Data quoted in P M Lee, `Bayesian Statistics: An Introduction',
Arnold 1997, Chapter 9, Exercise 5.

Iterations of the EM algorithm give the following values for theta
88.9793 88.9862 88.9862 88.9862 88.9862 88.9862 88.9862 88.9862 88.9862 88.9862 

The Gibbs sampler gives rise to the following conclusions:
We deduce posterior for theta has mean 89.0409 and variance 1.30669
and that posterior for phi has mean 128.26 and variance 359.461