1. John Stuart Mill ‘On Liberty’, in J. Robson et al. (eds), The Collected Works of John Stuart Mill (Toronto: Toronto University Press, 1963–1991), XVIII, 229–230.

2. Hick endorses an argument similar to Mill's in a number of places, e.g. J. Hick ‘On grading religions’, Religious Studies, 17 (1981), 451–467, and idem ‘The epistemological challenge of religious pluralism’, Faith and Philosophy, 14 (1997), 277–286. However, he does little to articulate the argument rather than merely endorse it. A slightly fuller treatment is given by D. Basinger Religious Diversity: A Philosophical Assessment (Aldershot: Ashgate, 2002), esp. 69–73. See also Plantinga's discussion of Hick in his ‘Ad Hick’, Faith and Philosophy, 14 (1997), 295–298.

3. For most believers, the religious beliefs, if any, that one claims are epistemically justified are exactly the same as those one claims are knowledge. However, a religious believer may hold the position that certain of his/her religious beliefs are epistemically justified to an extent but one not sufficient to raise them to the standard necessary for knowledge. It is not immediately clear what Mill's argument would suggest such people think about these beliefs. Given what Mill says elsewhere (e.g. in ‘Theism’), it seems that he would wish to accord some religious beliefs this weakly justified/warranted status. Be that as it may, in the main text I frame Mill's argument as one in favour of downgrading one's estimation of the epistemic status of one's religious beliefs so that one does not count them as knowledge, as this would certainly be something Mill would encourage religious believers to do. It would be interesting to consider how far a Mill-inspired argument (not then, we have reason to believe from ‘Theism’, an argument Mill would actually have endorsed) could be pushed against those who were prepared to abandon claims to religious knowledge, but wished to hold the line at claims to a weaker epistemic justification for their religious beliefs. Space considerations do not permit me to explore this avenue at greater length.

4. A. Plantinga ‘Pluralism: a defense of religious exclusivism’, in P. Quinn and K Meeker (eds), The Philosophical Challenge of Religious Diversity (Oxford: Oxford University Press, 2000), 172–192, 188.

5. Of course a Jew or Muslim ‘Calvinist’ would need to adapt the terminology of the main text somewhat, but I take it that essentially the same structure could be made to fit any variant of theism.

6. See my ‘How a single personal revelation might not be a source of knowledge’, Religious Studies, 39 (2003), 347–357, and idem ‘How can I know I've perceived God?’, International Journal for Philosophy of Religion, 57 (2005), 105–121, for further discussion of some issues relevant here.

7. The word ‘alternative’ gets scare quotation marks in the main text as elements of it are compatible with the solution to which it is in some respects alternative.

8. The traditional free-will defence would obviously be less germane on determinism, but in fact a Calvinist solution to Mill's charge may be had going down the second tack even without supposing complete determinism; God just has to narrow the possible life histories of the Elect down to those – which may still number more than one – by which they'd never have come to false religious beliefs (or false religious beliefs over core items).

9. Gellman, J. ‘In defence of contented religious exclusivism’, Religious Studies, 36 (2000), 401–417CrossRefGoogle Scholar, 406–407.

10. It is at this point that the topic intersects with so-called evolutionary arguments against naturalism. See J. Beilby (ed.) Naturalism Defeated? (Ithaca NY: Cornell University Press, 2002).

11. Thus, Plantinga is wrong when he accuses the proponent of Mill's argument of being ‘hoist’ on his ‘own petard’; Plantinga ‘Pluralism: a defense of religious exclusivism’, 188. The proponent of the argument may admit that if he or she had been educated differently, then he or she would not now believe in the argument that he or she is using to show that knowledge claims in a certain area are untenable. However, if he or she admits this, then as long as he or she does not claim to know that he or she knows this argument to be sound, no hoisting on his or her own petard is entailed.

12. For further discussion of this, see my ‘“Byrne's” religious pluralism’, International Journal for Philosophy of Religion, 58 (2005), 37–54, 49 ff.

13. Plantinga makes a similar point in his ‘Ad Hick’, 298. See also Sher, G. ‘But I could be wrong’, Social Philosophy & Policy, 18 (2001), 220–240.CrossRefGoogle Scholar

14. I am grateful to Anita Renusch for having drawn my attention to some of the literature on this topic and for her critical comments on a first draft of this paper. I am also very grateful to an anonymous reviewer for this journal who provided a number of points, large and small, which redirected its argument in more fruitful directions at various stages.