Which syntax are acceptable when citing a sequence of articles?

  • I am teaching my students to use LaTeX2e and BibTeX (albeit I am planning to upgrade to LaTeX3 and BibLaTeX soon!) with the package Natbib, and there one would use a command such as \cite{refBlah,refBloh,refBlih,refBleh} to produce something like “[19,31-33]” in the text and the corresponding entries in the bibliography (if the option [sort] is used, otherwise it could produce the reference in the order they appear, e.g. something like “[33,19,31,32]”).

  • I see articles (some written using Microsoft Word, some with LaTeX) giving the references as “[33], [19], [31], [32]” (generally in disorder), when I was specifically teaching my students to avoid doing that, arguing that the former syntax is more compact and elegant.

Is there any academic community where it is preferred to give the bibliographic references as “[33], [19], [31], [32]” rather than “[19,31-33]”, or is it just a vestige from the time where the generation of bibliographic references was not yet automatized?

  • 12
    If the LaTeX details in the question are distracting from the actual question (i.e. , are long-form ranges ever preferred over condensed ranges), maybe those details can be removed from the question. But the question itself seems completely independent of the tool of choice (LaTeX, Word, etc)
    – penelope
    Commented Jun 18 at 13:10
  • This question does not appear in practice because all major research journals one would submit to come with a bibtex standard file template that will render your *.bib files according to their individual styles. Commented Jun 19 at 10:52
  • 1
    @R.J.Mathar Not quite true: the bibtex standard cannot change a \cite{bleh}, \cite{blah} into \cite{bleh,blah} (but yes, it can represent the latter in the style preferred by the community, which is why I teach my students to use it.)
    – J..y B..y
    Commented Jun 19 at 10:55
  • 3
    In principle I like the dash, but nowadays having all numbers there as clickable links is much more convenient. So just by practical reasons use the version with individual numbers.
    – Dirk
    Commented Jun 19 at 11:41
  • 4
    @AndreiSmolensky that's proper in some fields/journals, definitely not in others. And for my thesis I stuck with the numeric superscript style most of the journals I've published in use. Lengthy lists of names/alphanumeric codes do rather break up the flow of the text, after all.
    – Chris H
    Commented Jun 20 at 9:55

5 Answers 5


Yes. IEEE (Institute for Electrical and Electronics Engineers) is home to a number of major Computer Science journals and conferences.

From their official reference guide:

Reference ranges in text will not include an en dash. All references will be written out. For example, “[1]–[4]” will now be “[1], [2], [3], [4]”

They do not seem to specify anything about the order of references (e.g. whether "[1], [3], [2]" would be more, less or equally as acceptable as "[1], [2], [3]").

I suspect at least one of the reasons is that nowadays, references in PDF documents are clickable links that jump to the appropriate reference in the reference list. If a text cites [1]-[4] (or [1-4]), and I'm interested in [3], I can't click on it to jump to it.

  • 17
    +1 for the clickable links explanation, that makes complete sense
    – Sursula
    Commented Jun 19 at 6:02
  • 4
    (At the same time, how can you know that you're interested in [3] before having seen [1-4] in the reference list?)
    – Karlo
    Commented Jun 19 at 10:47
  • 3
    I had the same concern as @Karlo, which, I believe, could invalidate the last paragraph of the answer. An alternative explanation is the one from Nicola Sap's answer (searchability via CTRF+F). Commented Jun 19 at 12:11
  • 6
    @Karlo Because a long list of example references does not often live in isolation. Usually, the formulation is more like "Approaches for basket weaving include manual [1], semi-automatic [2], automatic [3], and other modes of production [4], [5]. However, none of these approaches or their follow-ups [1], [2], [3], [4], [5] can be successfully applied to weave baskets underwater". And I'm really interested in automatic basked weaving approaches, [3].
    – penelope
    Commented Jun 19 at 13:19
  • 1
    That's interesting. I often see introductions like "Deep learning has revolutionized many fields [1-31]. However, ..." Perhaps they are trying to eliminate such a practice.
    – cag51
    Commented Jun 19 at 21:11

I think independently of the academic community, a non-sorted order only makes sense if the reader is supposed to look at them in a specific order.

I.e. [33,19,31,32] tells me that the key information is probably in [33] and [19,31,32] are just supplemental. But this is assuming a deliberate action by the author and not just the equally possible carelessness. (Though weirdness in the references are usually among the mistakes even underpaid copyeditors watch out for.)

