result may seem one in which the empirical data are closer to theoretical 

 expectation than they should be, for it will be recalled that the frequency surfaces, 

 as exemplified by the charts of figure 13, were not normal, but were skewed in one 

 direction or another, and were elongated rather than circular in form. The skewing 

 might not necessarily be detectable in the test, for the loss on one side may be ap- 

 proximately offset by the gain on the other, but the elongation should have its effect, 

 as is readily apparent if one imagines such elongation carried to its logical extreme. 

 Then the distribution would be in a band so that constant values would be found 

 when sampling longitudinally to the band, and values distributed in accordance with 

 the normal frequency curve, rather than the normal frequency surface, when sampling 

 across the band. At this extreme the catch magnitudes should be related to each 

 other as if drawn from the normal frequency curve instead of the normal frequency 

 surface. With intermediate elongation, such as indicated by the isometric lines of 

 figure 13, it is uncertain whether the distribution of catch magnitudes might be inter- 

 mediate between the type expected from the frequency curve and that from the 

 frequency surface, and hence fit neither; or whether it might still closely conform 

 to the type expected from the frequency surface as would easily be true if, in the 

 elongated surface, the form of the normal frequency curve were retained in the section 

 along its major axis. 



In any event, it is probably significant that the elongation of isometric shapes of 

 figure 13 is generally parallel to the coast, and also that the station grid is rectangular 

 rather than square, so that the mean spacing between stations in a direction longitud- 

 inal to the coast is greater than that in a direction perpendicular to the coast, the ratio 

 of the latter to the former averaging 0.44. Furthermore, by measurement it may be 

 found that the mean ratio of the minor to the major axis in the isometric shapes of 

 figure 13 is 0.47. Thus the sampling pattern was warped about the same amount 

 and in about the same direction as the egg and larval distribution patterns. One 

 compensates almost exactly for the other, and it is therefore less surprising that the 

 empirical data should fit the theoretical distribution, even though the latter did not 

 specifically take into account the elongation of the egg and larval distributions. 



Since it is impossible that hauls of indifferent quantitative accuracy, or that 

 sampling at a pattern of stations that did not adequately explore the area could, 



