266 DISCOVERY REPORTS 



the second net also on the surface was simultaneously and equally slowly being paid away from the 

 port. And so a continuous night and day sampling of the surface layer was carried out over a distance 

 of 30 miles. The principal and perhaps vital point of difference between this procedure and the 

 ordinary routine surface hauls we made was that when ' all out ' the nets of the consecutive series 

 were fishing much farther astern than the conventional surface net was ever allowed to do. It is 

 interesting, therefore, and perhaps of the highest significance to record, that it was these nets, fishing 

 far back in the wake, that produced the only instances of substantial daylight sampling of large krill 

 by stern nets on the surface, sampling involving catches running up to between 10,000 and 30,000 

 individuals, our data provide. It might well be then, that while lateral towing obviously provides a 

 highly efficient method of sampling the surface population, long- warped surface towing from the stern, 

 which is simpler to operate, might prove, by day as well as by night, an equally effective procedure. 

 Mr Peter David tells me that during the Antarctic voyage of ' Discovery II ' in 1950-1 he fished the stern 

 surface net on a much longer line than we used to do (p. 54) before the war.^ It is of further interest, 

 therefore, to record that on one of the three occasions when this net on its longer warp was used in broad 

 daylight in a region of known euphausian abundance (namely, in the Weddell drift) a substantial, by 

 pre-war daylight standards enormous, catch of large (over 20 mm.) krill, 2387 in all, was again obtained. 



The recent remarks of Hedgpeth (1957) seem to me particularly pertinent to the major problem 

 that confronts us in sampling these congregating and extremely active animals. He writes : ' Too many 

 ecologists, especially fisheries workers, employ statistical procedures without any clear idea of what 

 they mean and, what is worse, often apply them to data which are of doubtful biological validity. 

 For example, the idea that the size of a large fish school may be estimated by the tagging and recapture 

 method without reference to the schooling behavior of the fish is inexcusably bad biology and a 

 squandering of public funds. Adequate statistical procedures are tedious and time consuming, and it 

 would be well to spend a comparable amount of time at the outset in working out an adequate sampling 

 technique '. It would be well indeed, for as Hedgpeth has said, the old saw that ' statistics are the ulti- 

 mate degree of prevarication ' is too often true in ecology, especially in marine ecology, in which we grope 

 so much in the dark among diverse animals or animal communities, reacting for all we know in a diverse 

 manner of complex ways to the intrusion of our ships and nets. It would not in fact be at all surprising 

 if it should prove that it is ahead of a moving vessel that a surface net should be fished by day rather 

 than astern of the cleaving hull. This at any rate would be an experiment worth trying. We might also 

 try hauling nets on the surface from points some distance away in the wake back to a stationary hull. 



To conclude, it is obvious that our nets, that is, our stern-towed stramin nets,^ provide little 

 indication of the natural abundance, or absolute density, of the krill in the sea, (i) because the ship 

 scatters or splits wide apart their surface concentrations, and (2) because of the ability of these animals, 

 especially the larger individuals, to see or feel, or become otherwise aware of, the nets, and dodge them 

 by rapid movements. All that can be said is that direct observation and the evidence provided by the 

 boom net indicate that the density in a swar?n, and it seems clear enough that the krill spend their 

 whole existence in swarms, might be of the order of one to the cubic inch. However, the number of \ 

 samples (p. 53) being so large, and the stations that produced them (p. 50, Fig. i) so numerous,^ in 

 so far as the nets were fished in exactly the same way they can manifestly be said to provide a basis* for 



* Professor Hardy tells me that in the old 'Discovery' they also used to fish the surface net on a long warp. 



^ The boom net was used only on a few occasions and its full significance had probably not been fully appreciated until 

 I examined the samples in some detail long afterwards. 



^ South of the 50th parallel alone there are over 2500 stations, with many hundreds more to the north. 



'' Since, however, during the day the krill over 20 mm. practically avoid the surface net outright, and in some instances 

 (p. 271) may be concentrated at some deeper level, the daylight (0-5 m.) catches in this range, but in this range only, mani- 

 festly provide no such basis and have accordingly, both positive and negative, been discarded from the distributional data. 



