The default font of chem.SE web site currently is Georgia. Related CSS rule is:

html, body { font-family: Georgia,"Times New Roman",Times,serif }

The characteristic feature of this font are ‘text figures’, i.e. the digits having non-uniform height and/or baseline.

This leads to confusing typographical appearance of decimal digits. Consider a hypothetical question on H5N1:

chem.SE ...H5N1... text rendering

The digits, which should be normal size, resemble rather subscript digits or something between, in this case; i.e. might be confused with a chemical formula.

(Please note that the actual text appearance can vary depending on the operating system and the browser state and settings.)

This is how this text would look at mother SO site:

SO ...H5N1... text rendering

It does not use this font:

html, body { font-family: Arial,"Helvetica Neue",Helvetica,sans-serif }

Is Georgia font really suitable default font for chemical website like chem.SE, shouldn't it be changed?

  • 1
    $\begingroup$ "Is Georgia font really suitable default font for chemical website like chem.SE?" No, it is not a good choice. However, using a sans-serif typeface like Arial would be even worse, even if it could solve the problem with the old-style text figures. $\endgroup$ – Loong Sep 15 '18 at 15:22
  • 1
    $\begingroup$ Times New Roman, maybe? $\endgroup$ – user10259 Sep 15 '18 at 15:27
  • $\begingroup$ @Loong I wonder why (some) sans-serif would be a bad choice.. did e.g. Angew. Chem., Int. Ed. make a bad choice? $\endgroup$ – mykhal Sep 15 '18 at 15:29
  • 1
    $\begingroup$ Related: chemistry.meta.stackexchange.com/q/245/23561 $\endgroup$ – A.K. Sep 17 '18 at 19:11
  • $\begingroup$ i.stack.imgur.com/zWQVX.png $\endgroup$ – Avnish Kabaj Sep 25 '18 at 5:04
  • $\begingroup$ Assuming we come up with an consensus is there a chance that the website font would change? and if so what is that process? $\endgroup$ – A.K. Oct 3 '18 at 20:23
  • $\begingroup$ I don’t understand your comment about the Angewandte: I just checked the first random ASAP article and was confirmed in my suspicion that they use a serif font. $\endgroup$ – Jan Dec 5 '18 at 17:12
  • $\begingroup$ I was just making sure that they hadn’t changed something that went part unnoticed. Let me go through an article I have open rn. The keyword is in sans-serif (good). The title is serif (meh, but okay). Authors, abstract and full text with the exception of the article’s first letter are serif (good). The DOIs are sans-serif (good). The author addresses and supporting information notice are sans-serif (meh but acceptable). Image and table captions are in sans-serif (bad, but not terrible). Labelling in tables are sans-serif (acceptable). (t.b.c.) $\endgroup$ – Jan Dec 5 '18 at 17:32
  • $\begingroup$ Acknowledgements and Conflict of interest (section titles) are sans-serif (good). The keywords at the bottom are in sans-serif (hmmm). How to cite is in serif (good), the references are in serif (good). The footer and header are in sans-serif (good). Where did the Angewandte change to sans-serif, or am I misinterpreting your comment? $\endgroup$ – Jan Dec 5 '18 at 17:34
  • $\begingroup$ @Jan Let's assume I was talking about web. See the abstracts, or one of articles in a free sample issue, it uses sans-serif font, at least as I see it. They don't seem to care much, as the font-family for the text can be one of: Open Sans, sans serif, … $\endgroup$ – mykhal Dec 5 '18 at 17:42
  • $\begingroup$ Ah, I never actually use the web versions of articles because I need to save it if it’s interesting anyway. But yes, the web version is in sans-serif and harder to read. $\endgroup$ – Jan Dec 6 '18 at 15:21
  • $\begingroup$ Wiley in general has the worst User Interface in publishing, IMO. Angewandte (or W in general) seems to not care a tiny bit about how text is set, or style in general. And that's not even taking into account that most of the images they use are somewhat blurry. (cc @Jan) $\endgroup$ – Martin - マーチン Dec 12 '18 at 17:40
  • $\begingroup$ I am not sure what the discussion here actually achieved, except for the sake of the discussion. If it was aimed at changing specific things, then it should be followed up with a feature-request. I think it is somewhat clear that there is generally a preference for a serif font, but at the same time a general dislike if the Georgia font. I am not sure, if such a feature request actually has a chance with the developers, and if it does, how high on the to-do list it would place, but if we don't try, we won't know. $\endgroup$ – Martin - マーチン Apr 5 at 12:02
  • $\begingroup$ I am most decidedly pro-sans-serif. Hopefully we get to vote on such a decision. I agree that the current dislocations between text/math/chem formatting can annoy. $\endgroup$ – Buck Thorn Apr 8 at 10:09

I concur that georgia is suboptimal for our number heavy site; not to mention that zero and lowercase o look the very similar [0o0o]. I would prefer not to use \$ \$ to make numbers presentable.

Personally, I think Century would be a good choice but I've listed some fonts with nearly equivalent sizes for comparison (please no one pick calibri).

