Whenever I am creating figures for publication, I often wonder if I should be using a serif or sans-serif font. I browse the journals in my field and notice that there is no standard, just chaos.

I have typically chosen serif font to match the typography of the body text; however, I have read that sans-serif stands out among the serif body text. My rationale for choosing serif text is because it allows me to reproduce the symbols from the text exactly in my legends (and/or annotations). To me, this seems clearer.

Does anyone have a source or standard that recommends one or the other?

  • 1
    Some journals provide explicit recommendations. For example, Plos One asks for Arial.
    – Alexlok
    Feb 28 '14 at 20:56
  • Actually, I have seen PLOS, but they are one of the few that actually specifies font. Most just specify size.
    – Blink
    Feb 28 '14 at 20:58
  • 1
    Science does that as well. They opt for Helvetica (sans-serif). Feb 28 '14 at 21:05
  • Maybe I am losing something here, but are you assuming that journals will typeset your material the same way as you sent them your manuscript? Unless in some specific cases (Latex), they will typeset everything you send them.
    – user574859
    Feb 28 '14 at 22:18
  • Journals do not typically maintain your manuscript format, but they usually can not (do not) edit your figures (e.g., axis labels, tick marks, legend entries).
    – Blink
    Mar 1 '14 at 16:08

When it comes to typographic design, it can be dangerous to adhere to rules of thumb. Sometimes (actually most of the time) sans-serif fonts work, sometimes they don't. It would depend on tradition, trend, and overall feeling that the fonts project.

If the journal does not specify. I would usually favor sans-serif. The reason is that unlike my texts, I am not always sure how much the editorial team may size down my illustration. Sans-serif has a pretty good property that they are quite resistant to shrinking, and can still be legible at relatively small size.

In the mean time, if the publisher uses any software to smooth out the edge of the fonts after resizing (e.g. through aliasing,) serif fonts can sometimes appear broken at their thinner strokes.

There are, however, some illustrations that just don't look right with sans-serif. For instance, line labels and angle labels of trigonometry problem sets and formula like this one are much nicer with bold and/or italicized serif fonts, monotone ink-drawn anatomical charts (like this one) will just look very odd if we put on sans-serif labels. This timeline describing Shakespeare's Life may look ridiculous if sans-serif fonts are used.

In those difficult situations, look for serif fonts that are beefier or with more uniform stroke width. As they can likely withstand shrinking and aliasing. In addition, look for fonts that are slightly wider, and have a good "x-height" (literally height of the font "x"). Some possible candidates are Caslon, Baskerville, Garamond, and Palatino. Avoid cursive fonts, or fonts with some very thin lines like Times New Roman. A more in-depth discussion on squint-free fonts can be found in this blog page and this thread on SE UX.

  • "Avoid cursive fonts, or fonts with some very thin lines like..." I assume CMR (LaTeX) fall within that classification? I used Bitstream Charter for my last paper (to match the journal as close as possible), and that seems like a slightly "thicker" serif font.
    – Blink
    Feb 28 '14 at 22:03
  • Yes, I would probably take CMR out unless I know the exact dimension of the published illustration. And yes I agree Bitstream Charter looks very similar to the preferred ones I listed above. Feb 28 '14 at 22:07
  • 1
    +1 - great advice. IMO most of the "font" problems in charts stem from not having labels large enough to begin with (really a bigger problem for presentations than for charts in journal articles) and saving charts in inappropriate raster formats. Make sure to save your charts in either the vector format the journal wants or save them as high resolution raster formats without loss (PNG is the best - although many journals still want TIFF). So many charts in articles you can tell went through some stage as JPEG and everything, including the type, suffers.
    – Andy W
    Feb 28 '14 at 23:44
  • @AndyW: Agreed! The last paper I submitted included the images as PDFs (vectors), but for some reason they rasterized them and resized them... Fortunately, Bitstream Charter seemed to scale sufficiently.
    – Blink
    Mar 1 '14 at 16:14

I use Helvetica/Arial on all my figures, as it is a neutral font that doesn't detract from the point of the figure - to present data. It lacks the flourishes of most serif fonts, or stylistic features of othe sans serif fonts. As others have commented sans-serif fonts are more readable at small sizes, hence their overwhelming use in road signage. Since most figures are small when reproduced, readability is paramount.

  • 2
    My main issue with san-serif fonts are their failure to clearly distinguish 'l', 'I', and '1'.
    – Memming
    Aug 30 '17 at 18:45
  • @Memming If distinguishing between 'l', 'I', and '1' letters is important, then sans-serif fonts like Cabin, Fira Sans, Go, IBM Plex Sans and PT Sans might be helpful: they all present number 1 with a distinct "hook" on top and small letter "l" with a "tail" on the bottom. As a bonus, those fonts are free and subjectively very nice looking; just make sure the special characters you need are included with either of the font.
    – andselisk
    Oct 6 '19 at 9:49

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.