Responsive Type and Zoom

Typography that responds to viewport width (‘fluid’ or ‘responsive’ typography) can be useful when you want to ensure text does not get clipped or spill out of some design elements. Carousels, widget controls, or my Venn diagram are some examples.

I say viewport width because I rarely see responsive type consider the viewport height or the printed page. Not considering height can bw problematic for users holding their phone in landscape mode while trying to navigate pages with cookie consents and email form sign-ups. Not considering print can make for reams of wasted paper.

Responsive vs. Static

The alternative approach is to just use static text sizing, ignoring the size of the viewport.

To demonstrate the difference I forked a five-year-old Codepen that had just been updated last week, Precision responsive typography and started zooming the page, capturing screen shots along the way.

Then I duplicated the pen, removing all the responsive sizing styles, ensuring the text started at the same size in an un-zoomed window. Then I zoomed and screen-shat it.

Screens

All screen shots were captured in Windows 10 / Firefox 72.0b1, but the effect is the same across browsers.

The text in the screen shot is the same as the following image. The text in the screen shot is the same as the previous image.
The first/left image is the responsive type example with no zoom. The second image is the same HTML, but no responsive type, with no zoom. The text is the same size in both.
The text in this screen shot is smaller than the following image. The text in the screen shot is larger than the previous image.
The first/left image is the responsive type example at 200% zoom. The second image is the same HTML, but no responsive type, at 200% zoom. The size difference is apparent, but not overwhelming.
The text in this screen shot is significantly smaller than the following image. The text in the screen shot is significantly larger than the previous image.
The first/left image is the responsive type example at 300% zoom. The second image is the same HTML, but no responsive type, at 300% zoom. Firefox users cannot zoom any more, which means the text cannot get any larger for them. The text cannot get to truly 200% its original size.

While Firefox limits zoom to 300%, Chrome goes higher and at 400% zoom the user of the responsive type page can finally get the text to almost 200% of its original size.

Code

For comparison, I have embedded the two different CSS blocks within this poorly-styled details/summary thinger…

The original SCSS (comments stripped):

$min_width: 400;
$max_width: 800;
$min_font: 12;
$max_font: 24; 

:root { font-size: #{$min_font}px; }

@media (min-width: #{$min_width}px) and (max-width: #{$max_width}px){
  :root { 
    font-size: calc( #{$min_font}px + (#{$max_font} - #{$min_font}) * ( (100vw - #{$min_width}px) / ( #{$max_width} - #{$min_width}) ));
  }
}
@media (min-width: #{$max_width}px){
   :root { 
     font-size: #{$max_font}px;
   }
}

The entirety of the text sizing CSS in my pen:

:root {
  font-size: 150%;
}

Why It Matters

I have worked with users who scale text to this large. Some of them do it because they surf on their TV from their couch. Some do it when reading a recipe off their phone in the kitchen. Some do it when they need to present on a wall to a room full of a people. Some of them just have poor vision.

When people zoom a page, it is typically because they want the text to be bigger. When we anchor the text to the viewport size, even with a (fractional) multiplier, we can take away their ability to do that. It can be as much a barrier as disabling zoom. If a user cannot get the text to 200% of the original size, you may also be looking at a WCAG 1.4.4 Resize text (AA) problem — check out Failure of Success Criterion 1.4.4 due to incorrect use of viewport units to resize text.

I want to be clear — I am not picking on the specific example above. That example is only demonstrating what developers can do, not what they should do. There may be completely valid reasons to use these techniques and the example I borrowed shows one method to do that.

What to Do

Identify why you want responsive type. Is it based on user request? Surveys? Research? Or is it just that you or your team think it looks better? Or you want to try out this technique? The answers may tell you if you should even move ahead with responsive type.

Consider not setting a base font size. Maybe use the following rule (or similar or nothing) to inherit the font size from the browser, which may have been explicitly chosen by the user:

:root {
   font-size: 100%;
}

Then only set subsequent text size values using %, em, or rem units, avoiding values below 100%, 1em, or 1rem (unless scaling down in something already scaled up).

Be careful when using vw or vh units. Then be careful if using calc(). Be even more careful when using min(), max(), or clamp() (see CSS Values and Units Module Level 4 for the Working Draft spec language).

If you are going to use responsive typography techniques anyway, you must test it by zooming. Zoom across devices, across browsers, across viewport sizes (not everyone surfs full-screen), and across viewport orientations.

Also, don’t forget print styles. Consider print units in pt, and print to PDF to confirm it works without wasting paper.

Update: 7 January 2020

New York Times has an experimental interactive piece that demonstrates some of what I discuss above. Zooming out from a page should not make the text larger than zooming in.

At 80% zoom; about five words fit per line. At 110% zoom; about seven words fit per line. At 133% zoom; about six words fit per line.
The text is largest when at 80% zoom (first image), then much smaller at 110% zoom (middle image), then a bit larger at 133% zoom (last image).

Update: 20 May 2020

Scaling text by using the browser preferences to change the default text size will not affect any text on a page that is set in px units.

