Bottom margin vs. descenders question

Axel E. Retif axel.retif at gmail.com
Sun Jul 5 16:33:14 CEST 2020


On 7/5/20 8:03 AM, Peter Flynn wrote:


> On 05/07/2020 04:45, Axel E. Retif wrote:

>> [...] ---if you put an «x» and an «o» big enough together, you'll see
>> that the «o» is really a little bit bigger than the «x», otherwise
>> the eye will see the «o» («e», etc.) smaller.

> Why are you talking about width? Neither of them descends below the 
> beard of the type, and the question was about descenders.

Not width, but height. Please see the page I referenced in my first 
reply to this thread:

https://material.io/design/typography/understanding-typography.html

There you can see that round letters are a little bit bigger in height 
than letters like the «x»; it's more clear in the section “Baseline”, 
where you can see that the round part of the letter «d» is a tiny bit 
bigger than the «x». That was just to exemplify that what's important in 
typography is what the eye sees, not what rulers say.


> Let me rephrase it: if a face uses descenders which go below the beard 
> (and must therefore be set leaded), will a publisher take the measure of 
> page depth a) to the baseline, b) to the foot of the beard, or c) to the 
> bottom of the lowest descender (if there is one)?

What is important is a), the baseline, because in confronted pages one 
last line can have descenders (p, q, etc.) and the other no descenders 
at all ---what the eye sees as bottom of the pages are the baselines 
(the x, serifs of m, n, etc., not the serifs of the descenders of p's 
and q's). That's why in displayed formulas at the bottom of a page 
(La)TeX sets the sum signs at the baseline, but protrudes down integrals 
a little bit.

Please bear in mind that I said “what is important is the baseline” 
---to set the height of a page, for example with the geometry package, 
one should use a multiple of the baselines; let's say, if one is 
typesetting 11:13, the height of the page should be a multiple of 13 (in 
points, of course).


Best

Axel


More information about the texhax mailing list.