I've stumbled upon this article a while ago, and I've also seen refutations for the Muzio Gambit (Joseph Kling & Bernard Horwitz - Chess Studies on Endings of Games). I'm wondering what other gambits are proven to be unsound.
Edit: I'm looking for refuted gambits in the strict sense of the word, followed by proof. Even though a gambit might be playable against someone who doesn't know its refutation, the gambit is still considered refuted if it is proven that the side who initiated the gambit loses in all variations worth looking at against best play. So yes, I'm looking for scientific proof.
Also, if a database of almost all games played in a particular gambit shows that the side that initiated the gambit wins almost all the time, this doesn't mean that there isn't a way for that gambit to be refuted, it just means that if there is a way, most people haven't played it.
Above everything, I'm looking for material to learn from. An article that says that Vasik Rajlich or someone else has used a computer cluster and came to a conclusion regarding a certain position, how will I benefit from it if there is no proof? (And please note that that King's-Gambit-busted-by-Rajlich article was in fact a Chessbase April Fools' joke!)