Proposed Final Design for W3C Technical Reports style in 2016

Previous Topic Next Topic
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

Proposed Final Design for W3C Technical Reports style in 2016

Jim Allan-3
Jumping in a bit late. These are my thoughts not those of the Low Vision Task Force.

First, if we change the underline color to #707070 (just a tad darker) and it will pass with a contrast ratio of 5.0:1.

After a bit of research I could not find anything on contrast of only the underlining of links.

But I did find lots of information on reading and links. They are included below. Of note is the first study [1] using eye-tracking to see how we read hyperlinked text. They tested with colored links (no underlines), including grey hyperlinks.  Hyperlinks on "unfamilar" words caused rereading of passages. Color of the hyperlink was not a factor, except grey. Hyperlinks that had poor contrast (grey) caused rereading, because reduced contrast is more difficult to process. 

"Therefore efforts made in Web development to avoid using blue as the hyperlink colour and instead using a different colour may have no positive influence for the reader reading the text, but instead make it more difficult for the reader to know what is a hyperlink when they are expecting it to conform to the convention of hyperlinks being denoted in blue....
A hyperlink is not just a salient word in a passage of text, it denotes that more information that may be relevant lies behind that hyperlink....
These experiments have shown that coloured text does not hinder reading, but also that hyperlinks can cause us to reread previous content if the word is a low frequency/difficult word in order to re-evaluate the content. ...
The key lesson here is that Web designers should only hyperlink important words in pages, taking extra caution with words that are uncommon or ones that may be difficult to process.-

1. On measuring the impact of hyperlinks on reading (Fitzsimmons, 2013) -
pdf of poster session -

2. Guidelines for Visualizing links (Neilsen, 2004) -

3. Should all links be Underlined (Martin, 2007) -

4. Beyond Blue Links: Making Clickable Elements Recognizable (Loranger, 2015) -

5. A Comparison of Link Readability Techniques (Graves, 2004)

6. Comparing Link Marker Visualization Techniques – Changes in Reading Behavior (Obendorf & Weinreich, 2003) -

Jim Allan, Co-Chair Low Vision Task Force (W3C/WAI)
Accessibility Coordinator
Texas School for the Blind and Visually Impaired
1100 W. 45th St., Austin, Texas 78756
voice <a href="tel:512.206.9315" value="+15122069315" target="_blank">512.206.9315    fax: <a href="tel:512.206.9264" value="+15122069264" target="_blank">512.206.9264
"We shape our tools and thereafter our tools shape us." McLuhan, 1964

On Tue, Jan 5, 2016 at 8:43 AM, Jonathan Avila <[hidden email]> wrote:

Ø  I'll gently push back on your assertion, as I believe it is the combination of text and underline - together as a visual whole - that signals to most readers that a link is indeed a link. In other words, it's not "  link text ", nor is it _______, but rather, the combination of the two:  link text


This is also something I’ve always had heartburn over as well.  That is sometimes meaning is communicated not by color itself but by the difference in luminosity.    Technique G183 for links indicate that links must be indicated differently on focus/hover from the surrounding text – but in theory this could be met simply by a focus rectangle by meeting SC 2.4.7 (if A and AA are applicable).


A similar issue applies to change in background color on list items, page tabs, etc. that is focus and selection indication can be met by changing the background and thus communicate focus or selected state by the change in luminosity.  This change in luminosity may not be in the text but may be in the background.  In these cases SC 1.4.3 as written would not apply.   It’s unclear if SC 1.4.1 would apply in this case as it’s not color but the difference in luminosity.  Technique G183 is mapped to SC 1.4.1 and not SC 1.4.3 – which makes me think SC 1.4.1 might be broader than just color but that’s not how it’s written.




Jonathan Avila

Chief Accessibility Officer
[hidden email]

<a href="tel:703.637.8957" value="+17036378957" target="_blank">703.637.8957 (o)
Follow us: Facebook | Twitter | LinkedIn | Blog | Newsletter


From: John Foliot [mailto:[hidden email]]
Sent: Tuesday, January 05, 2016 9:18 AM
To: 'fantasai'; 'David Carlisle'; [hidden email]
Cc: [hidden email]; [hidden email]; 'W3C WAI Protocols & Formats'; 'public-low-vision-a11y-tf'; 'Wayne Dick'; 'Srinivasu Chakravarthula'
Subject: RE: Proposed Final Design for W3C Technical Reports style in 2016


fantasai [[hidden email]] wrote:

> > Fantasai, while I appreciate that you've made the link less faint, the

> > color used (#C0C0C0) still fails a color-contrast test.

> > (,bg=FF

> > FFFF)

> >

> > To be crystal clear, the WCAG 2.0 Recommendation only speaks of “text”

> > when referencing color contrast (so for example, it does NOT speak to

> > icons, etc.), however since **underlined text** is indeed recognized

> > as a link, I would argue that the color contrast requirement would be

> > in play here, as the underlining is part of the active text, and that

> > the visual indicator should be as visible as the text it is underlining.


> It's a reasonable argument on the surface, but actually, I think the contrast

> requirement for an underline isn't as stringent.


Therein lies one of our problems: this has not been definitively clarified. I believe this is an open question for the Low Vision Task Force (now copied on this thread), who are addressing issues related to this user-group that were not addressed in the original WCAG 2.0 publishing time-frame. I'll gently push back on your assertion, as I believe it is the combination of text and underline - together as a visual whole - that signals to most readers that a link is indeed a link. In other words, it's not "  link text ", nor is it _______, but rather, the combination of the two:  link text



> Unlike text, for an underline you

> only need to be able to distinguish that it's there, not distinguish which of a

> variety of shapes it is.


Correct, the low vision user needs to distinguish that it is there. If it is too faint in color contrast, it is not perceivable.


> I've tried increasing the contrast, but I run into a few problems trying to do that.

> I have to balance:

>    * contrast with the foreground color, so that it's visible

>    * contrast with the text color, so that it's easy to visually filter

>      out the link style and focus on the paragraph text

>    * contrast between visited and non-visited links, so that they can

>      be distinguished


> One thing I could do is to swap the darker color for unvisited links, and the

> lighter color (harder to see, but also less intrusive) for visited ones.


Balancing all of the functional requirements you’ve listed, I’d say that this is a compromise that might work, but I am loathe to be the final arbitrator of that decision. W3C also has a mandate to be conformant to WCAG, and how we thread this particular needle will require some consensus. I will take the Action to socialize this further inside of the WAI domain, and specifically will ask the LVTF to weigh in here.



> > Suggestion: could you lighten the line weight, darken it and perhaps

> > use dashes or dots instead? (see in-page links at WCAG -


> > I’m not a graphics person however, so feel free to explore other

> > alternatives. Jonathan Snook’s online color contrast tool is quite

> > useful

> > there:


> Lightening the weight or using dashes/dots instead would allow me to follow

> the letter of the WCAG rule without actually following its

> spirit: lighter-weight or discontinuous lines are perceptually lighter in color, even

> though the screen pixels will test at a higher contrast.

> So I don't think that's actually helping real people, even though it'll help the

> color-contrast checker.


LVTF, do we have any research or feedback on this point that either supports or dispels?



> I can do the opposite, though: make them thicker, so that they're easier to

> perceive even though the colors are the same. :)


> It would look like this:



This looks better *to me*, but again I’ll seek some other feedback going forward. Thanks for being patient Fantasai! J





> ~fantasai