13

I just tried submitting my first paper to arXiv. To be honest, this hasn't been an unqualifiedly fun experience.

One reason for rejection was that my paper contained line numbers. I am a bit confused as to why arXiv would reject papers for containing line numbers. One reason for me to submit to the arXiv was to potentially get comments on the paper - and it's a lot easier to comment on a paper that has line numbers. It's easier to discuss "lines 95-105" than "the third paragraph on page 5, not counting the table and its caption". I really like line numberings in papers I'm supposed to review.

Can anyone tell me why arXiv does not allow numbered lines in submissions?

(A quick internet search didn't turn up anything useful. )

Stephan Kolassa
  • 34,479
  • 11
  • 114
  • 185
  • 11
    Perhaps, the editorial team at arXiv values readability more than maintainability. – Aleksandr Blekh Jul 25 '15 at 19:17
  • 3
    I've been an ArXiv moderator (in CS) for several years, and I've never heard of such a policy. – JeffE Jul 25 '15 at 23:00
  • 12
    I would guess it is related to the principle that arXiv preprints should generally meet the content and formatting standards for journal papers. You wouldn't see a paper in a journal with line numbers. – Nate Eldredge Jul 26 '15 at 00:45
  • 3
    @JeffE: I can confirm that this exists as I have accidentally left on the line numbers myself once and had to resubmit the paper. However, I cannot cite a policy for this. – Wrzlprmft Jul 26 '15 at 10:14
  • @JeffE: the line numbers were detected automatically (not too hard, since I used the lineno package), and the rejection came just as automatically. I submitted to stats.ME and have no idea about other fields, but I would be (even more) surprised if this policy was restricted to just one field. – Stephan Kolassa Jul 26 '15 at 11:49
  • 1
    To second @Nate , the arXiv is a preprint server, and even though people do sometimes put articles on there for comment, it isn't really designed for that. (On the other hand I'm surprised they don't allow line numbers) – P.Windridge Jul 26 '15 at 12:32
  • 1
    If people are interested in discussing your paper, I'm sure they will go to the effort of discussing the actual material, not just "I like the material in line 41 on page 7, especially when combined with the idea on page 2, line 17". May I also suggest formatting your paper as you would like it to appear in a journal, and not looking like a draft and/or review manuscript. There's a reason journals don't include line numbers in their papers and keep them at 1.5 or double-spaced: readability. – David Roberts Jul 27 '15 at 07:34
  • 6
    I respectfully disagree with the statement "line numbers decrease readability". – Federico Poloni Oct 06 '15 at 02:39
  • 1
    @NateEldredge: While that may well be the reason, it sounds like a quite closed-minded reason to me, for various reasons. First, for referencing purposes, I'd actually find it a good idea to see journal papers print line numbers. Then, line numbers do exist occasionally in papers in their final layout, if only for limited sections of text, namely listings. Lastly, I'm not sure why several comments here suppose formatting standards of a journal are not adhered to: Any journal/proceedings format I know has a margin and hence more than enough space for line numbers without changing the layout. – O. R. Mapper Oct 11 '15 at 19:11
  • @DavidRoberts: Nothing prevents you from referring to concrete phrases from the text in your final comments. Still, don't you find the additional option helpful to scribble down notes while reading the paper simply by writing down the line numbers rather than copying a portion of the text? In particular as the line number inherently tells you where the line is approximately located, which is not the case when you only have some words from the text? Granted, the latter advantage is primarily important if you prefer reviewing papers printed out on paper, or if full text search is broken. – O. R. Mapper Oct 11 '15 at 19:26
  • I've never seen a paper with line numbers. Could you share a link to such a paper? – Earthliŋ Oct 11 '15 at 19:50
  • @O.R.Mapper If I have a printed copy, I write notes in the margins :-) and I like wide margins for that purpose. Standard LaTeX maths-intended packages tend to leave wide margins, too. If I'm refereeing, fine. But I don't need it. – David Roberts Oct 11 '15 at 21:14
  • 1
    @DavidRoberts: Mhm. I often find myself writing notes referring to several lines at a time wherever there is space. e.g. "line 2215: contradiction with line 978". It seems, optional line numbers that are shown or hidden based upon reader preferences are the way to go :) – O. R. Mapper Oct 11 '15 at 21:42
  • @Earthliŋ: when you write that you have never seen a line-numbered paper, do you mean that you have never seen line numbers in a final published paper (cf. O. R. Mapper's comment: lines are commonly numbered in computer code listings), or in manuscripts? In the latter case, the documentation for the LaTeX lineno package gives an example and also illustrates how you can modify the numbering. – Stephan Kolassa Oct 12 '15 at 07:15
  • @StephanKolassa I have never seen a published paper nor a manuscript with line numbers (except computer code). Personally I find the lineno documentation cluttered and hard to read. I'm guessing the reason is akin to Stack Exchange not allowing underline. Somebody didn't like it, disallowed it and it stayed that way. – Earthliŋ Oct 12 '15 at 13:55
  • @Earthliŋ: I agree that the doc is cluttered. That is because the author showcases all the different ways to customize the numbering. In actual use, you would of course not use bold numbers, roman numerals, flipping the numbers between the left and right margin etc., but, e.g., only number every fifth line. Which I personally find very helpful and not at all distracting. YMMV. – Stephan Kolassa Oct 12 '15 at 14:13

2 Answers2

8

In my experience, it is much more useful to give some context (a few words, name of the theorem, ...) to any editorial suggestions than line numbers. My documents exist as LaTeX source, so line numbers in some printed version don't correlate at all with line numbers in the text I currently work on. A few words of context are easy to search for in a text editor, line numbers would require referring to the printout to find same first. Besides, text changes (paragraphs get rewritten, shuffled around, new chapters are added, a new example appears, ...).

vonbrand
  • 9,982
  • 1
  • 25
  • 46
  • 4
    This. Specially if you get several comments, the moment you address the first one, the rest of the line numbers go down the toilet. – Davidmh Oct 12 '15 at 09:30
  • Hmm, maybe the default line numbering should be from the bottom. If one does changes in the beginning, the remaining numbered issues later in the document should stay numbered the same... – Per Alexandersson Mar 05 '24 at 18:14
3

It may be a way of discouraging authors from submitting galley proofs, which are typeset in the final format of the journal but generally have line numbers.

Doru Constantin
  • 505
  • 3
  • 8
  • 4
    Maybe, but why? I have recently reviewed a manuscript that was in this format, and I found the line numbers quite helpful for reading and taking notes. So, why would they want to discourage authors from submitting these so-called "galley proofs"? – O. R. Mapper Oct 11 '15 at 19:15
  • 1
    For copyright reasons. Posting the content of the paper as a preprint is (usually) OK, but not using the graphical identity of the journal. – Doru Constantin Oct 11 '15 at 19:43
  • But such a restriction would probably not apply to fields where it is customary to use the final layout already for the initial manuscript. – O. R. Mapper Oct 11 '15 at 19:51
  • 1
    I would also find that logical. However, the moderation page http://arxiv.org/help/moderation of arXiv states that: "arXiv cannot accept PDFs that have been downloaded from a publisher's website and contain a copyright statement". – Doru Constantin Oct 11 '15 at 20:12