Gelfand-Mazur — every real unital Banach algebra where every non-zero element is invertible is isomorphic to either $\mathbb{R}$, $\mathbb{C}$ or $\mathbb{H}$ — was first published without proof by Mazur. Mazur had a (rather short) proof, but the editor demanded he shortened it further. He refused to shorten it, and so it was published without proof.
Later Gelfand published a proof of a weaker version (only for complex commutative Banach algebras), probably without knowing about Mazur's result.