An infinite series of resolution III.2 designs for the \(2^ m\) factorial experiment (Q1185090)

Much combinatorial argument is given for a resolution III.2 plan, i.e. a resolution III plan for the \(2^ m\) factorial experiment which satisfies that at most two interactions are nonnegligible. Here which interactions are nonnegligible is unknown. A construction method of this plan involves the checking of the linear independence of \({2^ m-m-1\choose 4}\) possible sets of four columns from a certain matrix. After a tremendous process for this observation, it is shown that the design consisting of all the distinct treatments of weight \(m\), 1, 2, and 0 constitutes a resolution III.2 plan, where the weight of a treatment means the number of factors at level 1 in the treatment.