Ideas from the Foundations of Sampling Applied to the One-way Layout

This paper is an attempt to apply ideas from the foundations of sampling theory to a situation in the design of experiments, namely the one-way layout with two treatments and no technical error. Basic considerations concerning the role and choice of randomization are shown to correspond. There is no obvious value in the design context for analogues of πps sampling. An optimality result for estimation of the difference in treatment effect is given.