The first is that a miracle, understood as an event produced by a transcendent agent overriding the usual course of nature, involves a violation of the laws of nature. Larmer argues that events are explained by reference to both relevant laws and units of mass/energy in the sequences to be explained. He contends that a miracle need not be conceived as involving a violation of natural law, but rather as the creation or annihilation of mass/energy by a transcendent agent. In reply to the objection that this account would violate the first law of thermo-dynamics, he distinguishes two forms of the principle -- one metaphysical, one scientific -- and aruges that a miracle would not violate the principle considered as a scientific law. The second assumption is that miracle testimony cannot serve as evidence for theism. Larmer demonstrates that the logical ties connecting the concept of miracle to theism need not imply that one must be a theist to evaluate miracle testimony properly. All that is required is that one is prepared to entertain theism as a hypothesis. Attacking these assumptions allows Larmer to show that Humean balance-of-probabilities arguments, based on a presumed conflict between evidence which establishes belief in the laws of nature and evidence in favour of miracles, miss the point if miracles need not be defined as violations of the laws of nature. He argues that, in the absence of a general argument demonstrating that the testimonial evidence in favour of miracles conflicts with the evidence for the laws of nature, it is up to the atheist to demonstrate, on a case-by-case basis, why the testimonial evidence is to be rejected. His conclusion is that, contrary to what is usually thought, the burden of proof lies not upon the shoulders of the theist, but upon the shoulders of the atheist.