The prior distribution of the parameters has to be specified. This consists of specifying the distributional form of the priors, together with any of the necessary parameters. In the hierarchical model, the crack specific parameters already come from distributions with hyperparameters. It remains to specify and elicit priors for these hyperparameters.