Consider that a paper is published in a journal. It contains some insignificant errors which are not worth asking for an erratum. I was wondering in such a case would it be acceptable to fix those errors in the preprint version (for example on arXiv.org)? Since in this case it will be sort of a “post-print” version of the paper and technically an improvement on the paper, would it create a problem with the journal in a typical situation?

Edit: Let’s assume that the editor agrees that the errors are too minor for an erratum. What I want to know is if there will be any legal issues and more importantly: if it is a good idea scientifically?

  • 2
    The journal won't care. If it's a good idea or not is 100% opinion. Feb 20, 2021 at 4:10
  • I actually do this on occasion to winnow down typos and other minor errors in published versions that do not warrant an erratum. It is essential in such cases to make it clear (on arXiv for instance) that this is a post-publication version which correct minor errors. I've never had issues with a journal for this type of update. Feb 21, 2021 at 22:41

3 Answers 3


There is nothing wrong with updating a paper on the arxiv after it has been accepted or published, from a legal point of view. The paper on the arxiv is not necessarily identical to the published one. (In fact, it is rather the opposite: Changes introduced during the copyediting process are usually something which you are not supposed to include in the journal version.)

Whether this is a smart thing is a different question. As has been pointed out, people might get confused if the published version and the arxiv version differ, especially if they differ in terms of content: If someone cites the arxiv version, it might be that the same result is not contained in the published version, yet the reference might be at some point changed to the published paper. However, this is something which can (at least partly) be remedied by putting a suitable comment on the arxiv.

(Note that there are plenty of papers where the arxiv version differs from the published version, mostly because the authors did not bother updating it. Not that this is recommended, but it is common, so people often are aware that those versions differ.)


I would write the editor of the journal explaining the errors and the necessary corrections and asking what they want to do.

If they agree with you that the errors are too minor to warrant a published erratum they may want to post the erratum on their journal page, or wherever the article is available for download.

If they choose to do nothing, ask about fixing the preprint on arXiv (where you would of course refer to the published version).

I once corrected a minor error in a paper in a second paper I published several years later on the same topic. I've no idea whether anyone ever noticed either the first error or the correction.

  • 2
    This answer ignores the premise, which is that the errors are insignificant. Feb 20, 2021 at 4:08
  • @AnonymousPhysicist Judgement call, but I disagree. If the OP is worried enough to ask, then they should ask the editor. If they really don't matter the issue will be resolved and the OP can move on. Feb 20, 2021 at 12:45

This isn't a good idea. The problem is that once you make a change, if it's spotted, nobody (including you, six months in the future) knows what else might be different. E.g. a reader downloads the manuscript from the journal's website, prints it out, and covers it all over with comments. Next day they go to arXiv and download the manuscript. Then they notice that there's a word in the arXiv version and the published version that's spelt differently. It's obvious which one is correct, but still: given that one of the manuscripts has been revised, what else in the manuscript could've changed? Without crawling through the manuscript with a fine-toothed comb, they would have no way of knowing if there were any non-minor changes.

If it's really insignificant, just don't make any changes.

  • 4
    The way ArXiv is designed assumes the preprints will change. So the designers, at least, do not agree with you. Also, you can just run latexdiff on the latex source, which is usually downloadable from ArXiv. Feb 20, 2021 at 4:09

You must log in to answer this question.

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