I'm writing a scientific text, which includes a labeled diagram. Are these traditionally labeled in the same font and font size as the body text, or in a smaller font size, or in a different font?

  • Hi. Welcome to GDSE. There's no such rule AFAIK, and font choice is subjective.
    – Billy Kerr
    Commented Jan 5 at 11:27
  • @BillyKerr I understand that everything is allowed, but I hope that there is knowledge and intuition that I lack that could guide this choice.
    – Anna
    Commented Jan 5 at 11:32
  • Look at some science papers. See what others have done, and use similar fonts if you like them.
    – Billy Kerr
    Commented Jan 5 at 11:48
  • Doing what always has been done traditionally is seldom a good idea; especially in a field as conservative as design / typography in science. So I'd actually recommend against looking what others have done as @BillyKerr suggests. Working on a more thorough and more constructive answer.
    – Vincent
    Commented Jan 5 at 13:49
  • @Vincent - but if someone isn't a designer they may need help, and the work of others is as good a place as any to get ideas from, as a starting point. I also have the feeling from what I've observed of scientific papers, is that the designs are generally more conservative. This is not necessarily a bad thing.
    – Billy Kerr
    Commented Jan 5 at 17:11

4 Answers 4


As you can notice, there are different answers. This means that the word "should" does not apply.

Let me separate your question into several parts. Font-face and size. Although they are closely related.


It is pretty obvious that size denotes hierarchy. There are some already defined. Chapter titles, section subtitles, and the body text.

The body text size is what defines the minimum size the superior ones can be.

Although some other sections have less hierarchy than the body text, like footnotes, there is a limit on what they should be, because of readability.

But also they need to be different so the normal flow of reading the main body is clear. A keyword is a flow.

Footnotes can be differentiated by size, as mentioned in other posts can be in order of 2pts. 11-9pts 10-8pts.


In diagrams, the theoretical size is delimited is also by hierarchy and readability.

But in reality, it is limited by space and tools.

For a diagram, it would be more important to have a consistent size to show the internal hierarchy inside the diagram itself, than comparing it to the main text, but maintaining a good redability.

Arguably the information provided by a diagram is really important, so try to keep it at least the same size as the body or at least the same size as the footnotes.


Normally this changes in this is to provide differentiation, not over hierarchy but about context.

The basic change in font-face is using bold or italics, but normally these are reserved to be used inside the flow of the text.

A more "advanced" change is using the same font in different weights, extra bold, light, or different widths like extended or condensed.

And a change from sans-serif to serif is probably a more stylistic approach than hierarchy alone.


You might need to use a different font in case you want a clear distinction between the diagram and the flow of text, but in reality, this distinction and isolation should be given by well-defined spacing.

So: "should" the font be different? No. It might.

In practice, I suppose scientific papers traditionally depended on what Microsoft Word has defined by default. For decades it had Times Roman as the body default. And Powerpoint could have Arial as default...

LaTex also uses a serif font as default.

Another reason, I suppose, scientific papers keep those "archaic" design choices might be because 1. They want to look "timeless" or 2. They do not want to look very distracting, disruptive, or less serious.

As a recommendation, a modern approach would use a sans-serif font for both, text and diagrams, and choose one font that has enough internal styles and weights to make the differentiation you need.

I agree that Lato and Open-Sans are good choices, but ONLY if your paper is going to be distributed in PDF format with fonts embedded. If it is going to be distributed in Word format you need to use standard fonts, like Calibri, Arial, or Times... Times and Arial look old.

Unfortunately, Calibri font does not have a wide selection of weights.

You can use Verdana for diagrams, that have a good readability in small sizes if you have that constraint.

P.S. If by "label" you mean "Figure 1" and "Figure 2" text use the same size using different weights or italics.

  • Thank you very much! By "label", I meant single words in diagrams, not the figure caption.
    – Anna
    Commented Jan 6 at 11:12

It... Depends. There are two different questions to answer here, and they are pretty distinct:

  1. What is traditionally done for the typography of labels in a scientific diagram?
  2. What is a good, legible, and reader-friendly rule of thumb for typography for the labels in a scientific diagram?

The answers to these two questions are as different as the answers themselves.

