Print size and low vision: what the research says, and why e-ink is its own case
Accessible type is usually described as bigger type. The reading research describes a range with a real top end, a required size that changes with what you are reading, and one constraint that e-ink screens hit harder than backlit ones.
· 7 min read
AccessibilityE-ink basics
Ask what makes text accessible and almost everyone answers the same way: make it bigger. That gets you most of the distance. Vision science has been more specific for about forty years, and the specifics are the interesting part. There is a floor. There is also a ceiling. Which one binds depends on the reader, on the task, and on the screen.
Nobody publishes the number you want
Start with the surprise. WCAG 2.2 does not specify a minimum font size. Not in the success criteria, not in the Understanding documents, not in the cognitive accessibility note, not in the low vision requirements draft. I went looking and it is not there.
The 18 point and 14 point bold figures people quote from WCAG do something else. They define “large scale” text, and their only job is to relax the contrast requirement: normal text has to reach 4.5:1, large text only 3:1. A note on that definition says where the numbers came from, and it was not a vision experiment. They were borrowed from large print publishing convention, 14 point being the minimum size for large print and 18 point the larger standard size.
What WCAG asks for instead is a capability. Success Criterion 1.4.4 says text can be resized up to 200 percent without loss of content or functionality. The reader gets to decide.
Platform guidance is thin the same way. Apple’s design tips page gives 11 points as a legibility floor, close to the only hard number a major platform publishes. The fixed figures come from print instead: RNIB says 14 point ideally and 12 point minimum, and UKAAF sets large print at 16 point minimum with 18 recommended. Both describe documents you read straight through, which turns out to matter.
The good range has a top end
Gordon Legge and Charles Bigelow reviewed this literature in the Journal of Vision in 2011, and the shape they describe is a plateau with a cliff at one end and a slope at the other. Below a threshold called the critical print size, reading speed falls sharply. The consensus value for normally sighted readers is an x-height of 0.2 degrees of visual angle. Above that the plateau runs to roughly 2 degrees, a tenfold span they call the fluent range. At a 40 cm reading distance that means x-heights of 1.4 mm to 14 mm, or about 4 point to 40 point.
Then the part people skip. “Reading speed also declines for big print,” they write. The decline sets in between 1 and 2 degrees, gentler than the cliff at the small end, and they note this end is less well studied. It is a measured effect rather than a matter of taste. Type can be too large.
How much bigger depends on the job
Xiong and colleagues put numbers on this in 2018, measuring acuity reserve: the ratio between the print size a task requires and the smallest print a reader can manage at all. Their median values across low-vision subjects:
| Task | Print size, as a multiple of threshold |
|---|---|
| Spot reading a label, 40 wpm | about 1.2× |
| Fluent reading, 80 wpm | 1.9× to 2.1× |
| Reading at maximum speed | 3.0×, or 4.4× with macular involvement |
Three to four times more size for a paragraph at full speed than for a price tag. That spread is the most useful thing here for anyone laying out a page, because most pages carry both kinds of text. A date in a calendar cell is spot reading. A block of notes is not.
The far end deserves plain speech too. Legge published a simulation of 18 point text at 40 cm as it appears at declining acuity. At 20/200 the line breaks and the gaps between words are still visible and the characters are not legible at all. Careful typesetting is no substitute for magnification or speech.
Some readers want it smaller
The W3C’s low vision requirements draft has a line that cuts against the whole instinct: some people with tunnel vision and good visual acuity may prefer smaller letters so they can see more words at a time. UKAAF says the same from the print side. Which is why every text requirement in the W3C document is phrased as something the user can change rather than a size the author should pick. The standard declines to name a number on purpose.
Spreading the letters out backfires
Crowding is the interference one letter creates in recognising its neighbours, and it worsens in peripheral vision, which is where anyone with a central scotoma does all their reading. So extra space between letters ought to help. It has been tested repeatedly and it does not.
Chung, and separately Yu and colleagues, varied letter-to-letter spacing in a fixed-width font from half of standard to twice standard with normally sighted readers. Speed peaked at standard. Legge and colleagues ran it on four low-vision subjects reading heavily magnified text, and every one was fastest at standard spacing. Chung repeated it in 2012 with 14 subjects who had central vision loss and got essentially the same result.
Line and word spacing behave differently, showing either no benefit or a small one, and Blackmore-Wright and colleagues found that doubling both together did help subjects with macular degeneration. So the rule is narrower than “spread everything out.” Give the lines and the words room. Leave the letters where the font designer put them.
A small screen runs out of room before it runs out of size
Print size and display size pull against each other. Atilgan, Xiong and Legge measured the tradeoff in PNAS in 2020: reading prose, a line has to carry a minimum number of characters before a reader reaches 80 percent of maximum speed, 13 for normally sighted readers and 8 for low-vision readers, and that count barely moves across fonts or across laptop, tablet and phone. Their conclusion is the one that matters here. A small display, or a reader who needs magnified print, can shrink or entirely eliminate the range of sizes at which maximum reading speed is possible. That was measured on prose, note, and it does not govern a two-digit number sitting alone in a box.
Why e-ink is its own case
Everything above applies to paper and to any screen. E-ink adds a constraint a backlit display does not have, and it is about contrast rather than size.
An e-ink panel is reflective. It makes dark marks on a light background by moving pigment, and that background is never paper white. It cannot be, since it has no light of its own. Crossland and colleagues measured maximum contrast on an Amazon Kindle and a Sony e-reader at about 63 percent Michelson contrast, where 100 percent is the maximum available. Legge reports the figure and adds the part that matters: satisfactory for people with normal vision, problematic for some low-vision readers.
The two verdicts differ for a reason. People with normal vision identify letters down to 1 percent of maximum contrast and hold their top reading speed down to about 5 or 10 percent, so 63 percent is enormous headroom. People with low vision typically have reduced contrast sensitivity, and for some of them reading speed drops with any reduction from full contrast at all.
A measurement lines up with this. Gill and colleagues tested 27 patients with stable age-related macular degeneration in 2013, reading matched print sizes on paper, an Apple iPad and a Sony e-reader. The iPad was fastest, paper next, the e-reader last, and the authors attributed the difference to the bright, high-contrast display of the iPad. Ambient light widens the gap: contrast on any display drops under veiling light from a window or a lamp, a brighter display dilutes less, and a reflective panel has the least to give away.
So on e-ink, contrast is the scarcer resource. Gray secondary text that looks refined on a laptop costs real legibility on a reflective panel, and it costs most for the readers with the least to spare.
What this means for a document you use every day
Offer more than one text size and let the reader choose, because no standard will name a size for you. Do not treat small type by tracking the letters apart, since that is the one adjustment measured to hurt. Print the elements a reader has to hunt for in black rather than gray, because contrast costs no page area. And when a page divides into seven day columns on a four-inch screen, accept that large print is arithmetically out of reach there and say so.
The instinct to make it bigger is a good one. It runs out sooner than people expect, and on e-ink there is usually a cheaper win sitting right next to it.
Sources: WCAG 2.2, the W3C low vision requirements draft, UKAAF G003, RNIB’s Clear Print guidance, Legge & Bigelow 2011, Legge 2016, Xiong et al. 2018, Atilgan et al. 2020, and Gill et al. 2013.
Planned Pages generates a hyperlinked planner sized for your exact tablet, and your first month is free. Related: what makes a PDF slow on e-ink and choosing a weekly layout.