Right sides of 30 pork carcasses were fabricated into indicator cuts (Boston butt, picnic, anterior belly, posterior belly, anterior loin, posterior loin, and ham) and used to determine carcass composition by physical dissection and chemical analysis of soft tissue. Linear measures and chemical composition of whole-side grind were collected from left sides. Dissection of indicator cuts to predict dissectable composition of the side were much better indicators than linear measures ( R 2 = .70 to .94). Chemical analyses of soft tissue (lean + lipid) from the loin and anterior belly were better predictors of fat-free lean and extractable lipid composition of the side than linear measures; 89% of the variation in dissectable lean of the side was explained by the use of dissectable components of the anterior loin. The loin served as the best overall indicator cut across the various end points evaluated; 92% of the variation in dissectable lean of the side was explained using dissectable components of the shoulder. The anterior belly was a good indicator for percentage of dissectable lean, fat, fat-free lean, and extractable lipid.

Thirty pork carcasses were used to evaluate the interrelationships between compositional end points and the accuracy of predicting various end points from one another. Right sides were fabricated into Boston butt, picnic, anterior belly, posterior belly, anterior loin, posterior loin, and ham. Carcass composition was determined on the right side using physical dissection and chemical analyses of soft tissue. Left sides were ground (whole-side grind). Linear measures and chemical composition were made on left sides. Using dissectable components, high (r > .85) correlations occurred between percentage of fat-free lean and percentage of lipid, percentage of dissectable fat, and lipid in the whole-side grind; between weights of dissectable lean and fat-free lean; and between weights of lipid in the whole-side grind and lipid, dissectable fat, and dissectible lean. Dissectable lean percentage and fat (percentage and weight) could be predicted using weight of lipid, weight of fat-free lean, and side weight. Any physical measure could be accurately estimated from any compositional end point (r2 = .79 to .93). Regression equations using 10th-rib linear measures and hot carcass weights predicted compositional end points with R2 > or = .80, with the exception of dissectable lean (R2 = .56).