Maths may not have a replication crisis like some other areas, but when I go to maths events, it seems widely agreed there are far too many papers with incorrect theorems, it's just no-one cares about those papers, so it doesn't matter.
It turns out to be very, very common (as discussed in the linked article) that when someone really carefully reads old papers, the proofs turn out to be wrong. They are often fixable, but the point of the paper was to prove the result, not just state it. What tends to save these papers is that enough extra results have been built on top of them, and (usually), if there had been an issue, it would have showed up as an inconsistency in one of the later results.
The trunk is (probably) solid, but there are a lot of rotten leaves, and even the odd branch.
It turns out to be very, very common (as discussed in the linked article) that when someone really carefully reads old papers, the proofs turn out to be wrong. They are often fixable, but the point of the paper was to prove the result, not just state it. What tends to save these papers is that enough extra results have been built on top of them, and (usually), if there had been an issue, it would have showed up as an inconsistency in one of the later results.
The trunk is (probably) solid, but there are a lot of rotten leaves, and even the odd branch.