(Received June 5, 1995; revised February 19, 1996)
Abstract. The problem of identification of uniform mixtures via posterior means is studied. For linear posterior means a complete solution is given. It determines a family of prior distributions involving beta of both kinds and gamma. Identifiability via any consistent posterior mean is also investigated.
Key words and phrases: Mixtures, posterior mean, identification, characterization of probability distribution, uniform conditional distribution, conditional moments, linearity of regression, gamma distribution, beta type distributions.