I want to submit a paper to Elsevier (yes, I know, evil me - there are specific reasons). Elsevier allows sharing preprints "anywhere at any time". Later, one is allowed to "update a preprint with the accepted manuscript", provided the accepted manuscript is under a CC-BY-NC-ND license. There is no discussion of which license the preprint should be under. (https://www.elsevier.com/about/policies/sharing)

My questions is this: is it 1) technically possible on arxiv, 2) legally valid and 3) reasonable to submit a preprint with a more permissible license - say, CC-BY - and then switch the license to CC-BY-NC-ND upon updating the preprint with the accepted manuscript?

Previous questions on similar topics include [1], [2], but they do not discuss this specific issue.

  • Generally speaking, a permissive license needs to be honored going forward. You can't take away rights you previously granted. Not an answer since IANAL.
    – Buffy
    Jul 16, 2021 at 12:33
  • 2
    @Buffy Hmm... but I am not really taking away any rights, am I? The preprint would still be available under CC-BY; it's only the accepted version that would be under the more restrictive license. Sorry if that doesn't make sense at all, I find this whole topic spectacularly confusing.
    – Eike P.
    Jul 16, 2021 at 12:46
  • 1
    I don't know the answer to your question, but looking at the "Quick Definitions" section in your Elsevier link suggests that the existing arXiv version is a "Preprint" and the one you update after review would be an "Accepted Manuscript", so Elsevier will probably be fine with what you propose. Whether it would comply with the license of the existing version is an interesting question.
    – GoodDeeds
    Jul 16, 2021 at 12:54
  • 1
    @GoodDeeds, it might be wise to ask Elsevier. I think they are a bit fussy.
    – Buffy
    Jul 16, 2021 at 12:57
  • 1
    @Buffy Here's an answer on law.SX concerning the non-derivative question. Apparently, the license really forbids the usual way of working with an article, but copyright law may allow it nevertheless as "fair use": law.stackexchange.com/q/51333/34162.
    – Eike P.
    Jul 16, 2021 at 13:25

1 Answer 1


(Disclaimer: IANAL. Well, in the UK, lawyer isn't a regulated profession, only solicitor and barrister are, so I can claim I'm a lawyer all I want.)

Short answer: You can do whatever you want with later versions of the manuscript, but previous versions will stay with the licence you gave them.

This is seen a lot with open-source computer programs. To change the licence they need to contact all of the people responsible for the code, to request the change in licence. In your case, there is only one person to ask---you---and you presumably agree. You are the copyright holder, so you can decide.

So later versions of the article can be under a more restrictive licence, or in the case of the typeset versions of the article with the journal's markup, no CC licence at all. But v1 that you posted on the arXiv will forever be CC-BY. Notice that the descriptions on the arXiv website talk about reusers. Since you are not assigning copyright to someone else, you still hold the copyright, you just allow (depending on licence) others to do stuff with your work.

I personally always choose arXiv perpetual non-exclusive licence at the v1 stage, to keep my options open, as it's easier to free a paper than to enclose it.

  • Interesting. So you choose the arXiv perpetual non-exclusive license and then, later on, switch to a different license for that very same document? That might actually be a good idea. This way I can wait to see what Elsevier makes me sign and then, later on, switch to a more permissive license for the preprint, if there's nothing that prevents me from doing so.
    – Eike P.
    Jul 17, 2021 at 11:38
  • @jhin I rarely need to switch, to be honest. It's not often that journals require a different licence, you can licence the same work differently to the arXiv and the journal, and in mathematics there's not much to gain for others from going from standard copyright terms and CC-BY. Jul 17, 2021 at 21:09
  • Ah right, for pure maths I can imagine it's not super relevant. I'm mostly concerned about the figures in the paper, which I would like people to be able to reuse freely. (Especially after I recently contacted several journals for permissions to reuse figures, and most did not even respond...)
    – Eike P.
    Jul 17, 2021 at 21:45
  • I've not needed to worry about this, true. Images could be released into the public domain separately from the rest of the paper, and that might even be a good idea. Some journals are a bit snippy about the text being CCed, but images are generally not that much of an issue, because often the author does not hold the copyright for them anyway. Jul 17, 2021 at 21:49

You must log in to answer this question.

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