Table 3. — Statistics of length-weight relations for all data used in study. 



'All estimates are significantly different than at the 1% level. 



where a' = unbiased estimate of a 



(s^w'l) ^ mean square error about the re- 

 gression line. 



The mean square errors for this study are low 

 (Table 3). Thus the bias should be negligible. The 

 results of this study were examined by comparing 

 average weights of yellowfin used in the study 

 against predicted weights. Differences were 

 negligible as expected. 



The significant differences found among 

 samples and categories indicate that the variance 

 of estimated numbers of fish caught, estimated 

 from length frequency samples, could be reduced 

 by a sophisticated sampling scheme which is 

 stratified by category if not sample. Obviously 

 it would be simpler to weigh fish from each 

 sample rather than measure lengths, if one 

 desired to stratify by sample. Logistics rule out 

 this possibility. A formal cost-benefit analysis of 

 the effort required to develop an adequate 

 sampling scheme stratified by category probably 

 would rule out this scheme. The significant 

 differences among samples do point out the 

 desirability of obtaining large numbers of samples 

 rather than large sample sizes in further study 

 of length- weight relations. 



Acknowledgments 



E. Scott of the Southeast Fisheries Center, 

 National Marine Fisheries Service, NOAA, 

 Miami, Fla., measured fish under the supervision 

 of J. Wise of the same laboratory. I thank both of 

 these individuals for their helpful suggestions on 

 this paper. E. Holzapfel and M. Kimura of the 

 Southwest Fisheries Center, National Marine 

 Fisheries Service, NOAA, La Jolla, Calif., also 

 deserve thanks for performing most of the data 

 compilations and calculations used in the study. 

 I also thank D. Kramer of the Southwest Fisheries 

 Center for his technical editing of the paper. 



Literature Cited 



Batts, B. S. 



1972. Age and growth of the skipjack tuna, Katsuwonus 

 pelamis (Linnaeus), in North Carolina waters. Chesa- 

 peake Sci. 13:237-244. 

 Beardsley, G. L., Jr., and W. J. Richards. 



1970. Size, seasonal abundance, and length-weight rela^ 

 tion of some scombrid fishes from southeast Florida. 

 U.S. Fish Wildl. Serv., Spec. Sci. Rep. Fish. 595, 6 p. 

 Chatwin, B. M. 



1959. The relationships between length and weight of 

 yellowfin tuna (Neothunnus macropterus) and skipjack 

 tuna {Katsuwonus pelamis) from the Eastern Tropical 

 Pacific Ocean. Bull. Inter-Am. Trop. Tuna Comm. 3:307- 

 352. 



PlENAAR, L. v., AND J. A. THOMSON. 



1969. AUometric weight-length regression model. J. 

 Fish. Res. Board Can. 26:123-131. 

 Poinsard, F. 



1969. Relations entre longueur predorsale, longueur a 

 la fourche et poids des albacores Thunnus albacares 

 (Bonnaterre) peches dans le sud du Golfe de Guinee. 

 Cah. ORSTOM. (Off. Rech. Sci. Tech. Outre-Mer), Ser 

 Oceanogr. 7(2):89-94. 



WiLUAM H. Lenarz 



Southwest Fisheries Center 



National Marine Fisheries Service, NOAA 



La Jolla, CA 92037 



ELECTRICAL THRESHOLD RESPONSE OF 

 SOME GULF OF MEXICO FISHES 



Threshold voltage is the minimum electrical po- 

 tential to which an animal responds (Vibert, 

 1967). Usually threshold measurements are inex- 

 pensive and easy to obtain, and they provide 

 guidelines for designing electrical fishing sys- 

 tems. Bary (1956) and Kessler (1965) showed that 

 threshold voltage varied according to water temp- 

 erature, size of animal, and width of the pulse. 

 Earlier workers clearly demonstrated that 

 threshold voltages are affected by the position of 

 the animal in the electrical field. Klima (1968) 

 documented experimentally the mathematical re- 

 lationship between the ^ngle of the animal in the 



851 