Per Ben Norris's input I have evaluated some of the fonts against the 0oO and |Il1 tests. As this is subjective feel free to tell me what I may have evaluated wrong. It seems Ludica and Perpetua pass the best, though I am still biased towards Century.

enter image description here

  • 1
    $\begingroup$ I vote for calibri :P $\endgroup$ – Avnish Kabaj Sep 17 '18 at 19:47
  • 7
    $\begingroup$ no comic sans, am disappointed $\endgroup$ – orthocresol Sep 17 '18 at 21:39
  • 4
    $\begingroup$ Comic sans... now my eyes are bleeding. $\endgroup$ – LordStryker Sep 18 '18 at 17:02
  • 1
    $\begingroup$ @LordStryker in some ways still better than Georgia. $\endgroup$ – A.K. Sep 18 '18 at 18:27
  • 2
    $\begingroup$ Any proposed font needs to pass the oO0 (lowercase o, uppercase O, zero) test and the |1Il (vertical line, one, uppercase I "eye", lowercase l "ell") test. $\endgroup$ – Ben Norris Sep 20 '18 at 22:46
  • $\begingroup$ @BenNorris -see edit- thoughts? $\endgroup$ – A.K. Sep 24 '18 at 19:31
  • $\begingroup$ Under what circumstances would one ever need the vertical line in normal text-flow? $\endgroup$ – Martin - マーチン Sep 25 '18 at 15:17

I'm not a designer nor I have a major in typography, so the following is a pure subjectivism.

I think that half-measures are evil, and attempting to fit subscripts in chemical formulas with the oldstyle/hanging numbers is a half-measure. The web is still crippled in typographic features compared to the desktop publishing systems, but it's not an excuse for a poor font choice. I call it poor because every time I see numbers in Georgia not lining up, "The Adventure of the Dancing Men" comes to my mind.

Some time ago I did comparison tables for the cherry-picked free and open source fonts alongside with the default fonts installed in Windows that don't look particularly ugly:

Serif fonts

Sans-serif fonts


My main concern at the time I compiled these tables with XeLaTeX was to find the font that has better coverage for the symbols used by chemists and to find a font family of the sans-serif fonts where one can actually distinguish between I and l in chemical formulas (on the side note, I still think that choosing Arial as the default font by ACS for representing drawn formulas was a big mistake resulting in the urge to use lowercase letter "ell" instead of semantically correct lowercase "l": $\ce{HC\ell}$ vs $\ce{HCl}$).

Back to the main issue, as long as there is MathJax used, it feels like the main font should be a serif font to comply with Computer Modern (or its derivative) used by it. Among the serif fonts I personally find these three open fonts sticking out:

  • STIX Two/XITS. Made specifically for scientists, based on Times ligatures. One of the few fonts offering the proper symbol for the standard state. With a recent 2.0 release there were major improvements.
  • Linux Libertine. Mature, looks fairly good, great coverage. Used by Wikipedia, nuff said.
  • Noto Serif. Covers lots of symbols, installed by default in all modern systems. However, not looking scientifically enough to my eyes.

Also, Georgia font has very rudimentary ligatures support, and some titles are just painful to look at. The majority of other fonts do it better, and of course, the above mentioned fonts handle ligatures much better:

enter image description here

To sum it up: I think the numbers should be lined, serif font is better, and among serif fonts either STIX or Linux Libertine (or their derivatives, e.g. XITS or Linux Biolinum) would be arguably the best option. And no, I don't like Georgia [font].


I think the choice of oldstyle numbers was at least somewhat influenced precisely by the large number of numerical subscripts in chemistry meaning that plain-text typed H2O would at least somewhat resemble $\ce{H2O}$ even though it is $\text{H2O}$. However, I also think that after a significant time on this site the drawbacks become more obvious than the benefits.

Thus, I agree that ideally any scientific site should have newstyle number selection. (I forgot the technical terms for these numbers, feel free to point them out.) It would make many things more readable. As a (bad) workaround, I have often used simple dollar signs to turn oldstyle numbers into full-sized ones—a bad and unpractical solution.

When it comes to selecting a new font I want to propose: Select the font MathJax uses or one that is very, very similar to it! One of my pet peeves that I cannot do anything about is how different MathJax looks from standard text. For optic consistency, I prefer to enclose all chemical entities in $\ce{...}$ even when a simple inline text would be enough, e.g. when I’m talking about NaCl. If the font closely matched the one MathJax uses, that problem would be gone.


It's not the best choice, but I think it is preferable to what was before. You might want to check out the discussions when we got the graduate design, it's one of the points I mentioned: https://chemistry.meta.stackexchange.com/a/1673/4945

You can simulate how a sans serif font would look like still on a mobile browser. It simply doesn't match and creates weird breaking points in text flow.

screen shot mobile

I personally prefer serif fonts, I find them better readable, but there are certainly points in favour of sans serif fonts, too. The important point is that text and chemistry (maybe even maths) should match and not create an imbalance. With all the awfulness of MathJax, there are drawbacks of what a site can look like, right now I think it looks okay-ish for most purposes. The numbers are something I also really dislike, but I can live with this; I wouldn't want to do back to that really awful mix we had before.


One answer that would be more broadly for Stack Exchange in general is to allow users to select the font they desire from a small list in the site settings. Of course there would be a default/recommended font but then that many more people could be accommodated.


You must log in to answer this question.

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