¹ Clarendon Press, revised edition 1991.

² Ibid. pp. 133f.

³ Swinburne cites as an example of spatial order the arrangement of books alphabetically in a library, Ibid. p. 133; the movements of two individuals who are dancing in step with one another would count as an example of temporal order.

⁴ Swinburne remarks that this form of the argument ‘seems to me a much stronger one’ (p. 136). However, this remark may need qualification in the light of his article in J. Leslie (ed.), Philosophy and Physical Cosmology (an abridged form of which appears in Appendix B of the 1991 edition of The Existence of God). There he argues that the argument from spatial order can now be stated more forcefully than he had supposed, in view of recent developments in cosmology, which make possible an argument from ‘fine-tuning’.

⁵ On the traditional view, there are some things which exceed even the power of God, whether because these things are logically impossible in themselves (for example, making 2+2 = 5) or because they are logically impossible at any rate for God (for example, performing an action which presupposes embodiment). Talk of powers which are unbounded should be read in the light of such qualifications.

⁶ See Mackie, John, The Miracle of Theism (Oxford: University Press, 1982), p. 149Google Scholar. Swinburne replies in Appendix A of the revised edition of The Existence of God. We might consider this criticism as a response to (3).

⁷ Of course, there are precedents for such an approach. For instance, when Hume invites the proponent of the argument from design to become ‘a perfect anthropomorphite’, he is suggesting that there is an inconsistency between the argument's reliance upon an analogy between the products of divine and human agency and its resolve to establish the existence of an omnipotent God.

⁸ It may be objected: but perhaps the particles that now appear to us to be basic will be judged relatively superficial given further advances in our understanding. Swinburne allows that this is possible, but observes that we should expect any such progress to involve the discovery of particles which again belong to a relatively small number of kinds (op. cit. p. 140).

⁹ Of course, it cannot simply be reduced to the existence of these building blocks: there are emergent phenomena (consciousness for example) of which we can know nothing merely by studying basic particles.

¹⁰ Op. cit. p. 145.

¹¹ It seems to me that it is also possible to provide a theoretical rationale for this association of resemblance and shared causal origin; I have tried to do so in an unpublished PhD thesis (Oxford, 1991).

¹² Of course, this is not an analytic connection: the same thing exercising the same powers may give rise to quite different kinds of effect in different kinds of environment; the point is just that there is no cause for surprise if the effects of a single source should resemble one another closely.

¹³ This argument appears on pp. 140–1, op. cit.

¹⁴ Some may prefer to draw the boundaries of scientific explanation more generously, and classify in these terms any explanation which makes no reference to an intentional agent. If we proceed in this way, we will need to say: perhaps there are only two kinds of explanation, but scientific explanations just as much as personal can account for the fundamental regularities in the world.

¹⁵ See p. 146.

¹⁶ See p. 136.

¹⁷ Thus Swinburne remarks: ‘the order of the world is evidence of the existence of God… because its occurrence would be very improbable a priori…’ (p. 147).

¹⁸ See the remark in the passage just quoted: ‘The universe might so naturally have been chaotic.’

¹⁹ See The Miracle of Theism, p. 148.

²⁰ See Religious Studies (1983), 385f. A shortened version of this paper appears in Appendix A in the revised edition of The Existence of God.

²¹ See the passage I cited just now for both of these points. In fact, it is to the character of scientific laws that he turns in order to give content to the notion of simplicity. For instance, on p. 56, he writes: ‘There must be a criterion to choose between the infinite number of theories which are equally successful in predicting the observations already made… The history of science reveals that, in the absence of background knowledge, that criterion is basically the criterion of simplicity.’

²² See p. 53. By the ‘prior’ probability of a hypothesis Swinburne means its probability independent of (or ‘prior’ to consideration of) the evidence.

²³ In fact, the two objections are closely related. Mackie begins with what Swinburne says about chaos and wonders how Swinburne is entitled to consider future order at all likely. This further objection begins with what Swinburne says about order (and the probability of simplicity), and wonders how he can consider chaos very likely.

²⁴ The Existence of God, revised edn, p. 299.

²⁵ See p. 94.

²⁶ This consideration is implicit in Bayes's Theorem, according to which P(h/e.k) depends upon the ratio of two prior probabilities: P(h/k)/P(e/k). It follows then that even if e is to be expected (that is, even if P(e/k) is quite high), the probability of the hypothesis may remain high, providing that P(h/k) is high. And as I have just suggested, these two probabilities may be positively related: if h is likely anyway, and if e is likely given h, then e is quite likely anyway.

²⁷ I would like to thank Peter Byrne and Brian Davies for their comments on earlier drafts of this paper, which have done much to improve it. I would also like to express my gratitude to Professor Swinburne for his kindness in discussing these matters with me on many occasions; although he has saved me from many errors, no doubt some survive!