analysis of the differences meaningful. Each of the 10 most common serotypes previously reported (5, 6, 7) from 

 animal by-products was found in these samples by one or more of the techniques. 



A paired comparison technique for proportions (with a correction factor) as described by Snedecor and 

 Cochran (4) was used to compare the three wet compositing methods with the standard. The results are summarized 

 in table 3. There was no significant difference (a0.05) in the proportion of lots found positive by standard and 

 compositing methods. Admittedly, the paired comparison test has low power with the small number of discrepancies 

 involved. An examination of lot 42 and lot 44 using Fisher's Exact Probability Test (under the two assumptions that 

 one composite is equal to 10 individual samples and that the organisms are randomly distributed throughout the 

 original aliquot) also reveals that the composite methods will occasionally fail to detect salmonellae when they are 

 present. Unfortunately, Fisher's Exact Probability Test has low power (or is inappropriate) in testing whether the 

 standard culture method fails to detect salmonellae when they are present. (This is because the composite methods 

 reveal only that at least one of the 10 assumed samples making up the composite is positive.) Lot 35 (where all three 

 composite methods and of 10 individual samples were positive) and lots 3, 19, 33, 39, 48 and 58 (where two of 

 three composites and of 10 individual samples were positive) indicate this as a distinct possibility. Thus while 

 further testing may be necessary to completely resolve the issue, these findings indicate that the wet compositing 

 procedures described herein can be used to isolate salmonellae from animal by-products with an accuracy similar to 

 the standard methods employed. 



Table 3.- 



-Standard 



and wet composite cultural results 





Standard method 



Composite methods 



No. of pairs 



Aj 



A 2 



B 



Ai 



A 2 



B 



+ + + + 



+ + + 



+ 



TOTAL 1 



1 X 2 = 0.083. 



2 X 2 = 0.364. 



3 X 2 = 0.111. 



.it, 



37 



38 



7 



7 



3 



5 



4 



3 



12 



12 



16 



(.1) 



60 



60 



REFERENCES 



Anonymous. Recommended Procedures for the Isolation of Salmonella Organisms from Animal Feed and 

 Feed Ingredients. ARS 91-68, Agricultural Research Service, U.S. Department of Agriculture. 1968. 

 Harrington, R., Blackburn, B. 0., and Murphy, C. D. Culture Agglutination Procedure for Detection of 

 Salmonellae in Rendered By-Product Samples. Developmental Studies and Laboratory Investigations 

 Conducted by Diagnostic Services, NADL. APHIS 91-2, pp. 27-31, 1972. 



. The Efficiency of Dry Compositing for Detecting Salmonellae in Animal By-Products. Proc. 



76th Ann. Meeting of the U.S.A.H.A., 1972. In press. 



Snedecor, G. W. and Cochran, W. G. Statistical Methods. 6th Ed. Iowa State University Press, 213-215. 1967. 

 Wilson, S. T. and Germaine, F. W. Status of the State-Federal Salmonella Program. Proc. 72nd Ann. Meeting of 

 the U.S.L.S.A. pp. 129-138, 1968. 



, Walker, J. W., and Pfow, C. J. Status of the State-Federal Salmonella Program. Proc. 73rd 



Ann. Meeting of the U.S.L.S.A. pp. 33-45, 1968. 



. Status of the Cooperative State-Federal Salmonella Program. Proc. 74th Ann. Meeting of 



the U.S.L.S.A. pp. 436-448, 1970. 



40 