To say it a different way, if you set your text in px, then it will not scale when a user explicitly chooses a larger or smaller default size. You px-based text will stay the same size, laughing at your user.

Obviously full-page zoom overrides it, but many users still opt for a larger default size to avoid having to scale every site. So don’t set your text in px.

Update: 26 September 2020

Apparently I need to keep saying this. The following is becoming my default comment on articles that keep appearing without cautions.

Please be careful with maximum text size, particularly on sites/pages that face the general public or employees. If you prevent the text from scaling up 200%, then that is a WCAG SC 1.4.4 failure at Level AA. Viewport units have their own call-out as a major risk in WCAG.

No matter what technique you use, be sure that the page text can be zoomed at least 200%. And yes, I have written about responsive type and zoom, and have cautioned against min(), max(), and clamp().

Update: 29 September 2020

This in response to a specific request on the post Linearly Scale font-size with CSS clamp() Based on the Viewport. I referenced a simpler example from the post in my comment, but figured I would use the final example here to be sure I caught all the affordances in the code.

See the Pen Fluid typography example by Pedro Rodriguez (@pprg1996) on CodePen.

I made a video showing that as I zoom to 200%, the text stops scaling at about the 150% point.

Here I compare the original text, the page scaled to 200%, and then how the text should look if it was actually made it to 200%.

Unzoomed, about 7 lines of text fit in the window. At 200% zoom, about 4 lines of text fit in the window. When the text is actually scaled to 200%, maybe one line would fit in the window.

Here I show the <h1> at its initial size, when the page is zoomed to 200%, and when the text is actually 200% the size of the initial text. I aligned them on the baseline to make the difference between them more obvious.

The zoomed text is maybe 25% taller, with the actual 200% take showing how much different they are.

Similarly, even a simple demo taken out of context can be a problem. Your text should never get smaller when the user zooms, and it certainly should not be smaller at 300% zoom.

Update: 17 October 2020

Over at dev.to, Google’s developer support blog, Una Kravets posted min(), max(), and clamp(): three logical CSS functions to use today where she walks through how each of those CSS functions works and shows some examples.

There is a fluid typography example she borrowed from elsewhere that shows how text can scale and adjust to the viewport, using only the style declaration font-size: clamp(1.5rem, 5vw, 3rem). A video shows it in action.

If you have read this far, you may have spotted something that can cause a problem here. It’s not the use of clamp(), since the upper limit, 3rem, is twice the size of the starting 1.5rem. It’s the use of vw units. For some users, the text can never get large enough to bump up against the 3rem.

I made a video to demonstrate it in action, but you can visit the original demo to try it yourself. Remember that Firefox can only zoom to 300%.

This is a 1,024 pixel window in width, though I kept the height short and it has no impact on the demo. The text starts at 64px tall, at 200% zoom has only increased to 68px in height (a 6.25% increase), and at 300% zoom has only increased to 96px in height (a 150% increase).

If you use this code as-is, you have guaranteed a WCAG failure.

I filed a pull request (which they merged on 21-Oct) with a brief explanation:

When you use vw units or limit how large text can get with clamp(), there is a chance a user may be unable to scale the text to 200% of its original size. If that happens, it is WCAG failure under 1.4.4 Resize text (AA) so be certain to test the results with zoom.

…before merging another PR that removed the links. So, shrug emoji?

Update: 27 September 2022

I completely failed to link to W3C issues that may or may not have some impacts here:

Not a lot of movement recently, but it is instructional to see the challenges and goals outlined by all parties.

Update: 17 November 2023

I kinda like this:

Mudford cites Adrian Roselli, who appears to be the core source of the other warnings:

That may very well be true. I did not find others citing the very real (and easy to prove) WCAG failures developers were suddenly pushing into their projects. That was a little depressing, frankly (considering how easy it is to prove). I spent a lot of time grumping about this online, too. So yay me.

Anyway, Maxwell does the thing I did not — he tried to plot the curves and do the math to give developers guidance on how to write their CSS functions to create WCAG-conformant fluid type. I have not tested the assertions, but I definitely appreciate the effort.

It only took four years!

16 Comments

Reply

Hey Adrian, I’m working on a design system that’s using a vw plus rem scaling font-size: calc(1rem + 0.2vw) and it seems to be responding just fine to both browser zoom and font-size settings in the browser config.

Have I not explored all the scenarios in which this kind of responsive type would break or do you think this solution is okay?

Mark Goho; . Permalink
In response to Mark Goho. Reply

Mark, I think 0.2vw is such a small number that it might not matter. But be aware that for a user who zooms, the vw value in your calc() will not change with zoom. But since the rem is the major value it may not be a big deal except in very wide windows.

As always, test those edge cases so you at least know what happens.

Reply

I couldn’t replicate the zoom discrepancy between https://cdpn.io/aardrian/debug/XWJmjrN and https://cdpn.io/aardrian/debug/NWPPJRo . Both looks the same at 500% in firefox. What am I missing? Thanks for this post.

ahwei; . Permalink
Reply

Oh sorry I see it now if I view it in the ‘Responsive design mode’. Cool. Thanks!

ahwei; . Permalink
Reply

Hey Adrian, I think I might have a solution that respects zoom while resulting in the same effect.

