ABSTRACT
Two bones were available from each embryo and the experiment was therefore set out in an incomplete block design with two units per block. This experiment forms a balanced incomplete block design. It would be possible to analyse it via the special formulae for such designs but, partly because of the special simplicity resulting from the use of only two treatments per 'block', it is more instructive to proceed from first principles. The treatment pairs were randomized, but the following results are given in systematic order. The simplest analysis of such designs, the so-called within-block analysis, regards variation between pairs as arbitrary, each pair thus in effect having an associated parameter unique to that pair. For incomplete block designs like this, a second independent analysis can be made in which the pair parameters are regarded as random and the pair sums or means regarded as a second derived response variable.
