The analysis of organ weight data.

The increase in precision that can be obtained when the bodyweights of animals at the start of an experiment are used as covariate in the analysis of organ weights is demonstrated. Theoretical reasons are given why the analysis of relative organ weights does not eliminate the indirect effect of treatment on organ weight via its effect on bodyweight. An alternative method, that of analysis or organ weights using the current bodyweight as a covariate, is described. Results from these two methods of analysis carried out on simulated experimental data are presented. Analysis of covariance is demonstrated to be superior to analysis of relative organ weights.