But in order to increase the probability against the testimony of witnesses, let us suppose, that the fact, which they affirm, instead of being only marvelous, is really miraculous; and suppose also, that the testimony considered apart and in itself, amounts to an entire proof; in that case, there is proof against proof, of which the strongest must prevail, but still with a diminution of its force, in proportion to that of its antagonist.
A miracle is a violation of a law of nature; and as a firm and unalterable experience has established these laws, the proof against a miracle, from the very nature of the fact, is as entire as any argument from experience can possibly be imagined.
Andrew suggests that Hume is trying to have it two ways: he's trying to treat his argument as a priori in some places and a posteriori in others. Of the above passage, he says:
In the first part of the quote he seems to concede that, in principle anyway, testimony to the occurrence of a miracle could amount to a proof (note that, for Hume, the term “proof” simply denotes “such arguments from experience as leave no room for doubt or opposition” — Hume would never say that the epistemic probability of the occurrence of some matter of fact, as opposed to some logical or mathematical demonstration, was strictly 1), but that this proof would encounter an opposite proof, also deriving from our experience, of the hitherto observed course of nature (dead men don’t rise, etc.). But then Hume moves on to say that this opposite proof, derived from the “firm and unalterable experience” of the conformity of the world to some law of nature, is “as entire as any argument from experience can possibly be imagined.” The proponent of some miraculous event will have to maintain then that the evidence for the occurrence of some miraculous event surpasses the opposite proof from our past observations of the course of nature – which, I reiterate, is supposed to be “as entire as any argument from experience can possibly be imagined.” But how, if this kind of evidence can be marshaled for a purportedly miraculous event, can this event retain the appellation “miracle”? By definition the evidence against the occurrence of any miraculous event is as strong as can be imagined.
We don't have to speculate about what Hume would say to this, because the point was raised by Campbell and Hume replied, albeit briefly. Hume's reply (via Hugh Blair), which makes a bit more clear his account of proof:
The proof against a miracle, as it is founded on invariable experience, is of that species or kind of proof, which is full and certain when taken alone, because it implies no doubt, as is the case with all probabilities; but there are degrees of this species, and when a weaker proof is opposed to a stronger it is overcome.
That is, it's important to keep in mind (as Andrew does) that the notion of proof is that mentioned earlier in the Enquiry, in Section VI, in which it is a form of causal reasoning based on completely invariable experience. So in Humean terms it is entirely possible to have a full and entire proof of something and be wrong: its being full and entire has to do with the invariableness of the succession on which it is based. That there is a full proof (from testimony) of miracles is supposed for the sake of argument; all Hume does in Part I is to argue that this would have to overcome the full proof experience provides for the exceptionlessness of the laws of nature, in which case the psychologically more forceful proof would win. Thus I think Fogelin is right that there is no argument at all against miracles in Part I: the scenario envisaged is one in which you have a full and entire proof that miracles don't happen (the invariable course of nature) and an opposing proof, also full and entire, that testimony to a particular miracle can be trusted (by hypothesis). That this is the case is pretty much stated explicitly in the first paragraph of Part II:
In the foregoing reasoning we have supposed, that the testimony upon which a miracle is founded, may possibly amount to an entire proof, and tha the falsehood of that testimony would be a real prodigy: But it is easy to show, that we have been a great deal too liberal in our concession, and that there never was a miraculous event established on so full an evidence.
But if this is the case, there really is no contradiction or confusion in Hume: Part II is devoted to giving a posteriori reasons that, individually considered, explain why testimony for miracles never has reached the point of proof, much less of a proof capable of overtopping the proof of the laws of nature and, collectively speaking, give a reason to think that it could never do so in some cases -- the features of human nature (e.g., our excessive love of surprise and novelty, our passions and interests, our gullibility) conspire together to make it extraordinarily unlikely that any testimony could ever reach the point required, when they are likely to be operative. Note that this is not a proof a priori; despite the strong conclusion, it is based entirely on a posteriori reasoning. That is, the conclusion is not that there is no possible way testimony for miracles could ever reach the level of proof, but that our experience of how human nature operates in matters of religion is such that we are entirely reasonable in concluding that no human testimony can have such force as to prove a miracle for the purposes of religion (Hume explicitly makes this qualification). Thus Hume goes on to consider a case where you could have a miracle -- eight days of darkness -- that would avoid the causes for suspicion and doubt that arise in the case of religion.
So I don't think Hume is, in fact, confused on the matter; in fact, I think his argument is very precise (more precise, I think, than most readers of it have wanted to read it). The structure of the Section is quite clearly and explicitly laid out by Hume.
Part I: Supposing that testimony for miracles can amount to a full proof, how strong would the proof have to be in order to establish that the miracle had actually happened. (Answer: Very, very strong.)
Part II: But the testimony for miracles has never reached such a level, and in matters of religion we have strong reason to think it cannot. (This conclusion was made easier to reach by the reasoning Part I, which establishes a very high standard.)
Lewis Powell has an interesting response to Andrew's post at "Horseless Telegraph." But I think Andrew's interpretation is actually closer than Lewis's, for reasons that should be clear. I think, however, that it's a mistake to read the strong conclusion ("we may establish it as a maxim, that no human testimony can have such force as to prove a miracle, and make it a just foundation for any such system of religion") as a priori, since this conclusion is clearly based on the discussion of a posteriori reasons for thinking that testimony in religious matters can't be trusted. Thus Hume is a posteriori throughout (this is true even with the argument in Part I, which is also supposed to be based on our experience of human nature).
UPDATE: I should say, with regard to Lewis's interpretation, that I think his reconstruction of the skeleton of the argument of Part I goes wrong at (1): Hume never says that the proof for the laws of nature is a maximally strong proof, only that it is direct and full. And this makes sense; because while all laws of nature have direct and full proof from experience (this is why we classify them as laws of nature), they can't all be equally strong, because they are not monolithic, but deal with very different sorts of events. Rather, Hume's point is that there is nothing lacking to the experience to make it a proof. Since the testimonial proof is also supposed -- purely for the sake of argument -- to be full and entire, Hume's final paragraph in Part I makes no sense if it is not open to one of the proofs being superior to the other -- as he in fact says is possible, in response to Campbell, to Hugh Blair.
