Clarifications about em units, point size and line spacing

Fonts and More General Font Discussions
1

Hello.

As I understand from Erwin’s old post one em is equal to the current point size, and the em-units are to enable us to define a font w.r.t. that. For example, in a font with units_per_em=1024, a point with y=512 will be half the point size above the baseline. On screen that is translated into so many pixels as per the formula indicated by Erwin in that post.

Now I have seen that not all fonts at the same point size have the same vertical advance. When I use two fonts at 12 pt size, it is possible that one font is “taller” than the other. Even when both fonts are being typeset in “single” line spacing, this is so. This has always baffled me. I mean, then what is the meaning of the point size? I have had to say “Line Spacing: Exactly 18 pt” or some such arbitrary size (which “looks OK” for me) to get evenly spaced lines even when I am using the same point size.

I hope I understand correctly that:

  • the vertical advance is the sum of the ascent and descent.
  • if ascent + descent == units_per_em, then the actual vertical advance would be the same as the defined point size
  • if ascent + descent != units_per_em (it can be > units_per_em or < units_per_em), then the actual vertical advance would be: (ascent+descent)/(units_per_em) – this can even be smaller than the define point size!

Is the above correct?

In which case, what is the meaning of the line spacing setting in word processors? Do I understand correctly that double line spacing in fact means double the vertical advance (which is ascent + descent) and not double the point size?

[Just thinking out loud to ensure my understanding is correct…]

Thanks!

2

I usually design my fonts so that WinAscent + WinDescent = 2048 x 1.2 = 2458 — then 10 point text will be spaced at 12 points, i.e. six lines per inch (72/12).

DTP programs like PageMaker can control the line spacing in terms of pointsize, but word-processors and most applications default to single line-spacing using the line-spacing specified by the font designer.

Times New Roman defaults to 11.494 points at 10 point single spaced. Other fonts vary from Arial at 11.467 to Palatino Linotype at 13.494.

PagePlus allows me to set the line-spacing as a percentage of the pointsize, which is the best solution IMO. I don’t need to know or care what the font designer specified — I use the spacing that I think is best for my needs.

Irrespective of the default line-spacing advance, some fonts look bigger than other fonts of the same pointsize due to larger x-height and/or larger caps height.

3

Here a recent issue related to font size:

Michigan Supreme Court Rejects Challenge To Ballot Petition’s Font Size

4

With metal type, point size is the vertical measurement of the piece of metal alloy, regardless of what glyph is on the piece of metal type. So, the point size is measured from somewhere below the descender of p to somwhere above the ascender of h.

With electronic fonts, when used in most (all?) applications programs, point size is measured from the baseline to somewhere above the ascender of h. The space for the descenders is extra and just seems to be taken into account without usually being mentioned.

The Windows systems that I have used all seem to convert number of points to number of pixels by multiplying the point size by ((Win Ascent + Win Descent)/(Win Ascent)) by 4/3. So, for example, 24 point is expressed as 32 pixels or whatever greater number is needed to account for Win Descent as well.

When I make fonts, I usually use 2048 font units per em and I set the Win Ascent value to 2048 font units.

Sometimes I set Win Descent to -768 font units and sometimes I set Win Descent to -1024 font units.

For example, I used a value of -768 for Win Descent in my Quest text font.

For example, I used a value of -1024 for Win Descent in my Sonnet to a Renaissance Lady font.

These values are because I like to have multiples of 256 font units because I feel that this helps the rendering system produce a clear display. That feeling may or may not be well-founded, yet those settings do produce good results. I mention that I am somewhat mathematically minded and have not formally trained as a type designer, so maybe I have used my mathematical instincts more than is necessary.

I find that WordPad is a good program to test fonts.

For example, the following.

I set several lines of text using Quest text at 24 point.

I then set one line of text using Sonnet to a Renaissance Lady at 24 point, the text containing a letter t.

I then copied the t from the line of text that had been set using Sonnet to a Renaissance Lady onto the clipboard.

I then pasted that letter t into the middle of the second line of the text that had been set using Quest text.

The display of the third and subsequent lines of the text set in Quest text jumped down, even though the letter t had no descender.

I then pasted that letter t into the middle of the third line of the text that had been set using Quest text.

The display of the fourth and subsequent lines of the text set in Quest text jumped down, even though the letter t had no descender.

William Overington

14 August 2012