On Tue, 17 Feb 2009 20:00:21 -0500, Ron Koster wrote:
> At 10:02 AM 2/13/2009 -0800, David Hucklesby wrote:
>> I find that these percentages work best
>> cross-browser: 69%, 75%, 82%, 94% ... with a base font-size of 100%.
>>
[...]
>
> Firstly, from past threads, my understanding is that one shouldn't be going
> any smaller than 100% -- or at least should try not to -- if only to be in
> keeping with whatever it is that any particular user has set their own
> settings at, so wouldn't going as small as 69% (or whatever) be *too* small?
>

For the main content, yes, I agree. But side notes, copyright notices,
and the "small print" I'd use the smaller sizes, although I avoid smaller
than 75% myself.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

> Secondly -- and perhaps more importantly -- in recommending those specific
> percentages, are you saying that things go funny in some browser or platform
> if those exact percentages aren't used -- like if I used, say, 76% or 85% or
> something?
>

When scaling in EM sizes, for margins, padding, etc., I found these
sizes give most consistent results. I tested scaling from 1px to 40px
at a nominal 16px for the BODY font-size (at 96 DPI). Using the
calculated theoretical values, e.g. 13/16 to get 81.25% for a nominal 13px
font-size actually gave me 12px in Safari, which seems to ignore anything
after the decimal point.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 

> And what about for percentages higher than 100% (for headings or whatever)?
>

Use the calculated values - 125% for 20px for example. The rounding
issue does not seem to affect larger sizes so much, although rounding
up to the nearest whole percent, or .01em, works best here, too.


Cordially,
David
--


______________________________________________________________________
css-discuss [cs...@lists.css-discuss.org]
http://www.css-discuss.org/mailman/listinfo/css-d
List wiki/FAQ -- http://css-discuss.incutio.com/
List policies -- http://css-discuss.org/policies.html
Supported by evolt.org -- http://www.evolt.org/help_support_evolt/

Reply via email to