Davis (1934) showed that the undersized fish escaped during the 

 towing, not, as many others thought, during the hauling on board of 

 the neto Davis fastened a fine-meshed net around the cod end of the 

 trawlo This outer net was closed by a clever device before the trawl 

 was hauled on board. The fish contained in the outer net were in 

 good condition. Therefore, the meshes are open during the towing and 

 it is possible for the small fish to escape undamaged. Iversen (1934) 

 v^made observations and tests with outer nets in the Barents Sea and 

 came to the same conclusion, i.e., that even the herring were still 

 in excellent condition after passing through the meshes. Herrington 

 (1935) did rather similar work on board American trawlers of the 

 New England coast. 



Several tests have been made to decide through which parts of 

 the net the escape takes place chiefly. Todd (1911) found by setting 

 outer nets at various parts around the net that most fish escaped 

 through the rear-end — the so-called cod end— of the net« On the 

 other hand, Borowiok (1930) found that a fairly large amount of fish 

 escaped through other parts of the net. It is not impossible that 

 the results of these tests were disturbed by interfering factors. 

 Clark (1934) found with Borowik's method that most undersized fish 

 escape through the cod end of the seine (zeevischzegen) . Many other 

 observations support the idea that the fish only begin to try to 

 escape after they have entered the cod endo The cod-end meshes are 

 always the narrowest ones of the net. Therefore, when considering 

 the effect of mesh-width enlarging, it may be sufficient to compare 

 the effect of cod ends of various mesh widths. 



The best conception of the effect of a cod end of a certain 

 mesh width is obtained when the numbers of the various length groups 

 of a fish species are determined with the aid of a fine-meshed net. 

 Then, at the same locality, the percentage of the total number of 

 each length group that remains in the net of a certain cod end can 

 be determined. Data are then obtained whereby a selection curve can 

 be drawn, ioe., a curve which shows what percent&ge of the fish ©f each 

 length group that come into the net remains there o Figure 5 shows 

 several selection curves made during various tests of foreign 

 researchers* The curves shown in figure 5a for double-twine cod 

 ends of 5 l/2, 6 l/2 and 8 3/4 centimeters for haddock, are part of a 

 series of tests made in the same locality, with the same ship and 

 the same speed of tow (Bowman, 1930), Another series show the 

 selection curves for plaice for double"=twine cod ends of 5 1/2 

 centimeters (Bftckmann, 1932) and 10 centimeters (Davis, 1934 c)o 

 Ti'ie curves for single-twine meshes of 7 l/2 centimeters and 8 3/4 

 centimeters, together with the curves shown in 5c for sole, form 

 part of a series in which sailing trawlers were used instead of 

 steam trawlers (Buchanan-Wollaston, 1935) o 



4G> 



