I just came across an answer by a new user which looks very unusual to me: https://mathoverflow.net/a/139275/21095

Specifically, there is very little mathjax in this answer. Instead the vast majority of the math is pre-formatted text, including hard-coded bold and italic symbols, hard-coded subscripts and superscripts, etc. Please take a look at the code for the answer yourself, as I find it hard to describe.

Edit: According to the author of the post, it was composed using XCompose.

One problem is that the text does not even render in an alternative web browser I use called midori (not mainstream, but highly developed). I left a comment suggesting that the answer be rewritten using latex/mathjax code.

My questions are:

  1. How and why does the software accept and recognize these symbols? Are all the symbols unicode?

  2. Is this kind of code used elsewhere? Or is this a completely isolated case?

  3. Should such code be replaced with mathjax?

Update: It would appear that all the symbols are indeed unicode. I am unable of checking this statement thoroughly, but it seems plausible from the answer by Rohan Lean below, and from a cursory look at a list of unicode characters, e.g. here.

  • 2
    $\begingroup$ It also does not render with my browser; Safari 5.1.9 on a Mac. $\endgroup$ Aug 13, 2013 at 0:28
  • 1
    $\begingroup$ I removerd the tag 'bug' as I really do not see any bug here, now that it seems a given everything is unicode. (Obviously feel free to revert; it is mainly as things tagged 'bug' are looked at quickly by devellopers and this now seems mainly a community issue how to see this type of input or also a support question perhaps). $\endgroup$
    – user9072
    Aug 13, 2013 at 1:08
  • $\begingroup$ @quid: Thanks for removing the tag. The question is probably better without it, given the current information. Thanks also for the information on the function of the tag 'bug' concerning the developers. $\endgroup$ Aug 13, 2013 at 1:12
  • 1
    $\begingroup$ I have similar problems using Firefox 20.0 on a Mac. $\endgroup$ Aug 13, 2013 at 2:34
  • $\begingroup$ @Bill: Thanks for the information. I appear to have no problems viewing the answer with Firefox 22 on Linux. Can you please briefly describe the issues you are seeing? $\endgroup$ Aug 13, 2013 at 2:59
  • $\begingroup$ Hmm. On my office Mac, the rendering is correct on Safari. I'll check on my laptop again when I am at home. It looks like a font issue unless the editing last night fixed the problem. $\endgroup$ Aug 13, 2013 at 17:46

1 Answer 1


I am the author of that post.

All the characters are in Unicode, but some are encoded in more than two bytes in UTF8. Perhaps that trips up Midori?

I enter the characters using XCompose, in case anyone is wondering.

Edit: (As I unfortunately cannot post comments yet.)

Both Midori and Safari use WebKit to render web pages. I have just confirmed that even modern versions of WebKit have trouble displaying my post correctly. I am curious whether this is a bug in WebKit or whether I unknowingly violated some web standard with the characters I used. In any case, it seems that the problematic characters are some of the subscripts only. Can someone confirm that?

  • 1
    $\begingroup$ Dear @Rohan: Thank you very much for your answer, as it greatly clarifies the situation. I apologize regarding the unicode text. I was completely unaware that unicode had italic and bold text symbols. $\endgroup$ Aug 13, 2013 at 0:26
  • 1
    $\begingroup$ Dear @Rohan: Regarding your question, Webkit 1.10.2 appears to not display some italic symbols and some bold symbols. It displays at least some subscripts. $\endgroup$ Aug 13, 2013 at 1:20
  • $\begingroup$ On an up to date OSX system, the only characters that seem to cause trouble are the subscript ‘t’ and ‘p’, and I think those might be a font issue. Are the symbols that are not displayed correctly shown as a box? (I think that a screenshot might be helpful.) $\endgroup$
    – Rohan Lean
    Aug 13, 2013 at 1:46
  • $\begingroup$ Unfortunately, I'm also seeing many symbols as a box (not just subscripts and superscripts), both with Chrome and with IE10 on Windows 7, but different symbols are problematic in each case. $\endgroup$ Aug 13, 2013 at 2:08
  • $\begingroup$ In my case, the symbols simply do not appear. There are no boxes. I am using two different webkit browsers on linux. $\endgroup$ Aug 13, 2013 at 2:13
  • $\begingroup$ I suspect that this is a font issue, then. I am willing to go through my answer and replace the troublesome expressions with mathjax markup. Can anyone think of a better alternative? $\endgroup$
    – Rohan Lean
    Aug 13, 2013 at 2:22
  • $\begingroup$ (I am using Chromium on Arch Linux, by the way. It has no problems displaying my post.) $\endgroup$
    – Rohan Lean
    Aug 13, 2013 at 2:24
  • $\begingroup$ (Chrome dev on OS X mostly gets it right, but with a number of missing characters, rendered as boxes.) $\endgroup$ Aug 13, 2013 at 3:37
  • $\begingroup$ The post now uses mathjax. Does it work for everyone now? $\endgroup$
    – Rohan Lean
    Aug 13, 2013 at 3:58
  • $\begingroup$ @Rohan: The post appears to render appropriately when mathjax uses html-css or svg. However, there are still missing symbols when rendering with mathml. $\endgroup$ Aug 13, 2013 at 5:10
  • $\begingroup$ I still use Unicode characters in math-mode, but I would have thought that they are handled in exactly the same way as any other valid description of the symbol by mathjax. Does mathjax use mathml by default, so that my edit is not an improvement, or can most users now read the answer without problems? $\endgroup$
    – Rohan Lean
    Aug 13, 2013 at 5:49
  • $\begingroup$ I think the default renderer for mathjax is html-css, so that anyone who uses the default should be able to read your answer now. However, some people use mathml to render mathjax. Further, some other people either disable mathjax (e.g. by disabling javascript) or use a browser incapable of rendering mathjax. $\endgroup$ Aug 13, 2013 at 6:23

You must log in to answer this question.

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