An unbiased comparison of organ weights when an inequality in body weight exists

It is demonstrated in simulations that, if adjustment for inequality in body weight is to be made by covariance analysis and the comparison is carried out in terms of adjusted means of organ weights thus obtained, the test of significance for an organ weight change can be unbiased, or the chances for detecting spurious effects on organ weights remain at about 5% and the chances for detecting treatment effects are far above the 5% irrespective of differences in body weight. Since the use of relative organ weights frequently violates the unbiasedness, it is concluded that if a treated group differs from the control in body weight the comparison of adjusted means of organ weights by covariance analysis will be far more justifiable than that of relative weights.