This article describes a means by which to undertake Bayesian posterior inference via sampling techniques when the normalizing constant is not computable and hence unavailable. The strategy relies on the introduction of latent variables which removes any integrals associated with the inaccessibility of the normalizing constant.