Basically I’m using javascript to multiply the root font size by some value which is clamped at a min or max. So it respects both a users font size and zoom. Of course it affects all elements that use REM, but that tends to be what I want.

What do you think?

benton328@gmail.com; . Permalink
In response to benton328@gmail.com. Reply

I think if it never prevents a user from getting the text to 200% and it does not spin their fans into oblivion, then sure.

Reply

Hey Adrian, you say “Then only set subsequent text size values using %, em, or rem units, avoiding values below 100%, 1em, or 1rem (unless scaling down in something already scaled up).” What’s the problem with setting a small font to 0.875rem (the equivalent of 14px, I believe)? It will still scale, if the users needs to make it bigger through the browser font settings. In fact in the Comments section on this blog, you use both 80% and 0.8rem, which seems a perfectly reasonable use case…

Jonathan Douglas Holden; . Permalink
In response to Jonathan Douglas Holden. Reply

That statement was in the context of setting the base font size to 100%, which keys off the user’s own settings / preferences.

You are suggesting setting the base font size smaller, using my comments section (which is set to 80% the size of the base size of the rest of the content) as justification.

You don’t note that my content size is 130%, or computed as 20.8px. That means my 80%-size comments are 16.64px in a default browser set-up, or still larger than the user’s default font size (104% of default).

So, to answer your question, I suggest you do not do that. Look at the computed values in the dev tools, not the CSS declarations.

In response to Adrian Roselli. Reply

Aha! I wasn’t suggesting setting the base font to 80%, just noting that you do use 80%. What I didn’t check, as you say, is that the content font size is 130%. So that makes sense. You’re right, I should have checked the computed values! Thanks once again for your time.

Jonathan Douglas Holden; . Permalink
Reply

Hi Adrian,
I’m still trying to unpick the mechanics of WHY this happens. Is it only when you have a prefferred value entirely in vw (because vw doesn’t change)?

Can I ask if this (https://royalfig.github.io/fluid-typography-calculator/) has solved around the issue or is there still a risk?

In response to Nathan. Reply

Nathan, There is a WCAG failure that describes the issue with viewport units alone: Failure of Success Criterion 1.4.4 due to incorrect use of viewport units to resize text.

As for the tool you linked, I don’t know without breaking it apart. However, note how when the slider at the bottom hits a viewport width of 1,950px, the base font stops getting bigger (it anchors at 20px). That comes from the “Max Font Size” set above, which at 1.25rem means a user may not be able to zoom it enough for their needs.

The simplest answer, IMO, is to never set a maximum font size (so do not use clamp()) or, if you only care about minimum WCAG conformance, never set the maximum font size to a value below twice the base font size.

In response to Adrian Roselli. Reply

If it helps, I have done some extra mathematical analysis of this and have found the following….

Even when not applying a max, you cna hit a “200% not possible” violation. The three biggest variables in this outcome are:

(a) the steepness of the change (i.e. if a linear growth – the steeper the growth, the more likely 200% can’t be achieved)

(b) the max zoom of browsers (e.g. if sometimes you need to go to 300% to hit 200% text growth – eventually some browsers won’t be able to deliver the zoom scale needed.

(c) the worst case viewport width (on a growth curve from 2.625rem at 320px to 3.325rem at 1440px viewport and then larger to maintain linear growth beyond that, even at 300% zoom the 200% text increase is not possible on a 3840 screen. Whilst it is likely that a 4k screen is already upscaled, its not guaranteed.

So, since we don’t control screen sizes or browser zoom levels etc, there will always be a risk with responsive text and the only way around it is to avoid it or, when your growth curve hits a snag, add a breakpoint to go back to a value that avoids vw, etc.

Nathan; . Permalink
In response to Nathan. Reply

This is indeed interesting info. At the very least, having something to identify where the growth curve flatlines so you can add a breakpoint is handy.

In response to Adrian Roselli. Reply

A few extra developments. After running a bunch of maths and creating a “Fluid Type” calculator in Excel to try and protect the upper bound, I also have the thoughr that can’t techniques be combined to meet the accessibility criteria?

Say for example your zoom topped out at 194% (never reaching the magic 200%). If the user also changes their browser settings to enlarge text then that would bump it over the 200% (assuming em/rem units are involved).

Thoughts?

Nathan; . Permalink
Reply

Is this no longer an issue? Chrome on Android, Samsung browser, and Safari both seem to be zooming without breaking the layout now when viewport units are used.

Jonathan; . Permalink
In response to Jonathan. Reply

Jonathan, this post deals with desktop browsers, though I could have been more explicit in the post instead of relying on the screen shots to convey that. You have cited mobile browsers where zooming (pinch/spread) is a very different model. Those are analogous to desktop browsers that also use pinch-zoom (via touch screen or track pad).

Leave a Comment or Response

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>