282 DISCOVERY REPORTS 



catch-figures of the oblique nets, in the many instances where such catch-figures alone are available, 



in order to bring them more into line with the heavier gatherings from the surface nets. 



It will be seen from Table 60 that the ratios of the total horizontal (0-5 m.) to the total oblique 

 (loo-o m.) catches of the under 16 mm., 16-20 mm. and over 20 mm. classes of the surface population 

 are roughly 2, 5 and 4 respectively, and, accordingly, in the absence of any more satisfactory method 

 (see below), the multiples 2 for the under 16 mm. class and 4 for both the 16-20 mm. and over 20 mm. 

 classes have been appHed to raise the oblique (lOO-o m.) catches of these classes (in all instances where 

 such catches alone are available) to the orders of abundance they would evidently severally have been 

 found in had they been obtained in the surface net, or, as in the case of the adults alone, the night 

 surface net. The corrections principally affect the catch-figures of our very large collection of oblique 

 (lOO-o m.) day samples of the larger (over 20 mm.) krill, these animals, as already noted, being 

 virtually impossible to sample with the surface net except when used in the dark. In the result, it will 

 be seen, although not mathematically exact, in so far as they may be said to operate to the advantage 

 of the distributional data, they appear on the whole to be reasonable and to operate tolerably well. 



Table 60. Total catches of the horizontal {o-^ m.) and oblique {100-0 m.) stramin nets 



when fished simultaneously 



In plotting the distribution and relative abundance of the surface population on our circumpolar 

 charts I have used for the stramin net samples a conventional system of circles of increasing diameter 

 representing catches grouped in the following orders of magnitude, i-ioo, loo-iooo, 1000-10,000 

 and over 10,000. Referring again to Tables 57-9 and assuming all negative catches to have been in 

 fact very small ones, of the order of say i-io, it will be found that in 331 cases out of the grand 

 total of 381 simultaneous samplings where the surface catch was the larger, if the oblique catch be 

 multiplied by 2 or 4 as required, the figure we arrive at in each instance falls within the order of 

 magnitude represented by the horizontal catch or does not deviate materially from it. The corrections 

 applied do not, of course, always achieve this result. For example, in the sixth sampling of the staple 

 population shown in the first column of Table 59 we have a clear instance where the surface net must 

 have struck a swarm and sampled it with considerable eflFect, whereas the oblique net obviously missed it, 

 or virtually missed it, altogether, producing so small a catch that it makes no difference whether the cor- 

 rection be applied or not. However, they work so frequently in the right direction, bringing the oblique 

 catch-figures so often into line with the orders of magnitude represented by the surface samples, that 

 they can manifestly be used with advantage in the construction of the distributional charts. Even in 

 the relatively few instances where they do not operate to full advantage, as, for instance, where the 

 surface catch was 11,500 and the oblique only 50 (Table 59), it is obviously better, assuming there to 

 have been no surface net at this station, that the oblique catch should be represented by a minor 

 (lOO-iooo) order of magnitude rather than by one of insignificant size. Again, taking another example 

 from Table 59, where the horizontal catch was 10,248 and the oblique 408, it is far better, again 

 assuming absence of surface observation, that the latter should be raised to the second highest order 

 of magnitude rather than left as it stands. In the many instances where the oblique net failed to take 



