averaged 108 mm in size but by the second and third day of the 

 experiment averaged 106 mm. The twine used for the hand- 

 made larger cod ends apparently was not heat-treated. The dry 

 measurements averaged 154 mm and during the experiment the 

 mesh averaged 139 mm, a 10% shrinkage rate. No stretching 

 of the twine was observed during the experiment. There was no 

 consistent variation between meshes of the forward and aft 

 parts of the cod end as would be logically expected with larger 

 catches. In the small mesh there was a maximum range of 16 

 mm (0.6 in) between mesh sizes. In the large mesh the max- 

 imum range was 23 mm (0.9 in). A series of standard error 

 calculations (Appendix C) shows that the 95% confidence 

 limits are within 1 mm of the sample mean. 



The tows were conducted as described previously. On the 

 first day a number of problems were encountered. The twine 

 started to freeze before mesh measurements could be taken. 

 During Tow 2 a cover float flooded on the Christopher An- 

 drew, causing a marked masking effect. During Tow 3 the 

 Frances Elizabeth caught a large object that caused a door (ot- 

 ter board) to capsize. Tow 4 was scrubbed because of darkness 

 and the resulting change in fish population available to the 

 gear. For these reasons the first day's data were not used in the 

 overall analysis. All data presented, unless otherwise in- 

 dicated, are for only the second and third days of this experi- 

 ment. Appendix D presents the basic tow information. Appen- 

 dix E is a listing of the catch by weight per tow. The 

 "flounders" category consisted mostly of winter flounder, 

 though some American plaice were included. The "other" 

 category consisted mainly of skates; sculpin (Myoxocephalus 

 sp.), goosefish, Lophius americanus; crabs; and windowpane 

 flounder, Scophthalmus aquosus. 



Experiment Two 



This experiment was conducted from the fishing vessels Lin- 

 da B and Metacomet on 22, 23, 25, and 28 March 1978, in in- 

 shore waters off Gloucester, Mass. (Fig. 2). The experiment 

 consisted of four four-tow series by each vessel. The towing 

 order was chosen to minimize cod-end changes during the ex- 

 periment and thus consisted of the following: 



Day 1 Day 2 Day 3 Day 4 



Sm mesh Lg mesh w/cover Sm mesh w/cover Lg mesh 



Sm mesh w/cover Lg mesh Sm mesh Lg mesh w/cover 



Lg mesh Sm mesh w/cover Lg mesh w/cover Sm mesh 



Lg mesh w/cover Sm mesh Lg mesh Sm mesh w/cover 



Both vessels towed together at 2.5-3.0 kn. 



Thirty meshes were measured after each tow, and means, 

 standard deviations, and standard errors calculated (Appendix 

 C). The mean mesh size for the small cod ends on both vessels 

 was practically the same, equalling 99 mm (3.9 in) when round- 

 ed off to the nearest millimeter. These same cod ends were 

 used in the previous Scituate experiment and had averaged 106 

 mm (4.2 in). The large cod ends, which had averaged 139 mm 

 (5.5 in) during the Scituate experiment, had a mean mesh size 

 during this experiment of 131 mm (5.2 in). However, the dif- 

 ference between the average mesh size of the two large cod 

 ends, which was 3.5 mm during the Scituate experiment, had 

 grown to 5.3 mm (0.2 in). Both mesh gauges were tested 

 against each other by measuring 10 of the same meshes and 

 found to be reading the same. In addition, each gauge was 

 tested by pulling against a calibrated spring scale and found to 

 be calibrated correctly at 4 kg pressure. 



There were large variations in catch size and composition 

 between tows, even on a daily basis, making an actual catch 

 comparison between cod end sizes difficult. Many tows came 

 up with lost lobster traps and big pieces of waterlogged wood 

 that were in the area due to a large February storm. The Linda 

 B snagged 14 lobster traps in 6 tows, the largest catch being 4 

 traps. The Metacomet snagged 6 traps in 2 tows, one tow ac- 

 counting for 5 traps. The traps' condition varied from good to 

 broken up. There were no lobsters in any of the traps nor any 

 good buoys or lines attached. The traps were all found on sand 

 or mud bottom. Most of the traps were caught on the twine 

 forward of the trawl extension. No obvious effect on mesh 

 selectivity was apparent. 



The basic catch data are presented in Appendix E. The 

 "other" category consisted mainly of windowpane flounder, 

 sculpin, skates, crabs, and sea ravens. The Metacomet 

 grouped the ocean pout with the "other" category. There was 

 a small incidental catch of goosefish; lumpfish, Cyclopterus 

 lumpus; Atlantic wolffish, Anarhichas lupus; grey sole, Glyp- 

 tocephalus cynoglossus; and 12 lobsters (Homarus 

 americanus). One small Atlantic halibut, Hippoglossus hip- 

 poglossus, a 15-lb sturgeon, and a 74 cm haddock were caught. 

 Only a few small pollock were caught throughout the study ex- 

 cept for Metacomet Tow 11 where 140 pollock (13 kg) were 

 found in the cover, measuring 18-30 cm, the majority being 

 19-22 cm. 



Experiment Three 



This experiment was conducted from the Gloucester based 

 fishing vessels Joseph & Lucia //and Joseph & Lucia III, 13-15 

 August 1978, in the offshore waters of Georges Bank (Fig. 2). 

 The experiment consisted of three four-tow series by each 

 vessel. The series was initially chosen, as in Experiment Two, 

 to minimize cod end changes. However, due to problems with 

 the covers and a large catch of pollock on board (from com- 

 mercial fishing at night) that had to be landed early, the experi- 

 ment consisted of the following: 



Day I 



Day 2 



Day 3 (J&L II) 



Sm mesh w/cover Sm mesh Lg mesh 



Sm mesh Sm mesh w/cover Sm mesh 



Lg mesh w/cover Lg mesh Lg mesh 



Lg mesh Lg mesh w/cover Sm mesh 



Day 3 <J&L 111) 



Sm mesh 

 Lg mesh 

 Sm mesh 

 Lg mesh 



On the first two days both vessels towed in the same order; on 

 the third day the vessels alternated uncovered tows. The vessels 

 towed within a kilometer of each other at 3.5 kn. 



Thirty meshes were measured after each tow; and means, 

 standard deviations, and standard errors calculated (Appendix 

 C). The small cod ends on the Joseph & Lucia II and Joseph & 

 Lucia III had mean mesh sizes of 103.7 mm (4.1 in) and 109.6 

 mm (4.3 in), respectively. These same cod ends averaged 99 

 mm (3.9 in) in the second experiment and apparently stretched 

 during the night fishing that preceded the experiment on this 

 trip. The large cod ends both averaged 140 mm after a 2-h 

 break-in tow. However, during the experiment the mean mesh 

 sizes were 135.9 mm (5.3 in) and 140.8 mm (5.5 in), respective- 

 ly. 



The area fished had very few small fish of any species and 

 the catch was quite "clean" or lacking much "trash" fish. 



