We consider estimating the infinite dimensional normal means from the prior Bayes standpoint and study the robustness properties of the prior Bayes procedure. Assumptions are made that the prior variances of the parameter vector are decreasing, with some extension. Under squared error loss, the...