Dunnett’s method. Let X0 j (j = 1,..., m) and Xik (i = 1,..., s; k = 1,..., n) represent measurements on a standard and s competing new treatments, and suppose the X’s are independently distributed as N(ξ0, σ2) and N(ξi, σ2)

respectively. Generalize Problems 9.40 and 9.42 to the problem of obtaining simultaneous confidence intervals for the s differences ξi − ξ0 (i = 1,...,s).