CORRECTIONS APPLIED TO THE CATCH-FIGURES 283 



a single euphausian the correction of course cannot be applied. However, this does not matter a great 

 deal because as our tables show this always happened when the surface population was itself evidently 

 very scarce or represented at most by samples of minor abundance, by gatherings in fact rarely 

 exceeding 100 and never more than 500. 



Although these are rough calculations, any more accurate estimate of the factor, or factors, required 

 to bring the oblique into line with the surface catch is precluded by the wide scatter of the oblique 

 and horizontal catch-figures. An estimate (for the over 20 mm. class) based on the method of least 

 squares leads to a value of 3, but the error involved is so large that there is really nothing to choose 

 between this figure and the simple ratio of 4. 



These standard corrections have not been applied in cases where the oblique catch was already in 

 excess of 10,000,1 nor in such instances where simultaneous sampling revealed it to be larger than the 

 surface, all such instances being accepted as representing the true relative order of abundance of the 

 surface population wherever they chanced to be recorded. Nor has any correction been made for the 

 state of the sea. This is hardly necessary since, except in the shelter of the pack, these southern waters, 

 as Tables 51 and 52 show, are generally in the state of agitation in which the krill are most easy to 

 capture. The great majority of our samples were in fact obtained in such conditions, and with a mass 

 of data so large the relatively few occasions when the nets were fished in calm or smooth water, to the 

 manifest disadvantage it is true of the catch, are not likely to affect the overall picture of the distribu- 

 tion to any material extent. Some allowance, however, has been made for the persisting daylight in 

 which a number of our night surface hauls were made in high latitudes during the Antarctic summer, 

 the catch-figures for all such nets being multiplied by 3, since the average catch obtained when the 

 sun was a little below the horizon (Table 50) was approximately three times smaller than the average 

 catch in full darkness. This obviously seems too small a correction for the surface catches obtained 

 when the sun was low on the horizon, or at sunset or sunrise, the average catch in these conditions 

 being 27 times smaller than the average catch of the dark hours. It so happened, however, that at 

 each of the 22 stations worked in this failing state of the light the oblique catch was so correspondingly 

 small that the application of any larger factor than 3 would not I think in these instances have been 

 justified. Finally, where necessary, corrections have been made for time, all towings being resolved into 

 standard hauls of ^ hr. duration (see p. 59), and also for net aperture (see p. 157). 



Where relevant the above corrections have been used in the construction of all distributional charts 

 based on the data from the stramin nets, and, as explained on p. 59, in certain tables and figures 

 which appear on earlier pages. Although arbitrary they are obviously necessary and can be justified 

 on the ground that some such correction, even if uncertain, is better than blind acceptance of catch- 

 1 Although this is statistically inconsistent, it is in fact the only reasonable thing to do. The highest order of abundance in 

 which we normally sampled the staple population (krill over 20 mm. long), with either the oblique or surface net, was 

 10,000-100,000. Very rarely, in fact once in the oblique and once in the surface net, this order was exceeded and gatherings 

 of 100,000-1,000,000 were obtained. Only seven, however, of the 1700 odd oblique nets we shot produced catches of 10,000- 

 100,000, all from the richly populated East Wind-Weddell surface stream. These seven gatherings, therefore, must represent 

 it seems instances where the surface net would not normally have done any better, and by leaving them as they stand all I have 

 done has been to keep the highest order of abundance our stern nets can reveal within its normal limits. As I see it, biologically, 

 there could only have been one strong reason for multiplying these seven oblique gatherings by 4, and that would have arisen 

 if the surface net had consistently been capturing 100,000-1,000,000 euphausians of the over 20 mm. class. 



In plotting the distribution of plankton animals we cannot use mathematics on our charts. All we can do with our conven- 

 tional systems of circles or squares of increasing size is to show broad differences, and the whole idea is to show these dif- 

 ferences as clearly as possible, choosing a biggest circle that is not too large to obscure the rest of the data and yet not so small 

 that the differences are not well defined. Sometimes it is true my largest circles do obscure a number of negative stations, 

 but this is an unavoidable imperfection associated principally with the closely worked South Georgia whaling grounds, a 

 relatively insignificant part of the vast area we covered, and even there it does not matter very much since the facts of the 

 distribution in these island waters are shown later on separate seasonal charts drawn to a much larger scale. 



32-2 



