¹ Although it may be objected that Swinburne's credulity principle (The Existence of God, Oxford: Clarendon Press, 1979, pp. 254–71, 274–6)Google Scholar and Kenny's allowance for holding beliefs as basic on testimony (Faith and Reason, New York: Columbia University Press, 1983 p. 64)Google Scholar make them nonevidentialists, both obviously think that proper testimony can be traced back to evidence, and Swinburne expects that normal rational people will rely on evidence, not credulity.

² This argument comes at the end of the fourth and final chapter of the book, using premises argued for in the third chapter, which arguments are in their turn based on theory developed in chapters one and two.

³ The claim that natural theology is possible means that it is possible to offer strong evidence for the existence of God and arguments against disproofs of the existence of God (63).

⁴ I wish to call attention to the intentional contexts in this argument even though I will largely ignore problems these can create in assessing the validity of arguments and try to make my points in ways that do not depend on technicalities. The relevant remark at the moment is that I will understand ‘—— is rational (for me)’ as an operator with wider scope here than the belief operator, but the basic logical operators, such as implication and negation have the widest scope. Thus, using brackets and parentheses, I understand the first premise as follows: {[S believes (p & q & r)] is rational for S} only if [S believes that (God exists)], where p & q & r are understood to be some suitable statement of the articles of the Christian faith and not variables.

⁵ It is valid assuming we forestall certain kinds of counterexamples that arise when we quantify into belief contexts. Accordingly, we assume that the substituends of p and q are not only the same proposition, but the same formulation of the same proposition in each of its occurrences.

⁶ Note that I have changed the statement of the premise from ‘natural theology can be carried out successfully’ to ‘God's existence is provable’, because it does no violence to Kenny's intended argument and it makes the logical structure of the argument immediately apparent. Perhaps it should also be noted that Kenny intends this argument to hold for negative outcomes as well as positive, i.e. a proof that God does not exist would also count as a ‘successful’ carrying out of natural theology for purposes of Kenny's argument.

⁷ I believe that my points hold even if ‘provable’ is understood to mean something weaker than mathematically provable. Suppose that it means supportable by a persuasive argument. It is still difficult to see why I need to have the arguments myself.

⁸ One has to be careful about terminology here. I follow Alvin Plantinga's articles, some of which helped provoke this book of Kenny's, in using the term ‘Evidentialism’ to refer the thesis that belief in God is not rational unless it is based on the evidence of other beliefs. Cf. ‘Reason and Belief in God’, in Faith and Rationality, Plantinga, and Wolterstorff, , eds. (Notre Dame, 1983), pp. 16–93.Google Scholar

⁹ This is perhaps to link rationality with sanity, as we commonly do. Saying (and believing) what is true and justified according to the canons of rationality is not enough to make one rational, as is nicely illustrated by Kierkegaard's story in the Concluding Unscientific Postscript (Princeton, 1941) (174)Google Scholar about an escapee from an asylum, who decides to prove his sanity by always speaking the truth. To remind himself, he puts a ball into the tail pocket of his coat. As he walks, the ball bumps him with each step, and each time the ball strikes him he says, ‘Bang, the earth is round’. He speaks and believes the truth. Is he rational?

¹⁰ Plantinga, Alvin, ‘Is Belief in God Properly Basic?’ Nous, XV, I (03 1981), 41–51CrossRefGoogle Scholar, and ‘The Reformed Objection to Natural Theology’, Proceedings of the American Catholic Philosophical Association, 1980.Google Scholar

¹¹ Following Plantinga (Faith and Rationality,1983, pp. 78–82)Google Scholar I distinguish between evidence and grounds. Evidence consists of other beliefs which support the belief in question, while grounds are a basis in experience, e.g. experiences of various sorts, for beliefs but are not themselves beliefs.

¹² This confusion is clearly exposed by Mavrodes, George in his Belief in God (Random House, 1970), Ch. II.Google Scholar Mavrodes also shows that a clearer exposition of the notion of proof yields less sweeping claims about the epistemic virtues of proofs (Ch. II, Section 6).

¹³ Also shown by Mavrodes, George, op. cit. Ch. II, Sections 5–6.Google Scholar Suppose Fermat had written down a proof of his famous ‘Last Theorem,’ but that the proof had been thrown out with the trash by a servant. May we conclude that Fermat's Last Theorem has been demonstratively proven? Has it been proven for anyone? I should think not, and that the reason why is that Fermat proved it to no one but himself (in our story), and now the proof has been lost and cannot be duplicated. Thus the Theorem cannot be proven to anyone today, unless a proof is discovered.

¹⁴ This claim is not supported by Kenny's, terminology on pp. 62–5Google Scholar, where he constantly uses terms like ‘valid’, ‘sound’ and ‘proof’, terms whose applicability to inductive arguments is unclear, to put it kindly.

¹⁵ I wish to thank my colleagues at Calvin College for help with an earlier draft of this paper. Thomas V. Morris made me rewrite Section III. John Hare, Alvin Plantinga, and Del Ratzsch gave me helpful criticisms of the rewritten version, and the referee also gave me a number of helpful criticisms and suggestions.