Users need to perceive, comprehend and operate a digital service:

  • The service needs a way to get across to the user in order to be perceived. This could be visual (text and images), aural (through a screen reader), and so on.
  • The service needs to be comprehensible — the user needs to be able to understand what’s trying to be communicated to them.
  • The user then needs to be able to easily act upon the service — to operate it.

One key element of digital services for the NHS is content. We need to communicate and be understood using the written word viewed on a huge range of devices in a huge variety of contexts.

Given that content itself is a major part of our interface, choosing the right typeface is vital in making our services both perceivable and operable. We need our users who read text on screen to be able to read comfortably and easily, understand, and then be able to act on what they’ve read accordingly.

When choosing a typeface we have some considerations to take into account:

  • legibility and readability
  • historic connections and themes
  • tone and character
  • brand consistency
  • cost

In this post I’m going to dive into legibility and readability. So first up, a couple of simple definitions.


The Oxford Dictionary defines legibility as “the quality of being clear enough to read”. If something’s highly legible, it’s easier to recognise letters and words.

In the diagram below I’ve set out a sentence in four styles of font: blackletter, script, serif, sans-serif:

The sentence 'I'm sorry Dave, I can't do that' in four font styles

In the next diagram I’ve set out some commonly confused characters:

1, i, I, l and L. Then r, n and m. Then 0, O and o

You can see how typeface design balances legibility with other characteristics such as style and personality.


“Readability” can be a synonym with “legibility” but in this instance we’re talking about it in a specific way. Here’s Ellen Lupton, from her “Science of Typography” article:

… ‘readability’ describes the ease with which a text can be understood (as in the mental processing of meaningful sentences).

In this sense legibility and readability can be treated separately. Legibility is one piece of the readability puzzle, along with typesetting, layout, and the content itself.

What can we discover about the concepts of legibility and readability?

In the world of graphic design, rules have developed over time that help us achieve a legible and readable layout — pick a clear typeface, set your measure, size and scale comfortably.

Sitting alongside our graphic conventions, there’s a wealth of academic research into how we read, and it’s an extremely interesting rabbit hole to disappear down for a while (see below for some further reading). Finding out about models of reading and recognition, and examining the conclusions drawn, can help us with our own typeface research.

Recognising words

There are various theories about how (in general) people recognise words. Here’s three from Kevin Larson’s “Science of Word Recognition”:

Recognition of “word shapes”

Animation showing the outline of the word 'shape'. The word then fades in.

Fast sequential recognition of individual letters

Animation revealing the word 'sequential' one letter at a time

Parallel recognition of all letters within a word, which are then matched to an internal lexicon

Animation showing the word 'work' being matched from 'fork', 'word' and 'work'

It feels like the parallel recognition model essentially takes the first two models and combines them into something more complex and realistic. Even to a n00b like me, either of the first two models alone seems lacking.

Reading and understanding

Ideas around how we recognise words feed into theories of reading. How is it we gain understanding from these groups of symbols?

Again I found two common models with a third model that combines the first two:

“Top down” processing

This model promotes what the reader brings to the text in terms of experience. You bring meaning to text by understanding sentences and paragraphs. Think about using the surrounding sentence and grammar to make an educated guess about the meaning of a word you don’t know — you’re bringing your own knowledge to this collection of words.

This feels similar to the “word shape” theory of word recognition — starting from the abstract level if you like.

“Bottom up” processing

The reader proceeds from letter recognition to word recognition, then to sentence understanding, then to paragraph understanding, and so on.

There feels like a similarity here to the “fast sequential” theory of word recognition — we’re moving along in a linear fashion. It feels very physical, in contrast to “top down”.

“Interactive” or “metacognitive” reading

This model simply accepts that both top down and bottom up reading can happen simultaneously. The two processing modes interact as a reader progresses through a text.

You can bring all kinds of tools to bear in order to gain understanding of a text, from the atomic level letter recognition all the way through to second guessing what an article is about simply from the title:

Reading is at once a perceptual and a cognitive process. It is a process which bridges and blurs these two traditional distinctions. Moreover, a skilled reader must be able to make use of sensory, syntactic, semantic, and pragmatic information to accomplish the task. These various sources of information appear to interact in many complex ways during the process of reading

Context (and preference) is king

Context plays a huge part in our ability to recognise words, read and comprehend — and has a direct effect on the work that design and typography have to do.

Think about the difference between settling down in a quiet place with a novel, and reading your current speed from a display whilst driving. Think about the myriad of situations in between these two extremes — for example a printed newspaper needs to advertise itself (front page headlines) as well as provide content.

Think about the difference between printed material and digital screens. Do you prefer to read from paper or a screen? Which feels harder? Do you feel like you understand and retain more or less from a particular medium?

How do typefaces fit in?

The physical act of reading fundamentally requires us to be able to recognise letters and words. Making text legible requires us to make letter recognition as easy as possible. We can also aid “top down” aspects of reading with our content structure, typography, and layout — allowing users to scan and pluck out what they need without necessarily being forced to read every single sentence.

Research directly connected to typeface seems to split into two general contexts: studies to do with long form reading and studies to do with glance reading.

Studies into “glance behaviour” allow us to get an idea of how legible a typeface can be by restricting the time it can be viewed to a “glance”:

Glance test animation which flashes a word which may or may not be misspelt for 100ms

Interestingly, there are some type foundries bridging the gap between design and research.

