InNou2s (June, 1987), I have presented a detailed Bayesian analysis of Hume on miracles, including the crucial point that P(W/E) [E: Miracle m occurred] is the product of two factors, one of which may be exceedingly small, as when many other miracles could be substituted for m. Unfortunately, I failed to see then how that point leads to the radical objection to Hume mentioned above. I am indebted to the Editor for drawing my attention to three important articles in the Philosophical Quarterly by J. H. Sobel (1987), D. Owen (1987) and P. Dawid & D. Gillies (1989). But none of these seems to have noticed the very existence of the second fraction noted above. Dawid and Gillies, for instance, say explicitly (p. 63) that P(W/~-E) cannot be very low because faked miracles are common, because of mankind's propensity to the marvellous, etc. But, of course, under circumstances where there are millions of magician-produced 'miracles' with similar religion-inspiring, marvel-inducing etc., potential a second factor of indefinite minuteness (i.e. the value of the probability for the false witness to pick m, instead of one of the possible alternatives), may render the product P(W/--E) so small that P(E/W) could approach 1, and fully establish the credibility of E. Otherwise, these philosophers make numerous interesting points most of which I have not touched upon. Then again, my present paper contains what seem to me significant results which were outside the scope of their inquiry. These include the seemingly paradoxical conclusion that P(E/Y) ~P(E/W), as well as that the unconditional support for a hypothesis by a very trustworthy witness may sometimes lower that hypothesis's probability.