critical evaluations and few reports of faunal den- 

 sity specifically attributed to odometer wheels 

 (Belyaev and Sokolova 1960; Pearcy 1972; Carey 

 etal. 1973;Bieri 1974a, b). 



Methods 



Analysis consisted of comparisons of the actual 

 wheel performance with the performance ex- 

 pected of the wheels if working as designed. We 

 were concerned with accuracy and precision. Did 

 the wheels actually measure the distance towed, 

 and how much random variation was there in the 

 wheel counts? Exacting answers to these ques- 

 tions would have required a careful calibration of 

 the wheels under conditions encountered in sam- 

 pling at various depths. Such a deep-sea calibra- 

 tion would have required more effort than the 

 subsequent ecological sampling. However, partial 

 answers were obtained through the examination 

 of data collected during extensive ecological 

 sampling. 



As the trawl was dragged along bottom, the 

 wheels should have rotated with the rotations 

 counted on the hub odometers. Due to friction in 

 the wheel mounts and poor consolidation of the 

 sediment, each wheel slipped some portion of the 

 distance dragged. Thus the wheel counts should 

 have normally underestimated the distance actu- 

 ally towed. We are confident that the wheels did 

 not rotate while in the water column because in 

 those accidental cases where the trawl failed to 

 reach bottom, <10 rotations were registered. On a 

 single haul variation in slippage caused the left 

 and right wheel to register different counts. How- 

 ever, even if biased, the wheel counts should have 

 shown three relationships. First, wheel counts 

 should have been positively correlated with other 

 estimates of distance based on navigational fixes 

 and towing times. Second, the ratios of wheel 

 counts accumulated from all hauls should have 

 provided information on the magnitude of the 

 variation between wheels due to small-scale 

 sediment changes and operational characteristics. 

 Third, if sampling occurred within an area of 

 faunal uniformity, catch size might have been re- 

 lated to wheel count. However, since catch was also 

 determined by the actual, usually patchy, faunal 

 distribution, the absence of positive correlation 

 can not be taken as unequivocal evidence that the 

 wheels did not function as designed. 



Estimates of distance sampled based on changes 

 in loran A position or speed x duration of tow are 



792 



subject to major random error. In either case the 

 greatest error component is that associated with 

 determining the precise moment the trawl is on 

 and off bottom. Additionally loran A fixes contain 

 an inherent technical error, and speed x duration 

 estimates are dependent upon accurate deter- 

 minations of true speed over bottom. Since both of 

 these distance estimates were subject to random 

 error, correlation was the appropriate method of 

 comparison. The raw data were critically 

 examined to remove as much obvious misinfor- 

 mation as possible. If the distance by loran was 

 greater than it was possible for the ship to have 

 gone given speed and current conditions, the tow 

 was excluded. The greatest source of loran A error 

 seems to have been simple operator error yielding 

 tow distances that were far too great. 



The beam trawl was made of a thick-walled, 

 hollow aluminum tube bolted across the top of two 

 steel skids. The skids were lined with netting, and 

 an otter trawl type net attached to the trailing 

 edge of the skids (Figure 1). The basic configura- 

 tion and full operating details have not changed 

 since the initial report by Carey and Heyamoto 

 ( 1972). This beam trawl had many similarities to 

 those used for fishing since the 16th century 

 (Davis 1958). The bicycle wheels of an earlier ver- 

 sion of the trawl were replaced by heavier, spiked 

 aluminum disks to decrease damage to the wheels 

 during launch and recovery and while on bottom. 

 A Veeder Root^ model 54-794692 (Veeder Root 

 Corp., Hartford, Conn.) hub odometer was at- 

 tached to the axle of each wheel, counting once 

 each revolution (2 m distance). The counter was 

 housed in a thick-walled brass case filled with 

 silicon fluid to prevent air spaces which might lead 

 to crushing at high hydrostatic pressures. The 

 wheel and its mounting fork were attached to the 

 outside of each skid, free to pivot about the mount- 

 ing bolt. Surgical tubing was used to restrict the 

 angle of swing. Short lengths of angle iron were 

 welded to the bottom of the skids in front of the 

 wheels. These were intended to protect the wheels 

 and to prevent rotation while off bottom. The 

 trawl was paid out at a ship's speed through water 

 of approximately 2 kn. No weights were placed on 

 the bridle or towing line. The time of bottom con- 

 tact was estimated by an empirically determined 

 table of wire-out needed to reach bottom. A 

 Benthos model 1170 (Benthos Corp., Falmouth, 



I 



'Reference to trade names does not imply endorsement by the 

 National Marine Fisheries Service, NOAA. 



