This is pretty well worked over territory, but I wanted to run through some options for dealing with cases where we forget our evidence.

For convenience, let _p_, _q_ and _r_ denote the following propositions. We’ll assume these are true, which does involve taking sides in a dispute about just what constitutes a reign. The Charles’s in the propositions are the 17th Century Kings of England and Scotland.

_p_ = Charles II reigned for longer than Charles I.
_q_ = Charles II reigned for 24 years and 243 days, and Charles I reigned for 23 years and 309 days.
_r_ = Fred’s Year 10 history textbook says that _q_.

As a schoolboy, Fred reads that _q_ in his Year 10 high school history textbook. He subsequently comes to believe, and we’d naturally say know, that _q_, and quickly infers _p_. Some years later he forgets _r_ and _q_, but still believes _p_. His belief that _p_ is a strictly memorial belief – he hasn’t come to rebelieve _p_ via some lesser method. Perhaps he remembers some other things as well, say that Charles I reigned from 1625 to 1649 and Charles II reigned from 1660 to 1685, but nothing that entails _p_. Still, he remembers (as we’d ordinarily say) that _p_.

What exactly should we say about Fred’s knowledge and evidence in this case? I take it that there are four options on the table.

Option One. Since Fred no longer has evidence that _p_, he does not know that _p_.

Option Two. Since Fred came to know that _p_, _p_ became part of his evidence. So _p_ is part of Fred’s knowledge, it is also part of his evidence.

Option Three. Fred knows that _p_, since he learned that _p_ and has not forgotten it. But since all his evidence that _p_ has been lost to the mists of time, he has no evidence for _p_.

Option Four. Fred knows that _p_, since he learned that _p_ and has not forgotten it. Fred’s evidence that _p_ is _q_, or perhaps that _r_. The evidence is not accessible, and if Fred came to believe something else supported by _q_ (with no other reason) this could not be knowledge because it was not based in _q_, but _q_ is still his evidence for _p_.

I think I prefer Option Three.

Option One seems absurd – since it requires that we remember our evidence if we are to preserve knowledge, and that is too strong.

Option Two seems to make it the case that _p_ is self-evident for Fred, and this seems wrong. Not because it is contingent, or anything of the sort. Nor because it is a proposition about which Fred could be wrong. It is self-evident (to me, now) that there is a computer on the table in front of me. That is, that there is a computer on the table is something I know simply because it is part of my evidence. Perhaps it is self-evident that all men are created equal. (I go back and forth on this.) But it isn’t self-evident that _p_.

Options Three and Four are both odd, but I think in both cases the oddities can be lived with.

Option Three is odd because it says that we can know something without having any evidence for it. It might seem plausible that knowing things requires that we have evidence for them. But option three is consistent with the only slightly weaker claim that knowing something requires having, or having had, evidence for it. In any case, the possibility of innate knowledge suggests that this principle is a little strong.

Option Four is odd because it says that we have evidence that we can’t use for any purpose. A natural account of what it is to lose evidence is that we had the evidence and no longer have it. And it is very natural to think that when Fred forgot about _q_, he lost his evidence for _p_. But if we accept this option, we can’t say both of those things.

I think all things considered I’d prefer Option Three, but Option Four doesn’t look clearly wrong. Options One and Two do, at least to me, look pretty badly wrong. Perhaps though there is another option that I’m forgetting. Any thoughts?