I am curious to know what is the standard way (in AMS style) to cite a journal article that has an article number and/or a page range.

For instance, here is a BibTeX entry from MathSciNet of an article published in Proceedings A:

@article {MR4212412,
    AUTHOR = {Kallosh, Renata},
     TITLE = {M-theory, black holes and cosmology},
   JOURNAL = {Proc. A.},
  FJOURNAL = {Proceedings A},
    VOLUME = {477},
      YEAR = {2021},
    NUMBER = {2245},
     PAGES = {20200786, 15},
      ISSN = {1364-5021},
   MRCLASS = {83E30 (83C57 83E50)},
  MRNUMBER = {4212412},
       DOI = {10.1098/rspa.2020.0786},
       URL = {https://doi.org/10.1098/rspa.2020.0786}.

As you can see, the PAGES field contains both the article number 20200786 as well as the number of pages 15. On the other hand, it seems more logical to me to just write 1--15, because the article in question does have page numbers, and they run from 1 to 15. (The latter approach is in accordance with APA style; see here.)

In other cases, like articles published in Physical Review Letters (which is not covered by MathSciNet), an article has a six-digit number (e.g., 238104) and the $n$th page is identified by 238104-$n$.

How would you cite these articles? The AMS Style Guide completely ignores the issue.

EDIT. Since I posted the question, the format of the BibTeX entry of an article with an article number has changed in MathSciNet. The number itself is now preceded by the expression "Paper no.". For instance, the PAGES field of the given Proceedings A article now reads as follows:

PAGES = {Paper No. 20200786, 15}
  • 3
    $\begingroup$ My advice is to follow the MathSciNet style. $\endgroup$ Mar 29, 2021 at 13:20
  • 5
    $\begingroup$ @Iosif No, please. This advice leads to all sorts of problems with the way references are listed. Many journals just take whatever the authors send and publish it without careful checking. $\endgroup$ Mar 29, 2021 at 14:40
  • 6
    $\begingroup$ I want to voice my objection to closing this question. Unfortunately, more and more journals are switching to this format, in particular, Elsevier has already switched some of its mathematics journals, so mathematicians have better figure out a good way to cite these articles. $\endgroup$ Mar 29, 2021 at 19:08
  • 3
    $\begingroup$ I'd change it to PAGES = {20200786 (15 pages)}, $\endgroup$ Mar 29, 2021 at 19:23
  • 3
    $\begingroup$ The MathSciNet format is wrong, simply because 20200786 is not a page number. It may happen to work in a particular style the AMS is presumably using, but a different BibTeX style may well format it as the nonsense “pp. 20200786, 15” (or “pp. paper No. 20200786, 15” after the update). The only reliable way to include an article number in BibTeX is to put it in a NOTE field. The proper way is to put it in an ARTICLENUMBER field or some such; this of course requires modifying the style file to make it work (but the MathSciNet output already contains many such nonstandard fields). $\endgroup$ Jun 14, 2022 at 8:03

1 Answer 1


I suggest to format this bibliographic reference as follows:

Renata Kallosh. M-theory, black holes and cosmology. Proceedings of the Royal Society A 20200786 (2021). doi:10.1098/rspa.2020.0786, arXiv:2009.11339.

Rationale: The purpose of a bibliographic reference is to give the reader a very rough idea what the article is about and when it was written and to allow the reader locate the full text of the article as fast as possible.

The former purpose is achieved by including the title, author, and year.

The latter purpose is achieved through a multitude of ways.

If the article is accessed online, through a format like PDF or HTML, then the DOI link at the end of the reference instantly takes the reader to the referenced article.

If the article is printed on paper, and the reader wants to locate the referenced article online, then my impression is that by far the most common method is to supply a search engine with the last name and title of the paper.

This method almost always succeeds, but occasionally it is necessary to include precise and unambiguous information, especially if the journal is obscure and/or the article was published a long time ago. That's why the journal name and article number should be included, just in case.

I would suggest not to include the volume, issue, or page numbers, since these are completely redundant in presence of an article number and including four different numbers makes it confusing which one should be used.

I would also not worry about potentially confusing this number with some other information. Whether we like it or not, this format will become increasingly more common, and soon everybody will be aware that a number with more than 4 digits in a bibliographic reference must be the article number.

I also recommend against abbreviating authors' names or journal names. Not abbreviating authors' names makes it easier to search online for a particular author. This is especially important for some very common last names, e.g., MathSciNet currently lists 20026 authors with the last name Wang.

Not abbreviating journal names makes easier to search for really obscure journals. (The interlibrary loan service at my university, for example, requires unabbreviated journal names.)

Of course, a particular journal can (and often will) change the format of bibliographic reference to suit its style. However, many journals will preserve your DOI and arXiv hyperlinks if you include them. Additionally, an important factor to keep in mind is that if your paper is available on arXiv, more than 90% of downloads of your paper will be from arXiv and not from the publisher's website. (If you know a source with precise data, please indicate it in the comments.) Accordingly, the overwhelming majority of the readers will see the references the way the author formats them, whereas the publisher's version will only be seen by a minority.

  • 2
    $\begingroup$ If the only purpose of a bibliography entry was to allow the reader to find the paper, surely you could also omit the author names. But that is not the only purpose of a bibliography entry. I, for one, frequently like to know roughly how long a paper is, without going to the effort of actually looking it up, which is why I recommend PAGES = {20200786, 15 pp.} above. $\endgroup$ Mar 29, 2021 at 19:38
  • 2
    $\begingroup$ @VinceVatter: You seem to be objecting to a straw man. I indicated two different purposes, only one of which is to locate the paper. The other purpose definitely requires the author names. $\endgroup$ Mar 29, 2021 at 19:40
  • 2
    $\begingroup$ But then you removed the page count. $\endgroup$ Mar 29, 2021 at 19:42
  • 1
    $\begingroup$ @VinceVatter: Do you also include page counts for books that you cite in your papers? Does it really matter (enough to bother including it in the bibliography) that a particular cited result is proved in a 25-page paper as opposed to a 37-page paper? $\endgroup$ Mar 29, 2021 at 19:42
  • 2
    $\begingroup$ I would guess that the reason that people don't include number of pages when they cite arXiv papers is that the number of pages will most likely change. For a published paper, it will not change. I guess there is another principle at work here: your bibliography should not contain information that is likely to become incorrect. $\endgroup$ Mar 29, 2021 at 21:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.