Though I have solved this problem, I still want to hear the LaTeX community on this point. I also need to know if I have misunderstood anything, or if my fix is wrong.

I have the following mathematical expression:

S_p = \bigcup_{c \in C_p}^{\abs{C_p}} L_c \enspace \text{,}

which results in this:

enter image description here

Not the annoying space between capital S and its subscript.

It figured out a way to avoid that by using a negative thin space \!:

{S\!}_p = \bigcup_{c \in C_p}^{\abs{C_p}} L_c \enspace \text{,}

which gives me:

enter image description here

And the problem seems solved.

Both graphics are screenshots from a zoomed PDF document in Acrobat Reader X. I am using the elsarticle class (Elsevier's LaTeX class). The blue text was marked with the mouse.


  • Is my fix using the negative thin space correct?
  • Or is there a better way?
  • Is this a known problem with S and its subscript in general, or is is dependent on class used? I have not tried using the standard article class yet.

I have studied What is the right way to use the `\!` spacing command? and it seems to be an acceptable way to fine tune mathematical expression.

  • 2
    It would appear that you're using the mathptmx font package, which gives you text and math fonts based on Times Roman. Be advised that most Elsevier journals use a proprietary font called Gulliver rather than Times Roman. (See tex.stackexchange.com/a/45917/5001 for more information.) As such, it may not be worth your time and effort to optimize the look of some math expressions (by inserting negative thin spaces) in your paper, because these adjustments will have to be undone in all likelihood when the paper is typeset in Gulliver. – Mico Jul 16 '12 at 16:19
  • Thanks. That was basically why I asked this question. Perhaps I should have added that. I am ofcourse writing a paper for an Elsevier journal. And I know that the font I am seeing in my draft, will not be the font I see in print. The Gulliver font is ... not suitable for math though. Anyway that's both off-topic and subjective :) So I should probably do no fine-tuning in my draft? – Ole Thomsen Buus Jul 16 '12 at 16:49
  • The problem with fine-tuning the look of the math (and plain text, possibly) for a specific font family is that what looks good in that font family (say, Times Roman) may be unsuitable for publication using a different font family (say, Gulliver). Hence, all your fine-tuning have to be undone (or done entirely differently) by the journal. My suggestion to you would be, indeed, to stick with Computer Modern (or Latin Modern) for the preprints. – Mico Jul 16 '12 at 17:17
  • I will probably need an answer below that illuminates the points you make Mico. On top of my specific questions above, you can add the fact that I am using elsarticle with the intention of publishing in an Elsevier journal. The issue is on fixing the observed space above, which mean font changes, fine-tuning, both, or none. – Ole Thomsen Buus Jul 16 '12 at 19:05

You get better results with




S_p = \bigcup_{c \in C_p}^{\abs{C_p}} L_c ,

This requires an up-to-date TeX Live 2012 (or MiKTeX).

enter image description here

  • 1
    I guess you get a better-looking S_p, but now the gap between the "C" and "p" in "C_p" looks decidedly too wide. I suppose Michael Sharpe isn't fully done yet optimizing all the settings of his newtxmath package... – Mico Jul 16 '12 at 16:22
  • If you draw an imaginary vertical line to imagine the bounding box of those 2 letters, you'll note that the space looks like right. – Sigur Jul 16 '12 at 16:42
  • The C_p is fine. There also seems to be a space in my examples above. As per Mico comment above, I might be making problems for myself later, since I am writing a paper and it will be typeset in a different font in the end? And the problem seems to be font-specific. – Ole Thomsen Buus Jul 16 '12 at 16:53
  • @Thiago -- The problem is not with the shape of the bounding boxes per se. The problem, rather, is that the bounding box that contains the italic-C box has little material in the SE quadrant, whereas the italic-p box has little material in the NW quadrant, opening up a visual gap in the process. Keeping such visual gaps to a minimum is what the construction of kerning tables is all about, isn't it? – Mico Jul 16 '12 at 17:22
  • @Mico: The problem should only be with the p. The italic C has little material in the SE quadrant, yes, but the italic correction of the C should take care of that. (For horizontal sub- and superscript placement, TeX only cares about the italic correction of the nucleus; kerning tables don't play any role.) – Hendrik Vogt Aug 15 '12 at 15:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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