Fishery Bulletin 97(1), 1999 



in the volume of catch. The 

 above procedure was then 

 repeated to obtain data on 

 the flow of water at each po- 

 sition with 50 kg (25 bal- 

 loons) and 70 kg (35 bal- 

 loons) in each codend. 



Data analysis 



B>A>C>D 



I I 100 commercial codend 

 H 200 commercial codend 



2 . £ Water speed 



11 . 



1.0 



09 



S 08 



07 



06 



30 Kg 50 Kg 70 Kg 

 Position no. 1 



Data collected in each ex- 

 periment were examined by 

 using Cochran's test for ho- 

 mogeneity of variances and 

 transformed if necessary. 

 Data from experiment 1 

 were analyzed in a three- 

 factor fully orthogonal, bal- 

 anced analysis of variance. 

 The factors were codends, 

 positions, and weights. Data 

 from experiment 2 were 

 analyzed at each position in 



two-factor fully orthogonal, balanced analyses of vari- 

 ance (Underwood, 1981). Significant differences de- 

 tected in these analyses were investigated by Stu- 

 dent-Newman-Keuls (SNK) multiple comparisons of 

 means. 



Results 



Experiment 1 



The analysis of variance showed that there were sig- 

 nificant differences in flow rates between the type of 

 codend, position of the current meter and weight in 

 the codend, and significant interactions among 

 codend-type, positions and weights (Table 1). SNK 

 tests showed that mean water flow was greatest at 

 position no. 2 (1.162 m/s) in the 100 commercial 

 codend with a catch of 30 kg and lowest at position 

 no. 3 (0.709 m/s) in the 200 commercial codend with 

 70 kg ( Fig. 2 ). SNK tests also showed that, compared 

 with the 100 commercial codend, there was a signifi- 

 cant reduction in mean water flow in the 200 com- 

 mercial codend with 30 kg at position no. 2 (mean 

 difference of 0.071 m/s) and across all three weights 

 (30 kg, 50 kg, and 70 kg I at position no. 3 (mean dif- 

 ferences of 0.203 m/s, 0.176 m/s and 0.184 m/s, re- 

 spectively) (Fig. 2; Table 1). Although SNK tests did 

 not detect differences in mean water flow between 

 codends with 50 kg and 70 kg at position no. 2, nor 

 across all weights at position no. 1, these combina- 

 tions of weight and position showed trends similar 



In flume tank 



1 



30 Kg 50 Kg 70 Kg 

 Position no. 2 



30 Kg 50 Kg 70 Kg 

 Position no. 3 



Figure 2 



Differences in mean flow rates ±SE between the 100 and 200 commercial codends tested 

 in experiment 1 (< and > indicate direction of differences in SNK tests). 



to those described above — i.e. a mean reduction in 

 water flow with the 200 commercial codend (Fig. 2). 



Experiment 2 



There were significant differences in the mean wa- 

 ter flow between the type of codend at position nos. 

 2, 3. and 4 and between weights at position nos. 2 

 and 3 (Table 2). There was a significant interaction 

 between the type of codend and weight at position 

 no. 1. SNK tests detected differences in the mean 

 flow rates between the 100 and 200 panel codends at 



