A follow-up question to this: Citing a math theorem with errors

Let's say a paper A has some error in a proof of a theorem and then I publish a paper B citing the theorem and its proof and also indicate the error in the proof and provide a corrected proof.

And now say I want to use this theorem in paper A for another paper C, what do I do? (Let's say the paper wasn't corrected.)

  1. Copy all the stuff I said in my previous paper B to my new paper C, as if paper B never existed. (assume paper B is otherwise irrelevant to paper C.)
  2. Cite both papers A and B.
  3. Cite only B, since it also has a citation of A.
  4. Other?
  • 4
    Please don't engage in edit wars. This question has been locked for cooldown.
    – eykanal
    Commented Oct 20, 2021 at 18:07

1 Answer 1


I would recommend citing both your previous paper, and the other one, as well, explaining the genesis of the result correctly proven in your previous paper. That way someone just looking at the bibliography (without following the chain of past references!) has a better idea what's going on.

Sure, either way you are probably avoiding any charge of not giving credit, etc., but you're not explaining things as you (easily) could if you don't mention the other paper.

Further, just on appearances alone, you look more credible when you cite other people, not just yourself. :) At the very least, it shows that you are not a solipsist. :)

  • Also follow-up: if I cite this theorem in another paper D, then what do I do? paper E? paper F? (like even up to paper F, there's no correction published for paper A. Maybe author is dead.)
    – BCLC
    Commented Oct 20, 2021 at 17:30
  • 1
    Regarding your follow-up: even when the correction itself becomes "old news" (which is non-trivial to judge...), I'd tend to cite the original + my correction of it. Not create a long, unhelpful (misleading) chain of backward references. When it's so-old news that everyone knows its story, and it's become folkloric, sure, you can stop citing. Still, sometimes people are not aware that various "modern" concepts may have arisen a looooong time ago. :) Commented Oct 20, 2021 at 17:36
  • 3
    For a/the thing already set up in A and B, even if cited in C, if paper D does not use anything else from C, just cite A and B. In particular, if the citation is in reference to that particular result, citing A and B correctly explains where it came from. To my taste, piling up chains of references is at best unhelpful, and sometimes is like playing the game "telephone". But tastes vary. Commented Oct 20, 2021 at 17:39

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .