The Bayesian approach of evidential reasoning is fairly widely used, e.g. by forensic scientists in crime investigation. However, its application can be controversial because it relies on a numerical parameters, in the form of probability distributions, that are difficult to scrutinise, e.g. by juries. This is particularly important when these parameters are not derived from databases or mathematical models, but are expressions of expert opinion in their own right. Argumentation driven approaches to evidential reasoning provide a means to scrutinise expert opinion, but existing techniques are not capable of reasoning about probability distributions. This paper introduces a novel approach to enable arguments concerning probability distributions to be formulated and used in Bayesian evidential reasoning. As such,it extends conventional argumentation driven evidential reasoning approaches to scrutinise the probability distributions that underpin a Bayesian analysis of evidence.