An analysis of data from the complete dissection of one side from 285 animals from a breed comparison experiment was conducted to determine whether there were important biases in the prediction of percentage lean in the side based on dissection of the rumpback joint and, if so, whether the dissection of any of five other sample joints would have been better.

For estimating lean content of the side, it was found that the lean content of any sample joint except the hand was a more precise predictor than standard carcass measurements alone, while including the latter in a prediction equation based on a single sample joint dissection improved precision further. For prediction based on lean in the sample joint alone, any one of the ribback, ham and rumpback was best. If other carcass measurements were included (in particular, C fat depth, eye-muscle area and trimming percentage were found to be useful), there was little to choose between the joints. For all joints there were differences between sire breeds (Large White, Canadian Yorkshire, U.S. Duroc, U.S. Yorkshire, Danish Landrace or Norwegian Landrace), sexes (hog or gilt) and feeding regimens (ad libitum or scale) in lean content of the side at constant lean content of the sample joint, but the regression slopes differed only when the rumpback or streak was used as a predictor, and then only between ad libitum and scale-fed pigs. It will therefore be important in any future such experiment fully to dissect a sample of each class (breed, feeding regimen and sex) to provide unbiased estimates of the differences in lean content. A sample of 30 from each class should allow the differences to be estimated adequately.