Use a serif face for a novel interior. Not because it is more readable, but because it is what readers expect a book to look like, and because the strongest evidence says the choice does not affect legibility either way.
The claim that serifs aid readability is not supported by the research. It is one of the most confidently repeated ideas in book design, and the study that isolated serifs as the only variable found nothing.
The Study That Settles It
Arditi and Cho built custom fonts differing only in serif size, at 0, 5 and 10 percent of cap height, and measured both legibility thresholds and reading speed. Their conclusion:
"Five percentage serif fonts were slightly more legible than sans serif, but the average inter-letter spacing increase that serifs themselves impose, predicts greater enhancement than we observed. RSVP and continuous reading speeds showed no effect of serifs... our data exhibited no difference in legibility between typefaces that differ only in the presence or absence of serifs."
And from the full text: "In Experiments 2 and 3, the presence or absence of serifs made no difference in reading speed, for all participants, both normally sighted and those with low vision."
That design matters. Most font comparisons change several variables at once, so any difference could come from x-height, stroke weight or spacing rather than serifs. This one held everything else constant. The small advantage that did appear was traceable to the extra letter spacing serifs impose, not to the serifs themselves.
A 2022 print study reached the same place from a different direction. Comparing Times New Roman against Helvetica in a paperback layout with 36 readers: "For the paperback layout, no significant difference in reading time (p = 0.50) or reading speed (p = 0.56) was found between the two fonts... Thus, these two font types can be used as interchangeable standard typefaces."
The authors of a 2025 review put the general finding plainly: "there is no such thing as the most legible typeface, as typeface legibility varies depending on the reading situation."
What the Accessibility Bodies Say
You would expect them to prescribe sans serif. The UK Association for Accessible Formats explicitly declines to:
"Many people express a preference for either a sans serif typeface or a serif typeface. However, there is no definitive evidence as to whether serif or sans serif typefaces are easier to read for most people, and some evidence that it makes no difference. Examples of legible typefaces include Arial, Verdana, Trebuchet, Times New Roman."
Note that Times New Roman, a serif, is on their list of legible faces.
What they care about instead is letterform differentiation: "It will therefore be easier to read a document that has a clear uncluttered typeface, and preferably one in which different letters are clearly distinguishable." That is the criterion that actually matters, and serifs are not what determines it.
Two caveats worth carrying. UKAAF's own summary checklist contradicts the passage above by recommending "sans serif", so cite the detailed guidance rather than the checklist. And the American Printing House for the Blind takes the opposite line, stating that fonts for low vision "should not have serifs", with no reasoning or citation attached.
Where Serifs Do Show an Effect
Two studies find something, and both are narrower than the folklore.
In readers with low vision from a specific optic neuropathy, serifs helped only in combination with uniform stroke width: "low stroke contrast... yielded significantly lower thresholds and higher sensitivity when a font contained serifs relative to no serifs." That was 19 patients with a particular condition, on screen.
The same research group's earlier work found the effect is an interaction, not a rule: "Sans serif fonts with low stroke contrast yield better performance and if a serif font is used, high stroke contrast yields better performance than low stroke contrast." That cuts against a simple serif preference in either direction.
One study frequently cited as serif evidence is not. A 1996 print study found normal readers 5 percent faster with Times than Courier, but Courier is a monospaced slab serif, so the comparison is proportional versus monospaced spacing, not seriffed versus unseriffed.
So How Should You Choose
Since legibility does not decide it, other things do:
Convention. Novels are set in serif faces. A sans serif novel interior reads as a design statement, and most novels are not making one.
Letterform differentiation. The criterion the accessibility bodies actually name. Check that capital I, lowercase l and the digit 1 are distinguishable, and that b, d, p and q are not simple mirrors of each other.
Stroke contrast. High-contrast display faces like Bodoni thin out badly at body sizes on uncoated paper. That is a real print consideration and has nothing to do with serifs.
Test at size, on paper. Screen previews flatter type. Print a page at your actual trim size and type size and read it.
Dyslexia Fonts Do Not Work
Related, and worth stating because the belief is widespread. A 2026 meta-analysis of 15 studies, 91 effect sizes and 688 participants:
"the results indicate that dyslexia-friendly fonts have no consistent or reliable effect on reading performance in terms of speed or accuracy. The overall effect size was negligible, g = -0.04, 95% CI [-0.15, 0.07], z = - 0.67, p = 0.5."
A study of 170 children with dyslexia found they "did not read text written in Dyslexie font faster or more accurately than in Arial font. The majority preferred reading in Arial."
The one positive result dissolved under control. When Arial's letter spacing was matched to Dyslexie's, "the difference in reading speed was no longer significant", and the authors concluded "the efficacy of Dyslexie font is not because of its specially designed letter shapes, but because of its particular spacing settings."
Spacing does work, though. A study in PNAS found "a simple manipulation of letter spacing substantially improved text reading performance on the fly (without any training) in a large, unselected sample of Italian and French dyslexic children." So the intervention that helps is spacing, available in any typeface.
What Deckle Offers
Deckle ships 24 body fonts, mostly serif faces suited to book interiors, with several sans serif options.
Two are worth explaining in light of the above. Atkinson Hyperlegible is designed around exactly the criterion UKAAF names, maximising the differentiation between similar letterforms rather than betting on serifs. OpenDyslexic is included as an option, and on the evidence above you should not expect a reading gain from it. Offering the choice is reasonable, since some readers prefer it, but preference and performance are different things.
If accessibility is your goal, the more effective levers in Deckle are the ones the evidence supports: type size, line spacing, and turning justification off, which both WCAG and UKAAF ask for.
For the wider typeface decision, see choosing a typeface for a book interior.
Common Mistakes
- Believing serifs guide the eye along the line. No study supports it, and the one isolating serifs found no effect.
- Switching to sans serif for accessibility. UKAAF says the evidence does not exist, and lists Times New Roman as legible.
- Adopting a dyslexia font expecting a reading gain. Meta-analytic effect size is essentially zero.
- Choosing a high-contrast display serif for body text. It breaks up at small sizes on uncoated paper.
- Judging a typeface on screen. Print it at final size.
FAQ
Q: Is serif more readable than sans serif? A: No. The study isolating serifs as the only variable found no difference in legibility or reading speed.
Q: Why are books set in serif then? A: Convention and reader expectation, which are legitimate reasons. They are just not legibility.
Q: What about readers with low vision? A: The evidence is mixed and interacts with stroke contrast. Type size and spacing matter far more.
Q: Does OpenDyslexic help? A: The meta-analysis says no. Where a benefit appeared it came from letter spacing, not letterforms.
Q: What should I actually check when choosing? A: Whether similar letters are distinguishable, and how the face holds up printed at your real body size.