1. By separating the sources of variability in an experiment involving haematological observations it has been shown how to estimate the usefulness of repeated observations on each animal.

2. In the illustration provided it is concluded that when monozygous twin calves are used, random variability accounts for so much of the total variability that a great improvement in accuracy can be achieved by taking repeated observations on each animal. When unrelated calves are used, little additional accuracy is obtainable by taking more than about three observations on each.

3. It is shown that, in consequence, the relative efficiency of monozygous twin and unrelated calves is likely to vary, depending upon the number of observations on each animal. Thus, to benefit fully from the increased efficiency offered by monozygous twins, more observations are required on each animal than in the case of unrelated animals.

4. Although the figures refer to haematological measurements on calves, the statistical method evolved is of wider application.