Re: Add Font Family back to adaptation

Hi Wayne,

The problem is that starting here:
“The user can substitute a font family for the author's font family...”

That is a user-agent requirement, not a content requirement.

Also, the user can do that now. Unless you have found a way authors can block that change?

We must take it a step further and identify the content issues when the user does that, what is the harm?

Taking font-icons out of it, the only harm I’m aware of is when the text expands out of the containers. In that case, sizing is the only issue.

If sizing is the issue, then we take a point on the normal distribution of font-sizes (e.g. one standard deviation to the right), and include that as part of the sizing criteria.

As I said, if that is more than about 10% overall then I suspect that it will get enough push-back to shunt into Silver or personalisation, but happy to try.



From: Wayne Dick <>

I would start the bullet with this language.
The user can substituted a font family for the author's font family so long as the the following condition is met.

There ratio of the user's font family character size and the average character size taken over font families in the script for the writing system in use does not exceed 1.2.
The average character size for a font family is the width of the Unicode language block measured in pixels divided by the number of characters in that block.
Note: Unassigned characters will be excluded from the Unicode language block as well as any characters that are identified by language experts as inessential to the count. The character count will thus equal the size of the Unicode block minus the number excluded character codes.

The average character size for font families in a given script is the mean average character size taken across a representative sample of font families.

On Tue, Jun 6, 2017 at 11:30 AM, Wayne Dick <<>> wrote:
The need is conflict pairs in font families. The most common are: Capital I, Lower case l and the digit 1 (that is actually 3 pairs taken 2 at a time). The other pairs are S (capital 'S') and 5 (digit 5), B (capital 'B') and 8 (digit 8), O (capital 'O') and 0 (the digit zero) and in some script cases Q (upper case Q) and 2 (digit 2).
Letter spacing and size are different issues. The capital 'I' and lower 'l' is the worst. Frequently the only difference is a slight difference in height. These pairs do not pose a huge problem in text, but outline numbers and passwords are a different situation.
Use case: You forget your password and get a temporary password on your mobile phone.You are forced to copy the password by hand to your computer, and if you cannot distinguish one of these pairs, then you are in serious trouble.

On Tue, Jun 6, 2017 at 9:19 AM, Alastair Campbell <<>> wrote:
Hi Wayne,

For what purpose? I thought the point was that we assume people can over-ride fonts, therefore we cover that need with sizing?

Having looked at the distribution of font-sizes, do we need to increase the spacing aspects of the current bullets?

Note that there is a balance: Layouts can take a certain amount of buffer before they look odd in regular use. If we push past that point (my feeling is around 10-15% increase) then it will be either rejected by the design & dev community, or moved into a personalisation SC.

Going past that point means a more significant override, so either discarding the author-styles (like Linearise assumes), or adding alternative layouts with personalisation. In either of those cases, it would move out of the text-adaptation SC anyway.

I’m more concerned with colour which isn’t covered yet, does anyone have further ideas on including that?


From: Wayne Dick <<>>
Date: Tuesday, 6 June 2017 at 16:50
To: LVTF - low-vision-a11y <<>>
Subject: Add Font Family back to adaptation
Resent-From: LVTF - low-vision-a11y <<>>
Resent-Date: Tuesday, 6 June 2017 at 16:51

It is time to put font family back into Adapting Text. Steve has addressed the icon substitution issue well. Namely add failure to mark icon fonts to 1.3.1. I have looked at the distribution of font face sizes. Within each writing system:

  1.  Identify a mean character size within the script for that system.
  2.  Compute the ratio R of average character size by family to mean character size over all families in a representative sample.
  3.  Using language experts, establish boundaries that make sense close to 1 standard deviation from the mean.

Note: Mean +/- 1.2 should do based on the physiology of the human eye.


Received on Wednesday, 7 June 2017 08:51:17 UTC