MIT Agelab and Monotype conducted research into glance behaviour, asking participants to perform menu selections on a device (i.e. a satnav or similar) whilst in a driving simulator.

This research pitted Frutiger against Eurostile, and concluded that the humanist Frutiger was easier to read and comprehend in this context. Since then Monotype have developed Burlingame with a specific focus on glance legibility.

Another foundry, FontSmith, is carving out a domain in accessibility and legibility with FS Me (in collaboration with Mencap) and FS Millbank.

Readability studies run more towards measuring overall reading time versus understanding and / or fatigue. It’s important to note that increased reading speed doesn’t always result in better comprehension. With eye tracking technology (as it happens, one piece of our user research arsenal), researchers can find out more about patterns of reading.

What’s interesting here is that studies seem to show minimal differences in reading ability between things like serif vs sans-serif and so on. People can get through regardless.

Lupton puts it beautifully in her “Science of Typography” piece:

What we might expect from the science of type is a seamless web of rules. Such is not forthcoming. In its drive to uncover fixed standards, the research has affirmed, instead, human tolerance for typographic variation and the elasticity of the typographic system.

What have we learnt?

In a way, a lack of hard scientific findings around typeface choice and readability is reassuring. Humans certainly aren’t simple, and we’re certainly not homogenous. But our ability to recognise and understand textual communication seems to be pretty adaptable and robust.

Underlying readability is the foundation block of legibility. If you’re having trouble recognising the letters or the word in front of you, then this is going to be a problem.

It seems sensible to form a few opinions that we can use as a yardstick:

  • It’s not “all bets are off” — the aesthetic conventions we work with haven’t been disproved.
  • Distinctive characters in a font can aid legibility and word recognition.
  • Open “humanist” style typefaces can be regarded as having higher legibility.

We can test our decisions:

  • with “glance testing” like the example above
  • in context — such as asking someone to find a particular heading on a phone
  • using eye tracking to help give us an indication of how readable our pages can be

The typeface of choice for NHS.UK digital services needs to be inclusive and legible. It will be one piece of a complex system of design and content, but getting it right will be a sizeable step along the way to our goal of being operable, perceivable and comprehensible.

Some further reading

It’s the tip of the iceberg, but here’s some interesting reading to get into this stuff:

A great jumping off point from Ellen Lupton
The Science of Typography

Kevin Larson and the science of word recognition
The Science of Word Recognition

Bringhurst’s principles of typography applied to the web by Richard Rutter

Some more definitions of legibility and readability from (Monotype)
Fine typography: legibility
Glossary: legibility
Glossary: readability

Sans versus serif
Serif vs sans for text in print
Serifs and font legibility

Does a large x-height make fonts more legible?
Type anatomy: x-height

Testing glance legibility
Using Sunflower Plots and Classification Trees to Study Typeface Legibility
Study: Typefaces Key to Reducing Driver Distraction
Monotype & AgeLab Develop Streamlined Method for Testing Typeface Legibility under Glance-Like Conditions
Revealing Differences in Legibility Between Typefaces Using Psychophysical Techniques: Implications for Glance Time and Cognitive Processing

Looking at readability
The Effect of Typography upon the Perceptual Span in Reading
A Study of Fonts Designed for Screen Display

About preference
Screen vs. paper: what is the difference for reading and learning?
Subjective Impressions Do Not Mirror Online Reading Effort: Concurrent EEG-Eyetracking Evidence from the Reading of Books and Digital Media

More on Monotype’s Burlingame
Introducing Burlingame, A Safer Font For Your Dashboard
A New Typeface Designed for Split-Second Legibility

More on FontSmith and FS Me
FontSmith: FS Me
FontSmith case study: Mencap
The New Face of Mencap

Stop press: a good post drawing on lots of legibility research
Letter and symbol misrecognition in highly legible typefaces for general, children, dyslexic, visually impaired and ageing readers


  1. Comment from kairo

    So, what font is going to be used? Is it one that already exists (i assume there are lots of fonts which fit this criteria)
    Will this research lead to developing an entirely new font? Is that cost effective?

    • Comment from Mat Johnson

      Hi Kairo –

      We’re looking at developing a shortlist of fonts – this post aims to show some of the things we need to think about when making a choice, as opposed to putting forward a choice right now.

      I think it’s very unlikely we’ll be commissioning at an entirely new font as, like you say, there are going to be existing fonts that should help us with high legibility.

  2. Comment from Ralph


    The discussion of cognition => comprehension routes really reminds me of the project. They’re pushing the glance comprehension aspect of this discussion. Imagine getting your NHS advice at 1000 words per minute 😎

  3. Having used FS Me at my previous employer, I am a huge fan of that for its clarity and style. I’m interested to see you using Neue Frutiger here – might there be a chance we can eventually use that font across all our domains?

    • Comment from Tim Allan

      Hi Andy,

      We’re still working with a number of different typefaces, testing for clarity, readability and legibility. This is an under-researched area so we’re looking to contribute more to the body of research there. You mention FS-Me, and that along with Frutiger Neue is one we are currently looking at. I don’t know the exact position is for it’s use across domain, so if you are involved in any projects that want to use it, I would check you have the appropriate license in place.


Leave a comment