Question #1 is hard to answer without becoming rude, but I'll try: Typography, and design in general, is traditionally not a concern in science. There are few traditional rules, and those that exist are based more on being grandfathered in rather than yielding good and legible results. I'll just be bold (pun intended) and say that following traditional typography and design rules as they exist in science is a bad idea. Your message will not stand out, will be hard to parse or even decipher if you do.

Disclaimer: this answer is a bit jaded and exaggerated, but it's a rather accurate reflection of my experiences as a designer in science.

Question #2's answer consists of a few points.

  • Choose a readable sans serif typeface (as per Julian's answer). Good candidates are Roboto, Lato, Open Sans or Montserrat (all links to Google Fonts). If the body text of your article is set in a sans serif, choose the same typeface.
  • Ideally, you'll want to make your labels of roughly the same font size as your body text, or maybe a bit smaller. If you go for smaller, make sure the point size differs by at least ~20%, so eg. 8pt labels next to a 10pt body text. Normally, you'd want different kinds of text to have sufficient contrast to help differentiate between their functions. But a reader confusing labels with body text is highly unlikely, so those being of the same font and size is possible, if not ideal. You may want to give your labels a colour or make them bold or italic if you end up choosing the same size as your body text.
  • Your labels should never be smaller than ~8pt (slightly depending on the typeface and font variant). Smaller font is hard to read, especially for supervisors and peer reviewers approaching retirement age.
  • More important than the exact size, is being consistent with that size. All labels should have the exact same size (and font, and colour), unless there is a hierarchy among them. In that case, use (distinctively) different typography to reinforce that hierarchy.
  • Do realise that you will not always have control over the relative size of your diagram in relation to your text, especially in traditional scientific publications and magazines. Therefore, it's safer to err on the side of large and legible for any text you inlcude in your diagram.
  • Leave enough white space around your labels to prevent overcrowding the diagram. If necessary, use pointer lines.

tl;dr (too long, didn't read)

Traditional typography choices in science kind of suck, so you may want to do your own thing rather than follow convention. Choose a sans serif typeface, and match your body text typeface if it's in sans serif. Exact size is not hugely important, as long as the difference between body text and label is obvious. Don't go below ~8pt, even after the diagram is shrunk, eg. to fit in a magazine.

  • Thank you, very helpful list!
    – Anna
    Commented Jan 7 at 9:43

For diagrams it is in general best to go for a well-readable sans-serif font. If you have a serif or more stylistic font it makes sense to use a different font for diagrams, otherwise use the same.

The amount of information on diagrams often forces you to have small type.

  • 2
    If the amount of information in a diagram forces the labels to be small, it's a bad diagram. /unpopularhottake
    – Vincent
    Commented Jan 5 at 15:00
  • Not nessecarly. Depends a lot on the content of the diagram. When you are talking simple flowchart – yes agree. If you have more complex time/money chart for instance you don't want the months label to be too big... Commented Jan 11 at 17:57

You may be asking in the wrong forum. Designers are paid to design, to be original. They devalue tradition, which is just what has worked reasonably well in the past. And reasonably well is all you need. The content of your text is far more important than its appearance. This is not (I hope) advertising. The only real constraint on appearance is that it not be distracting. If you have in mind some specific publisher, they will have a house style. It will be OK. Follow it. It will specify most things like font size for captions.

Figures themselves are more of an issue. The possibilities are too numerous to be covered by any house style and, too often, are produced by some software with whatever the default options are. If those options are customizable, consider them carefully. Do not make labels too small. Do not overcrowd graphs. DO look at what others do, and copy the ones that you can understand. You cannot control how the printer will size and position your figure. Prepare for the worst case.

  • Good Designers do not devalue tradition, often even embrace tradition. They should know the rules what is 'allowed/best'. While its true they sometimes bend/break these rules in order to make it more original. But designers also are able to give non-designers simple best-practices guidlines that work in 95% of cases. Commented Jan 11 at 18:03
  • True, I spoke in anger because several here have devalued tradition. And scientific typography is one field where obtrusive novelty is simply abhorrent, not to mention self-defeating. Non -designers need only follow reasonably good precedent in their drafts.
    – Philip Roe
    Commented Jan 12 at 3:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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