A much clearer way to achieve the same would be something like "(see [33], as well as [19,31,32])", which would be what you should teach your students to use if they think the numerical order [19,31-33] is misrepresenting things.




A practical reason for individually bracketed, explicit references ("[1][2][3]"), is that sometimes readers will search (Ctrl+F) for the places where, in a manuscript, a reference was mentioned.

For instance, when doing a retrospective review of the applications/mentions of a certain study in the literature, one would

  • find (via search engines/crossref) a manuscript that cites the study of interest and download it
  • find what number the study of interest is in this manuscript's bibliography
  • search for where that number appears in the text, to determine in which context the study was cited

From this point of view, different styles have different "searchabilities":

style searchability
❌ [19,31–33] Ref 32 can not be found via Ctrl+F.
❗️ [33,19,31,32] or [19,31,32,33] All references can be found (Ctrl-F, "32") but the search results will contain false hits: page numbers, measurements, years and other values.
✅ [33][19][31][32] or [19][31][32][33] All numbers can be searched and, by including the brackets in the search query (Ctrl-F, "[31]"), the search will mostly contain valid hits.

It is true that journals' websites will often provide an HTML rendition of the manuscript with back-links from the bibliography to the main text. However, I'm interpreting the question as referring to PDFs, and not necessarily published in journals: journals will generally have their style guide that the authors must adhere to, or they will employ an internal re-formatting and proof generation service.


If the document is not meant to be published (e.g. an internal essay or dissertation) or is only meant for print, then the practical considerations above may not be relevant.

In this case, the rendition with an en-dash ([19,31–33]) is, in my opinion, preferable, as it conveys the desired information unambiguously and concisely.


I can't see any reason* why the list would not be sorted: if the author wants to highlight the contributions from one of the references, they will do so by singling them out in the text **:

A promising hypothesis [32], later supported by more experimental evidence [19][31][33], ...

As a reader, I would not expect the order to matter, and – if anything – I would consider it odd if a list was not sorted.

(PS: * comments and other answers make me realize that there can be such reasons: Especially Lime's answer provides one; ** mlk's answer said this before and better than me)

  • Note that pdfs can also have clickable links, not just HTML versions of articles!
    – J..y B..y
    Commented Jun 19 at 10:50
  • 1
    @J..yB..y I am aware of that. But PDFs don't usually have back-links from the references section to the main text (even if they would be capable of supporting those)
    – Nicola Sap
    Commented Jun 19 at 10:53
  • but ref 32 will be found with a ctrl f in the bibliography, wouldn't it?
    – BioBrains
    Commented Jun 19 at 11:37
  • 3
    @BioBrains The use-case is "Given reference 32 from the bibliography: at which places in the text is reference 32 mentioned?" Commented Jun 19 at 12:17
  • 2
    If auto-sorting of references isn't enforced by the style, I actually prefer to sort any reference list by when the papers were published, which does not necessarily coincide with the numbering assigned to them in a manuscript
    – penelope
    Commented Jun 19 at 12:40

Under most circumstances, if I am citing several papers at once then the most appropriate order to cite them in is chronological. This will normally not be the same as the numeric ordering in the bibliography, where in my field at least the primary method of sorting is alphabetical.

I would still prefer [33,19,31,32] over [33], [19], [31], [32]. If numerical order happened to match chronological, I would prefer [19,31,32,33] over [19,31-33]: if a paper is worth citing, it deserves to be cited explicitly.


Regarding the order of citations, it is useful to note that the order may send a message. In a case where group A and group B claim to have discovered a technique independently and have an ongoing priority dispute, writing [B,A] rather than [A,B] could be seen as taking sides.

A diplomatic impasse can be avoided by having a rule such as "always order papers alphabetically in the bibliography, and always order reference numbers increasingly". In this way, the only risk is alphabetical bias.

If you want to take sides, you can always use more words in the text: "we refer the reader to [B], where the technique was introduced, and [A], which expanded on it / rediscovered it independently".

  • I think in cases where there is already a dispute, alphabetical bias becomes a big risk. People could still interpret [A,B] as suggesting priority (unless you explicitly mention this rule when you cite). Commented Jun 19 at 18:01

You must log in to answer this question.

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