Just to clarify some things:

It was not the Smithsonian, but the Library of Congress, that dealt with the fellow whose recording was used in O Brother Where Art Thou. In almost all cases where we are dealing with old field recordings, it's the Library. The Smithsonian's folk music archives are primarily derived from Folkways and the other Moe Asch labels on the one hand, and their great festivals on the other. Because of that, most of their stuff came with explicit or implicit releases allowing them commercial use of the material; obviously Asch had such agreements with his artists, as Folkways was at least nominally a commercial enterprise, and the festival offers a pretty good gig in exchange for a signed release. But keep in mind also that every single year they have stellar, grammy-level talent like Beausoleil, and don't often release that stuff on CDs, so I suspect they are not getting artists to sign releases for every performance, and consequently cannot release many of the festival performances on CD.

The Library's a different story. We have collectors going back to the 1890s, when really no one was thinking about these issues. We have the first ethnographic field recordings in the world, in fact. Lomax and the other collectors of the 30s and 40s didn't usually obtain signed releases either. Many of our other collections came from amateurs on the one hand, and academics on the other--people who didn't even think about commercial issues when amassing their collections.

Even so, the Library does have to seek permission for every item we publish online or release on a CD, unless a signed release was obtained at the time of collection. The typical formula is to send a letter to the last known address requesting permission; if it comes back "unknown," you have made a good-faith attempt, and you file that returned letter as proof of the good-faith attempt. (If a researcher from outside the Library wants to publish any of our materials, then it falls to them to make the same kind of attempt.)

After the attempt is made, if there is no objection from an informant or descendant, the Library usually goes ahead with the publication. If it's an online presentation, material can always be removed if heirs turn up with objections. If it's a CD release, they're usually small batches and as a practical matter I don't think there's ever been a complaint to the Folklife Center after an item was published. (There have been objections from survivors of deceased performers that have prevented certain items from being played on the radio, on at least one occasion that I remember.)   

Keep in mind that the Library never makes any kind of profit, and is by nature educational and public. Any money we earn goes toward furthering our educational and archival missions, which are undertaken on behalf of the American people. Moreover, as several people have commented, the sums involved are tiny. The amount the singer would have made for mechanical royalties on a track of typical LC CD release wouldn't pay for the trouble of collecting it. So it would be hard to imagine people claiming any kind of "damages" unless (1) they are professional performers and the Library release might compete with one of their own CDs or (2) something was published that was downright embarrassing. Since the Library doesn't want to publish anything downright embarrassing, and knows how to avoid the other situation, the chances of a real conflict are slim. Nevertheless, the legal niceties are scrupulously observed.

All that said, Bob, I don't see anything wrong with what you're doing, either ethically or legally. (I'm not a lawyer, though!) Until you start publishing or selling the material, no one could reasonably object. I agree that if you're going to give copies away, you should check with the performers if possible...especially with anyone who is a pro performer and has "official" CDs on the market. I think they'd be foolish to imagine your giveaways were competing, but it would be honorable for you to check with them.

Finally, I agree that a university library or other archive might be interested in your collection.